Author Archives: Chris Makara

About Chris Makara

Since 2003, I have developed an in-depth of experience in Interactive Marketing & Digital Strategy, with a focus on SEO, Social Media & Demand Generation. I am an avid Football Fan, Golf Enthusiast & Ambidextrous Bowler. I can also be found on LinkedIn and Twitter.

[HOW TO] Brilliantly Scrape Twitter Data With Microsoft Excel/Google Sheets

Who doesn’t love data? Not just Data who was in Goonies, but cold hard Twitter data?

Back in the day (I guess I am showing my age here), the character Data had all the gadgets in Goonies. No matter the situation he seemed to always have some contraption that bailed himself (and others) out of trouble. Twitter has tons of data to bail you out of a jam. Whether it is content creation ideas, targeting the right followers, or even seeing what your competition is up to.

While there are countless tools that provide insight into Twitter data, chances are you have said, “there has to be a way to do this in Microsoft Excel.” If not, now I might have you thinking.

The bad news is it is not a straight forward process. The good news is, I figured it out how to export Twitter data to Excel and will share the details on how to do it through some trial and error after reading this post from Matthew Barby. If you know of a better/cleaner way to construct any of the formulas, please let me know in the comments below.

🆕 MAJOR UPDATE  – The previous method described in this blog post no longer works. However, I did manage to create a Google Sheet that does. Click here to get the Google Sheet Twitter Scraper.

With this Excel spreadsheet, you will be able to pull in most data from any Twitter user, including first and last name, location, bio, URL, number of Twitter followers, number of Twitter users they are following, number of tweets, how long ago they last tweeted, and the content of their latest tweet. This info is very handy if you are trying to analyze certain Twitter users. Continue reading

6 Superb Ways IFTTT Can Increase Your Digital Marketing Productivity

I am all about doing things efficiently. If that means using some sort of automation to get things done, so be it. One of my favorite tools as of late is “If This, Then That.” Or simply known by many as “IFTTT“.

The way IFTTT works is simple, you create recipes to automate tasks. These recipes are based on the premise of “If this trigger happens, then that action takes place.” Simply think of it as a “cause and effect.”

IFTTT supports over 70 channels that you can utilize to create your recipes. Such channels as Facebook, Feedly, Evernote, Google Drive, Twitter, and more. Let’s take a look at 6 ways using IFTTT can help increase your productivity.

Continue reading

How to Build a Chatbot: The Essential Guide for 2020

How to Build a Chatbot:

The Essential Guide for 2020

Interested in how to build a chatbot? 

In this post, I’ll cover everything there is to know about chatbots.

But first things first…what is a chatbot?

We all have an image that pops up in our heads when we think of chatbots.

Maybe it’s inspired by where we’ve seen (or think we’ve seen) them used in the past…or perhaps hearing the word conjures up random images of tiny robots we remember from a cell phone commercial or sci-fi special.

Or maybe you already have a general understanding of what chatbots really are, but you want to know how you can make one work for your business.

Whatever your knowledge level is, this post is designed to give you a streamlined crash course on all things chatbot – what they are, why businesses use them, how they work, etc.

Ready to see how to build a chatbot?

Let’s get started.

Chapter 1:

What is a chatbot

A chatbot is essentially a program that uses machine-learning and pre-constructed “rules” to automatically respond to user-generated messages.

A pretty broad definition, yes, but it’s fitting, as chatbots are used on a huge range of websites and digital platforms, for all kinds of different reasons.

Ever asked Siri, Alexa, etc. any question, about anything at all?

Those are a few obvious examples of chatbots.

Clearly, you have to follow a few standard guidelines when making requests in order to get a satisfactory response, such as starting your questions with the bot’s name, speaking clearly, using common terms vs. obscure slang…

But for the most part, these programs are built to hone-in on keywords and generate responses accordingly.

For example, take a look at this sequence flow of the bot that is currently running on my website. 

You can test the bot by clicking here.

Other examples of everyday situations where you may be depending on a chatbot’s use include scheduling an Uber ride, ordering a pizza via the Pizza Hut or Dominos app and communicating through an automated customer service chatbox featured on a retail or service company’s website.

Whether you’re truly familiar with what a chatbot is or is not, as you can see, you’ve probably been engaging with them for years now without giving the concept much more than a fleeting thought.

Chapter 2:

Chatbot Use Cases

If you want to drill down to basics, chatbots were essentially created to serve as a solution to the heavy burden placed on companies’ customer service associates.

Think about it – with the onset of e-commerce, the number of sales, interactions, questions and requests being transacted and responded to every second is multiple times that of what could have ever been managed by the on-the-clock staff of a single brick-and-mortar business of the past.

These days, a clothing company can be headquartered in Los Angeles, California, and have service representatives answering questions from customers in Beijing, Stockholm and Sydney, helping customers in Washington D.C. and Honolulu make purchases online and following up on order inquiries on products that are being shipped to Orlando, Munich and Athens.

If every one of those customers speaks to a service rep to ask their questions or resolve their claims, the clothing company would have to double, triple, quadruple (you get the picture) their staff to keep up.

That gets expensive, sometimes prohibitively expensive.

Additionally, companies may not have the office space to house that many employees.

Some companies have handled space challenges by implementing large, warehouse-like call centers or allowing certain representatives to work remotely.

But long waits on the phone continue to be an issue.

Online support via e-mail or human-managed chatboxes relieves some of the burden on call centers.

But at the end of the day, the central problem still exists: The number of customers to representatives is way too high to be reasonably managed.

That’s where chatbots come in.

These automated programs can be built into online chatboxes or smartphone apps to help serve the basic, everyday customer requests that tie up phone lines.

If the majority of these basic requests can be resolved through a series of automated messages, then reps can be freed up to handle the more complicated claims and issues needing human attention.

So that’s the first major business use of chatbots, in a nutshell.

But they can do so much more. Here’s a breakdown of some of the most significant ways businesses use chatbots.

Deliver 24/7 customer support.

 

Even if the consumer demand on your business is easily handled by your regular support staff, supplemented by seasonal reps during periods of high-demand, there may be a time when no one is available to help.

Maybe it’s late at night and everyone’s gone home for the day.

Or perhaps you have customers in another country, with only a small time frame available to speak with support staff.

By implementing a chatbot on your website, you can offer these customers round-the-clock support – without the expense of adding another shift or hiring more people.

Help customers navigate – and make the right choices.

 

Not everyone that lands on your website will be a returning visitor.

Some people might have just heard your company name somewhere or been referred to your site by a friend. Therefore, while they know what they ultimately want from you, they may not have any clue as to what steps to take to get there.

Or maybe they aren’t picking the best option to meet their specific needs, due to a lack of knowledge.

A chatbot can talk customers through a basic set of exploratory questions to help them get going in the right direction.

For example, say you’re a custom printing company.

A customer, Sara, hears about your company on a radio commercial and navigates her way to your website via Google search. She knows she wants to create a special birthday card for her son’s 5th birthday, but she has no idea how to get started.

A few seconds later, a chat window pops up asking her if she would like assistance creating her first greeting card.

She replies yes, and answers additional questions as the bot explains how to choose a template and add colors, designs and copy.

After a few minutes, Sara closes the chatbot and starts creating her card.

In this case, the company’s use of a chatbot may have alleviated one less call from their busy phone queue.

Or, perhaps they avoided losing Sara as a customer due to her leaving the site after a few minutes of not finding what she was looking for.

In addition to helping customers navigate their way to solutions, chatbots can also be used to help them make the best decision possible.

In a different take on the example above, the initial bot may walk Sara through a few exploratory questions to pinpoint the product type that will best serve her needs.

These questions may reveal that rather than a birthday card that her young son may not be that interested in reading, a big, colorful banner to hang at the party may be a much more satisfactory solution.

Happy customers are good, but happier customers are better.

By implementing a chatbot, you can help customers find solutions that will push their satisfaction to the next level.

Inspire more subscriptions.

 

Similar to how chatbots can be used to help guide customers toward the best solution for their needs, they can also be used to encourage visitors to subscribe to your blog or newsletter.

In this case, it’s best to place a chatbox pop-up on a blog page where active readers will be likely to see it.

It might be used to ask for feedback with the intention of starting a conversation that eventually leads into telling readers where they can subscribe.

By gaining subscribers, not only are you earning your brand a spot in this potential customer’s inbox as a recurring marketing opportunity, but you’ve also now captured their contact information to be added to a drip campaign or used for future sales messaging opportunities.

Create new, engaging customer interactions.

 

There aren’t that many ways to make a customer service phone call or e-mail inquiry a unique and memorable experience.

Even with charismatic representatives and quick replies, the process is always essentially the same: call, talk, solve/get transferred for additional help, talk, get put on hold, etc., or the back-and-forth exchange that’s standard with any e-mail interaction.

But chatbots are an entirely new playing field.

Even though “rules” must be followed to some degree in order for the program to work effectively, communications can support features that aren’t typical to your average brand interaction.

Some apps support emoji use, make sounds or play music, allow customers to peruse store locations or available products, etc. Exchanges go from bland support messaging to funny, colorful interactions that customers will remember, and ideally, find helpful.

Best case scenario?

The customer finds an answer/solution and enjoys the interaction so much they show/share it with their friends and family.

Did someone say, referrals?

Earn more leads.

 

Let’s face it, your sales reps are much more engaging (and happier) when they aren’t cold calling day in and day out.

They get sick of asking the same qualifying questions over and over because, well, they’re human.

But chatbots aren’t human.

They can ask that same set of questions repeatedly, all day long, and still respond with the same level of enthusiasm and patience.

So it makes sense to leverage a chatbot’s basic step-by-step messaging ability to ask site visitors a series of exploratory questions as to why they’re there, what their needs/pain points are and what they value most from a product or service provider.

Feedback generated in these exchanges can be used to update marketing collateral to better reach prospects or supplement sales rep training programs.

If a visitor’s responses indicate an elevated level of interest, your chatbot can ask if they’d be open to speaking with a sales representative, and, if so, subsequently ask for their contact information.

These responses can be automatically sent to your sales department for follow up, and Voila! A warm lead.

And even better – a warm lead sourced without any human effort.

I know that you’re thinking. “That’s what you can do with a chatbot. But why do it? What kind of return can I hope to gain?”

Funny you asked…

Chapter 3:

Benefits of having a chatbot

The whole purpose of a chatbot in serving one of the needs above is to realize a business benefit of some sort, obviously.

I briefly touched on a few of these above; however, the following offers more detail on some of the big benefits of implementing the use of a chatbot into your marketing strategy.

Saves time.

 

Chat messages are (obviously) short, sweet and to-the-point, in contrast to much lengthier e-mail messages.

In the time it takes to build out an e-mail drip campaign, your chatbot could have already engaged with dozens of people via a messaging platform.

Earns higher response/engagement rates.

 

Because of the immediate nature of “instant messaging”, there’s an elevated sense of urgency to read and respond to the questions/comments that just “pop in”, vs. the slower engagements expected on channels like e-mail.

Not to mention societal behavior and ever-shortening consumer attention spans lend themselves to short-natured chat messages.

Additionally, when users are engaged in any kind of instant messaging conversation, it monopolizes their full attention at the time. E-mails delivered to an inbox may get lost in the shuffle and have to fight for attention amidst the avalanche of other messages.

Makes CTAs personal.

 

When a commercial shows up during your favorite TV show, there’s no personalization directed to you, specifically. Nor is there the pressure and expectation of an immediate response.

Direct mail pieces go from your mailbox to the trash can. Sales e-mails are often disregarded as spam and deleted.

But when a chat message pops up, it’s literally a one-on-one conversation demanding an equally quick response from YOU.

Personal, direct, immediate chat messages are more likely to be answered than universal CTAs mentioned in mass media or anonymous e-mail blasts.

Puts you ahead of the curve.

 

Chatbots and messaging platforms aren’t going away anytime soon.

In fact, the number of users are steadily growing as our society becomes more and more mobile, with shorter and shorter attention spans and more and more media distractions.

Getting your business on board now will only help you ride the wave into a future where messaging platforms are becoming the preferred form of B2C communication.

Knowing the what and why behind chatbots is important, no doubt.

But it’s all for not without the how…

Chapter 4:

How to make a chatbot

These basic steps will help you through the process of building a chatbot for your business needs, whether you’re already an expert…or all of the above is news to you. Hey, everybody’s got to start somewhere.

Identify your end goal.

 

This first step is an obvious one – you have to know where you’re going before you can build a marketing plan to get there.

You need to decide what you want your chatbot to do and what kind of benefits you ultimately want to gain.

Is your customer service team struggling to keep up with calls? Do you want to gather some warm leads to help your sales team move away from frustrating days of cold calling?

There are no right or wrong answers.

And you can always create more chatbots later.

But it’s important that you have a decision made before you get started.

Your end goal directs the entire process.

Decide where your chatbot will be featured.

 

The platform or web page where your chatbot will live also affects how it’s built. This decision is largely based on your end goal, as well as your knowledge of your audiences.

Places where you can potentially feature a chatbot include the following:

  • Various pages of your website
  • Smartphone/tablet app
  • Twitter Direct Message
  • Facebook Messenger

As an example, say you’re a company that sells sports gear and gym equipment directly to consumers and you want to boost sales of your larger machines like treadmills, ellipticals, stationary bikes, etc.

Your team decides to build a chatbot to help generate new leads for your sales managers.

Knowing your end goal, you start thinking about where you’d like to feature this chatbot. Your first thought is to place it on the landing page of your website, hoping to capture the attention of visitors who may be perusing available equipment before heading out to the store to look at and buy it in person.

But you also know that you offer products for nearly every sport imaginable. So the chance that on any given day there will be an influx of visitors to your site who are specifically on the hunt for a new treadmill or rowing machine is slim.

You wonder if your social media platforms may offer the opportunity to reach a more specific, and potentially more relevant, audience.

Digging into your Twitter and Facebook demographic data, you find that both offer access to a younger audience, between the ages of 18 and 40.

Deeper insights into your Facebook audience indicate that many followers are active and/or regularly engage in some kind of recreational sport.

You consider the idea that younger, active Facebook users may be more open to engaging in a chat on Facebook Messenger than perhaps any given visitor to your website.

Knowledge of your audience and consideration of consumer behavior reveals that, in this case, your best option may very well be to feature your chatbot on Facebook Messenger vs. your initial thought to place it on your site’s landing page.

Nail down your content strategy.

 

Once you know what you want your chatbot to be used for and where it will be featured, you can start exploring how discussions will flow from start to finish.

  • What are customers most likely to ask?
  • Do you have any pre-existing marketing collateral featuring FAQs and corresponding standard responses?
  • What tone and personality do you want your chatbot to have?
  • Will it have a name?

Don’t feel like you have to reinvent the wheel here. In fact, you really shouldn’t.

  • Reach out to your internal teams for insights.
  • Reference past collateral for lists of FAQs.
  • Review old case studies and sales materials to get a more complete idea of the customer journey.

For instance, if you’re the sports gear and gym equipment retailer mentioned above, you would need to consider how you want to approach a conversation about training equipment needs with followers.

You find that your sales team has a collection of helpful talk tracks that are shared with new sales associates during their onboarding process.

Many start by asking customers about their exercise regimen, whether or not they have a gym membership and/or if they have kids that play on sports teams. Then, based on the answers, the talk tracks offer additional topics of conversation and questions sales associates can ask to gather more information and lead into a conversation about gym equipment.

Referencing documents like these can help you develop the tone, direction, and even the kinds of questions you might ask to capture leads for your salespeople, direct people to your website for more information – or even make a sale online.

As a final mention, remember that the tone of your chatbot should match that of your brand, and be consistent with any other form of communication your company shares publically. It’s all about consistency.

The one exception to this rule is that chat copy should be a little more conversational and a lot shorter than most other forms of B2C communication.

Once you’ve pulled together some reference materials and have a feel for how you want to approach the conversation, you can get down to actually building out the copy.

Come up with an awesome intro message.

 

When visitors first see your chatbot pop-up, what will your intro question or statement be?

It sounds like a basic endeavor, but there’s some deeper consideration involved.

You need to make sure that your opening message captures attention and kicks off the conversation in the right direction, while also being straightforward and short enough to get the point across in a matter of seconds.

Keeping your end goal in mind along the way is another absolute must.

Also consider that your initial message needs to be more generic and open-ended so it can serve as the lead message for many different possible conversation tracks (more on this in step 5).

And don’t forget – your bot is fighting for attention amidst all the graphics, products, comments and news feeds on your chosen platform (landing page, social media channel, etc.).

So your pop-up needs to stand out.

Refer to the research and company collateral you gathered and considered in the last step.

These documents can help ensure that you start the conversation with a question that targets a popular customer pain point, or perhaps with an intro statement that offers guidance on a topic commonly discussed with your customer service team.

Whatever makes the biggest bang from the get-go.

Because when you really think about it, this first message may very well be THE most important message of all. It’s the make-or-break moment where a visitor will decide whether to engage with you.

Or not.

So make it good. No pressure though…

Build out the rest of the conversation(s).

 

Here’s where you really get to work on imagining what conversations will be like and developing different talk tracks for each.

Essentially, you start with your intro question, then consider the many different directions a conversation can go, based on how the user responds.

If this response, then one talk track ensues. If that action, then another, and so on and so on.

Conversations begin to branch off of one another and oftentimes reconverge in different areas – be open to letting things expand organically.

[One thing to mention here – after the intro message, be sure to mention that the user is speaking to a bot, so they realize the limitations of the conversation. You should also include information on where users can go to speak with a human if they have a more complicated issue or need.]

For instance, say you’re a pet shop owner and you find that many people are overwhelming your customer service lines asking basic questions, the answers to which can easily be found on your website.

These questions may be regarding what time your store opens, what pet food brands you sell or whether or not you offer grooming services.

You decide to build a chatbot to free up the phone lines for more complicated calls and feature it on your website’s home page, popping up seconds after visitors arrive.

You make your intro message very simple and open-ended:

“Thank you for visiting PetWorld! Can I help you with something?”

If the customer clicks on the pop-up, the next message dispatches. [If they close it out, you can feature a message such as: “No problem. I’ll just hang out here while you look around. Click if you need me!”]

Next message:

“Hi, I’m PetWorld Chatbot, nice to meet you!”

A second after…

“What can I help you with? You can ask me things like ‘Where are you located?’ or ‘Do you have Purina dog food’ by typing your questions below.”

A second after…next message:

“If at any time you wish to speak with a representative, feel free to call us at 234-555-5555 between 8 am and 6 pm.”

From here, you can refer to your business’s FAQ list to build out different talk tracks, inspired by the most popular customer questions.

Your bot will offer replies based on keywords in the questions, so it may take some tailoring over time to get it right.

But you’ll get there!

To keep things organized when planning, many people find it helpful to draft potential conversation tracks visually, in formats resembling a flow chart or corporate hierarchy chart.

You can go as simple as shapes and arrows in Microsoft PowerPoint and/or Google Drawings, or you may want to start drafting in an online tool such as Mobile Monkey that offers a form-like template where you can enter and test various conversations.

Additional message “add-ons” to consider throughout this process are clickable buttons, emojis, GIFs and graphics.

These may help boost engagement and enhance conversations to make them more memorable or intuitive.

Trying to consider all of this at once can be overwhelming.

But when you take it one conversation at a time, you’ll find that many messages can be mirrored across different talk tracks and that some questions end up reconverging in the same places, saving you from having to come up with brand new copy for every possible ask.

Research and invest in a chatbot tool that meets your needs.

 

This is a rather obvious step, as unless your lead developer is going to take your conversation tracks from here and run with it, you need to research and purchase a chatbot building tool to help bring your plans to life.

Different chatbots exist for different platforms and different needs.

Again, another reason why it’s essential to make a firm decision on the end goal you’re trying to achieve and the platform you intend to use at the beginning of the process.

Research what’s out there and don’t be afraid to ask for demos or to speak with a knowledgeable sales rep, when you still have unanswered questions after reviewing the tool’s website.

Here’s a short list of the best chatbot builders (or social media management suites that offer chatbot-building options) to get you started:

Once you’ve decided on the right tool for your end goal and chosen platform, spend some time checking out all the features to get familiar with how things work and what’s available.

Then finally, you’ll need to add your conversation copy into the format provided and get ready for testing.

Not sure how to build a chatbot? I can help! 👉 Click here to chat 🙂

Test your conversation tracks, over and over again.

 

Coming up with every possible question, response, keyword, etc. while building out all of these potential talk tracks is impossible.

Even for people who have built chatbots multiple times in their careers.

There’s just no way to think of everything the first go around.

But it’s no biggie. Because you can test as many times as you want.

The majority of chatbot tools and platforms let you preview and test conversations without officially launching them.

This way, you can walk through all possible interactions before any customers do, and if/when you hit a snag or realize you left something out, you can easily edit and re-test as you go.

Some chatbot platforms support natural language processing (NLP) which uses machine learning to understand the context of a conversation and provide accurate answers.

Launch and observe.

 

After you’ve completed steps one through seven above, it’s time to go live.

Launch your chatbot!

Celebrate!!!

But don’t expect to get it perfect the first time.

Be sure to monitor how customers are using the bot and whether there are areas where confusion ensues or conversations dead-end.

Remember, with the help of a user-friendly chatbot tool, you can always make tweaks, test them offline and then push them live in a matter of minutes.

Evaluate and adjust accordingly.

 

Finally, after your chatbot has been in action for a time period suitable for achievement of your end goal, you’ll want to evaluate its overall performance.

  • Did it help you earn significantly more leads?
  • Cut down on overtime shifts?
  • Upgrade your customer service ratings?

If so, keep going! You may even want to apply this strategy to another important end goal to see if you can achieve similarly positive results there.

If not, don’t sweat it.

Go back and see if you can identify any bottlenecks or points of confusion. Try moving your chatbot to a different landing page or platform. Add emojis. Tighten the copy.

Feel free to experiment. Your chatbot tool makes adjustments and testing easy, so you can edit and evaluate over and over until you’re satisfied with the results.

Not sure how to build a chatbot? I can help! 👉 Click here to chat 🙂

Chapter 5:

Why your business needs a chatbot

Chatbots are only one small facet of a much bigger initiative that’s been forming over the years known as messenger marketing.

Messenger marketing is truly as simple as it sounds: marketing your products or services to customers via a messaging app.

There are plenty of reasons messenger marketing has gained so much traction in this day and age. I’ve already mentioned a few of these above, but the most significant consumer trends driving this evolution are:

  • Shorter attention spans
  • Expectation of instantaneous responses
  • More time spent on mobile devices
  • Distracted by multiple forms of media

Messenger marketing solves for each of these.

Messages are short, immediate, mobile-friendly and personal. You’re meeting the customer where they’re at, and tailoring your message to their behavior.

But the biggest reason you should jump on this trend – and sooner rather than later?

To stay a step ahead of your competition.

According to internet marketer Larry Kim, “There are more daily active users of popular messaging applications than social media applications, yet less than 1% of companies are doing chat marketing.”

So what are you waiting for?

Build your chatbot and put it into action before the rest of the industry can even define what they are.

This guide is all you need to get started.

Not sure how to build a chatbot? I can help! 👉 Click here to chat 🙂

Now It’s Your Turn

So that’s how to build a chatbot.

Let me ask you: What most excites you about chatbots?

Is it the time saved by use of automation?

Or is it leveraging another channel to drive leads and sales?

Leave a quick comment below to let me know your thoughts on building out chatbots.

Tweet Analytics: The Definitive Guide to A/B (Split) & Multivariate Testing Using Twitter

You are probably familiar with the acronym ABC which is short for “always be closing.” This is a mantra preached by many in the sales world.

Lucky for me, I’m not a salesperson.

So I like to live by a different philosophy – ABT.

In this case, “always be testing.”

Look, I am a huge fan of data and analytics. Particularly when it comes to social media analytics.

There is so much that can learned by knowing what does or doesn’t work.

But did you know that you can use Twitter to A/B, split test or even multivariate test?

Not only just sharing different variations, but the ability to easily view your tweet analytics side-by-side in Microsoft Excel.

It’s true and today I am going to reveal how I use Twitter Analytics and Excel to do perform your own tweet analysis.

Here’s some of what we’ll cover:

There is a lot to go through, so let’s get started.

Download My Ready-To-Go Excel File  Click here to download my Microsoft Excel file so you can easily view your tweet analytics.

Different Types of Testing

The two most common types of testing are A/B testing and multivariate testing.

With A/B (sometimes referred to as split-testing) you are running a test of two items. In this case, some possibilities for an A/B test are:

  • Changing the CTA of the tweet (using “Please RT” against “Share this”, etc)
  • Test a tweet that uses an attached image against one that doesn’t use an image
  • Perform a hashtag analysis where you compare the effectiveness of one hashtag against another

The idea with an A/B test is to run the control (A) against a controlled variation (B).

Here’s an example:

In multivariate testing you are changing two or more things in the variation against the control. A few examples are:

  • Impact of the CTA by different hashtags and attached images
  • Changing of CTA’s impacted differently with the placement of the URL in the tweet
  • Test the effect of engagement rate of using the blog post title against a section title of your blog post

Here’s an example of a multivariate test:

The important thing about running A/B or multivariate tests is to know what you are testing for. Are you interested in engagement rate, number of URL clicks, likes, retweets, etc.

Knowing this up front can help you create better tests.

Why You Should Test Using Your Twitter Tweets

There are many different ways you can test. For example, there are WordPress plugins that can switch out your blog titles.

But, what if your site doesn’t get a ton of traffic?

You will be waiting forever to get results of what does or doesn’t work.

Other options for testing include using paid social ads. While this does work and can get you results rather quickly, maybe you don’t have a budget to do this?

So what’s the free solution?

Use your Twitter account.

Of course, if you want to use this method you should have a decent size following or a very engaged user base.

Before You Can A/B Test With Twitter, You’ll Need to Setup Your Analytics Account

When you create a Twitter account, you will also be able to get access to your Twitter analytics. It’s a pretty easy process to get setup.

If you already have Twitter Analytics up and running, you can go ahead and skip to the next section.

So if you don’t already have a Twitter account (I won’t judge you), go ahead and create your account. Once your account is created, you have “turn on” Twitter Analytics.

Simply go to Twitter Analytics and login.

Once logged in, you will see a button to click in order to setup your Twitter Analytics account.

It’s important to setup your Twitter Analytics account as soon as you can since data won’t be tracking until you have activated your account.

Depending on if you are using a new account or already had one setup, your date range of data will vary for your tweet analytics.

Creating and Scheduling Your Tweets

In order to run any sort of test, you will need at least two things (versions) you will want to test. This can be something as simple as changing the CTA in a tweet from “click here” to “read now”. Or you can have two entirely different texts to test. Additionally, you can test tweets with or without images.

Test Tweets from Existing Content

For me, the easiest way to create tweets come from work that I have already done.

Simply put, I scan the blog post I will be sharing to see what I would like to test. This is one of the ways I go about creating evergreen social media updates – see #4.

Basically, you can look for:

  • Section headings
  • Stats
  • Quotes
  • Tips/Tricks

For example, there is a section of my blog post called “Why You Need Multiple Social Media Tools”. Within that section you have a nice eye-catching statistic like

“Facebook has 1,900,000,000 monthly active users. Yep, that’s 1.9 billion.”

blog post variations to test

You could decide to test the section heading against the blog post title as well as the statistic.

Test Tweets With or Without Images

Another option for testing could be to test whether the use of an image with the tweet effects engagement rates. To do this, you would have a tweet with an image, then another tweet exactly the same but without the image.

Additionally, you could test tweets that both had the same image, but different text.

And when it comes to images, here’s a great tool that offers countless social media image templates to use.

Test Blog Post Titles With Tweets

I’m sure as you were pounding away at the keyboard on your latest blog post you more than likely came up with a handful of blog post titles. (No blog yet? Read this to start a blog).

There are some you probably thought were garbage, while others made it difficult to decide if it should be the one you ultimately roll with when you publish your post.

Why not take your different titles and use them as the text of the tweet?

You can then see which one(s) get you the most engagement. Who knows, it could be the one you least expected…but you’ll never know until you test it with your social audience.

Test The Impact of Hashtags on Your Tweets

Are you using hashtags in your tweets? If not, you could potentially be missing out on more engagement.

According to this Buffer post, tweets that use hashtags can get double the engagement rate. But be cautious if you plan on using more than two hashtags as engagement can drop.

hashtags increase engagement rate

If you aren’t using hashtags, you might be missing out. But you won’t really know unless you do your own hashtag analysis.

Create a File With Your Tweets to Test

No matter which route you decide you want to test in Twitter, I recommend creating a file to save all your variations of Tweets. This will make it easier for you when you want to share the tweets.

Here is what you will need to do:

  • Create an Excel file (or Google Sheet) with three columns
  • In Column A, enter the text of your tweet
  • For Column B, enter the URL of the blog post
  • If you are going to use an image, note the location of it in Column C

You could do this in Word, but I prefer Excel in that I can sort/filter if I have quite a few tweets to schedule. Plus it makes it easier to get your content into a social media automation tool.

Scheduling Your Tweets to be Tested

Depending on how you want to test may determine the approach you take to schedule your tweets. Some people prefer to post both variations at the same time and base their results on just those two tweets.

You can even identify the best times to tweet for your followers, and choose to post during those times.

best times to tweet

Other people, myself included typically like to share the different variations again and again over a period of time.

Going this route allows me to do a few things:

  • Get a larger sample size of people who viewed the tweet
  • Opportunity to show different variations across different days/hours/timezones

Good news is that whether you want to only post each test/variation once or multiple times, you have several options to schedule these Twitter tests.

Post the Tweets Manually

This approach would work best if you plan on doing a one-time test. Meaning that you want to simply post the tweet one time and see what happens.

Simply login to your Twitter account and post your tweet from the Excel file you created.

Once posted you can check the tweet stats on Twitter for that single tweet.

Now, unless you have a huge Twitter audience you probably won’t get too much insight from posting a single tweet. As you know, only a fraction of your audience will actually see your tweet.

Therefore, I recommend posting the same tweet multiple times to maximize exposure and to get enough data to make an informed decision. This can easily be done through scheduling your social media posts.

Use Hootsuite to A/B Test Your Tweets

One of the largest social media scheduling tools is Hootsuite. This tool will allow you to do many things on Twitter including social media monitoring, team collaboration, and the ability to schedule tweets.

Remember the Excel file you created with your tweets? With some slight modification you can use it to upload your tweets to Hootsuite.

Here’s what you need to do.

The format that Hootsuite will accept is a 3 column CSV file (a CSV is a comma separated file you can make in Excel). These columns are:

  • Date and time – What day and time will the tweet in this row be published
  • Your message – This is the content of the tweet. It should be at most 117 characters
  • URL (optional) – While it is optional to use in the eyes of Hootsuite, we will need to use one so that we can more accurately monitor the data later. This will take about 23 characters when the URL is shortened

To make your Excel work in this format, you will need to insert a new column to the left of Column A. This will push all your columns right one letter (Column B becomes Column C, etc).

You will then need to delete the column with any images you had listed. Unfortunately, Hootsuite doesn’t allow the use of images in a CSV to be attached to your tweet (If there is a way to do this, definitely let me know).

So now you should be left with 3 columns of data:

  • Date and time
  • Your message
  • URL

Before we save and run off to Hootsuite, you will need to set up days and times to post.

Perhaps you already know this info, if so you can enter it in the cell of Column A. Please note that the proper format for this is MM/DD/YYYY  HH:MM:SS PM.

If you don’t know what day/time you want to post yet, I would just enter some random future dates and times into these columns.

Or if you want you can paste the following in the first 3 rows and then drag down the dates to the last row. It should auto-increment the day/time for each row. Just be sure to change the actual date to at least tomorrow. Here are the temporary dates:

  • 2/11/2016 7:30:00 AM
  • 2/11/2016 8:00:00 AM
  • 2/11/2016 8:30:00 AM

Then once your CSV is uploaded, you can rearrange when you want certain tweets to be published.

It’s important that you save the file as a CSV. To do that in Excel, click “File > Save As” and for the file type select “CSV (Comma Delimited)” from the drop down and then click “Save”.

Next, you will need to login to your Hootsuite account and click the “Publisher” tab in the left menu. Once in the publisher tab, click on “Bulk Message Upload”.

You will be prompted with a screen like this:

hootsuite bulk upload csv

In just a few short steps, you will be able to upload your list of tweets to test:

  • Simply click the “Choose File” button and browse your computer where you saved your CSV file.
  • If you used the date format I mentioned above, just be sure to click the radio button for “mm/dd/yyyy hh:mm.”
  • Next you will need to select the social network (Twitter account) you want to schedule these tweets for.
  • The final step is click the “submit” button.

Once your CSV file has successfully uploaded, you will be redirected to calendar of scheduled posts. From here you can move around and edit any of the days/times of when the tweet should be published.

Some things to note about using Hootsuite to publish your tweets for testing:

  • Hootsuite only allows you to upload and schedule one version of an update. Meaning that you can’t typically upload duplicate tweets. You will have to re-add them later once they have been published.
  • There is not an easy way to recycle your tweets to use again and again automatically if you want to post the tweets multiple times. You will have to re-add your CSV file once all scheduled tweets from it have been sent.
  • If you want to attach an image to your tweet, you will need to manually click on the tweet to edit it and upload your image.

If you are not already using Hootsuite, you might need to decide if it is right for you to schedule your tweets for testing. An alternative to Hootsuite is Buffer.

A/B Test Tweets With Buffer

Buffer is another social media tool that has a primary focus around scheduling Twitter updates. If you would like to use Buffer to schedule your tweets, unfortunately it is a manual process.

To add your tweets for testing to Buffer, you will need to first select your Twitter account to test with and then click on the text input box to create your tweet.

buffer schedule tweets to test

To make the copying/pasting easier using the Excel file you created above, you will want to use this formula in cell D2:

=CONCATENATE(A2," ",B2)

You can then drag (or copy) down this formula to the end of your list of tweets. When you are ready to add them to Buffer, simply copy the text from Column D and paste that into Buffer when you create your tweets.

The tweets you add to Buffer will automatically be slotted into the next available time you have setup. You can easily drag a tweet to move it up or down in your queue.

Additionally, if you want to create a custom time to post the tweet this can be done by hovering over the tweet and clicking the “edit” link.

Next, click on the date to select a particular date. If you want to set a custom time outside of your default schedule, simply click on the “custom time” link and enter the time you want to tweet this post. Be sure to save your tweet!

buffer custom post time

If you want to post the tweet right away, you can hover over the tweet and click “share now.” This will send your post to your Twitter account immediately.

buffer share now

Should you want to simply upload a CSV to populate your tweets automatically, there is a way to do it.

Use a CSV File to Upload Tweets for Testing with Buffer

One of the nice things about Hootsuite is that you can upload a CSV with your tweets. With Buffer, it’s not an option.

However, I’ll show you how to upload a CSV to Buffer.

In order to upload a CSV to Buffer we are going to do it with the help of a site, Bulklyfull disclosure, I created Bulkly and it is currently in beta.

The first thing you will want to do is create an account and connect your Buffer account.

Once you have granted access to Buffer, all of your social media accounts in Buffer will show up in Bulkly. Bulkly will send updates to your Buffer account for you.

Bulkly offers a handful of options to load up your Buffer account, but for this post we are going to focus on the “Content Upload” option.

Unlike Hootsuite, Bulkly will allow you to group together social media updates.

Notice the pending, active, and completed columns that each have groups of updates within them.

It makes it very easy to stay organized.

And this comes in very handy if you are going to want to post these updates again and again in order to get more exposure and data. Additionally, it allows you to keep these updates separate from any other updates you have grouped in Bulkly which will allow you to set more or less frequent sending to Buffer. When you are finished running your Twitter test, there won’t be any need to sift through hundreds of scheduled updates to remove only a few – you just have to delete the group or drag it over to the “Completed” column.

One thing to note about Bulkly is that it currently uses your posting schedule as it is setup in Buffer. This means that when a tweet is sent to Buffer, it will be added to the next available spot in your Buffer queue.

Before we get started, we’ll need to tweak the Excel file to work with Bulkly. If you aren’t interested in messing with a CSV, you can easily create your tweets online instead. Click here to see how.

In order to group your tweets in the CSV, you will need to format it like this:

  • Column A (URL) – This is the URL you want the tweet to link to
  • Column B (Image URL) – Enter the URL (optional) of an image you want to attach to the tweet
  • Column C (Content) – In this column you will enter the text of your social media update

To assign these tweets to a group you will need to create at least one group. In Cell A2, you will enter the group name. Then on row 3 you will start entering the URL, Image URL, and content as outlined above.

When you have completed entering your data, it’s important that you save the file as a CSV. To do that in Excel, click “File > Save As” and for the file type select “CSV (Comma Delimited)” from the drop down and then click “Save”.

From your Bulkly dashboard, you will want to click on the “Content Upload” link and then click on the “Upload Content (CSV)” button.

 

Browse for your file. Once uploaded you will be directed to your “Pending” queue. This will show the posts you have uploaded, but not actually scheduled.

From here you can edit and of the items you uploaded. When you are ready, you’ll need to select the Twitter account(s) you want to send these updates to as well as how often you want to send them to Buffer. Click on the “Activate” button to move everything in this group to the “Active” queue.

 

Within a few hours your tweets will be dripped into your selected Twitter accounts at the intervals you selected.

Create Social Updates Without a CSV

Now if the whole CSV talk was a little much, Bulkly gives you the option to simply create the tweets directly on the site. From your “Content Upload” screen, click on the “Add Content Online” button and enter the data through the interface.

Once you have entered your tweets, be sure to activate your group to start sending to Buffer.

Getting Your Tweet Analytics

Once your tweets have been posted a handful of times, you can start to export your Twitter data. To do this, we will be using the Twitter Analytics account you previously set up.

Go to your Twitter Analytics account and click on the “Tweets” tab at the top:

twitter analytics tweets tab

In your tweets tab, by default you will see your last 28 days of tweet activities. This date range can be changed to show a maximum of 90 days’ worth of data starting from when you first setup your Twitter Analytics account.

This page is full of key data for your tweets. Among the data displayed are impressions, engagement rate, link clicks, retweets, likes, and replies. All of this data is shown in the right sidebar.

The meat of the page displays your most recent tweets of the date range. It will show you the tweet, impressions, engagements, and engagement rate. Twitter defines engagement rate as:

“Total number of times a user has interacted with a Tweet. This includes all clicks anywhere on the Tweet (including hashtags, links, avatar, username, and Tweet expansion), retweets, replies, follows and likes.”

Engagement rate is calculated by taking the number of engagements and dividing it by the total impressions.

Your most recent tweets can be toggled using the “Tweets”, “Top Tweets”, “Tweets and Replies”, and “Promoted” tabs. This easily allows you to segment your tweets at a quick glance.

Exporting Your Twitter Analytics

In order to know how your various tweets used for testing performed, we need to export your tweet data out of Twitter.

Luckily, this is a very easy process.

Just click on the “Export Data” tab and wait for it to download.

export twitter data

As I mentioned, you can export up to 90 days of data at a time. So if your Twitter testing has lasted longer than 3 months, you can set the date range to be sure you download all the data. You will just need to combine the data into a single Excel file to get the whole picture of your tests.

After you have downloaded all your data, it’s time to move on to the fun part – manipulating the data to see what does or doesn’t work.

Analyzing Your Tweet Analytics with Excel

Raise your hand if you just cringed at the sight of using Excel.

Don’t worry, I won’t bore you with all the intricacies of how I did this – just some of the key components that make this possible.

But before we go any further, you’ll need to download the “ready to go” spreadsheet for analyzing your Twitter A/B testing data. Normally, I like to provide everything you need to know in the blog post, but there are various behind the scenes items going on to make this all work.

You can try building this out yourself as detailed below, but it’s probably best to skip the headache and grab my working file.

Download My Ready-To-Go Excel File  Click here to download my Microsoft Excel file so you can easily view your tweet analytics.

Your Twitter Export

The first thing you will need to do is open your exported Twitter file. Sometimes, just clicking on the downloaded file will automatically open correctly and import all the data.

Should this not be the case, you can manually import the data in a few steps.

The first thing you need to do is click on the “Data” tab and then on the “From Text” icon.

import twitter csv into excel

Next, you will be asked to find the file you want to import. In this case, you will need to find your exported Twitter file. After, you select the file location a wizard will load (nope, it’s not Gandalf).

Step 1 of the wizard is asking you what is the data type for this file. You will need to have the “deliminated” radio button selected and then click “next.”

twitter csv import to excel step 1

For step 2 of the wizard, you need to indicate the delimiters of the file. In this case, the file is delineated by commas. Uncheck all boxes and have the box selected for comma. You will know it is working when your rows of data become organized into columns. Click next.

twitter csv import to excel step 2

The final step is to tell Excel where you want to insert this data. You should be placing it in an empty sheet so you can click in cell A1 and then click “ok.”

twitter csv import to excel step 3

All of the data from your Twitter export will be added to the spreadsheet. If you exported multiple date ranges from Twitter, just repeat this same process. However, you can insert the data at the end of the data you already imported. Just be sure to remove the row containing the header info for each set of data you import.

Once all your data has been imported, rename your Excel tab as “Raw Twitter Data” and save as an Excel file (not a CSV).

If you want to try building this out yourself, or are just interested as to what is happening in the file keep reading. Not interested in this kind of stuff, I won’t judge – you can click here to get to the part of reading the data.

Add Helper Columns To Help With Data Manipulation

Unfortunately, in order to really segment the data we need to add some “helper” columns in our Excel file. These will have various formulas and formatting.

Before we add these new columns, you need to create a table of the data you already have. The easiest way to do this is to click “Control + T” inside of the data. This will bring up a dialog box asking what data you want in the table. It should auto select all your columns and rows. Be sure to have the box checked for having a header row and click “ok.”

Once your table is created, you can easily create the additional helper columns. These columns will need to be created in the first available column to the right of your table.

The benefit of creating a table out of your data is that it will automatically copy down and formatting or formulas you create in the new columns.

So let’s add some helper columns.

Column AO

The first column you will need to add is called “Short URL” (type the name in cell AO1). The purpose of this column is to extract any URL that was in the tweet.

To extract the URL, you will need to use the following formula in cell AO2:

=IFERROR(MID(C2,FIND("http",C2),IFERROR(FIND(" ",C2,FIND("http",C2))-1,LEN(C2))-FIND("http",C2)+1),"")

Once entered, you will see the shortened URL from the tweet if there is one. The table you created should have automatically copied itself to the last row of your data.

Column AP

This column will be used to unshorten the shortened URLs. The reason we need to do this is that you will have different shortened URLs for each tweet. When we are building out our pivot table later, having a bunch of different URLs will make it impossible to group the data together.

Therefore, you will need to add this module to your Excel file.

Here’s how to do it:

  • Click on the “Developers” tab in Excel. Not seeing the tab, here’s how to add it.
  • In the “Developers” tab, click on the “Visual Basic” icon
  • Once the Visual Basic editor is open, click on “Insert > Module”
  • Paste this code into the editor:
Public Function unshorten(url As String) As String

Static oRequest As Object

Set oRequest = CreateObject("WinHTTP.WinHTTPRequest.5.1")

With oRequest

.option(6) = True

.option(12) = True

.Open "HEAD", url, False

.send

unshorten = .option(1)

End With

End Function

 

  • Click “File > Save”. You will be prompted to save the file as a macro enabled workbook. Click on “no” in the pop up
  • When the “save as” box pops up, select the file type as “Excel Macro-Enabled Workbook”

This module will now unshorten your short URLs. All that is left is to add the formula to the column and name your column “Unshort URL.” In cell AP2, enter this formula:

=IFERROR(unshorten(Table1[@[Short URL]]),"")

Please note that if you renamed your table, you will need to change it in this formula. Otherwise, the table you created should be named this already.

Once entered, this column will unshorten your URLs. Please note that depending on how many rows of data you have, it can take quite a while to unshorten. For example, it took close to 2 hours to unshorten about 6,000 rows of data.

If you have thousands of rows of data, I would let the URLs unshorten. Once completed I would copy the column and paste the values directly on top of the formula for this column. The reason being is that when you add the additional columns below, Excel will try to recalculate the formulas again depending on your settings. By having just values for the unshortened URLs, Excel won’t have anything to recalculate.

Column AQ

Depending on how you track your social media analytics you may have query parameters on your URLs. This column serves as a way to remove those parameters so that you are left with a clean URL.

For example, instead of having a URL like:

Your clean URL will look like this:

Here’s the formula you need to put in cell AQ2:

=IFERROR(LEFT(AP2,FIND("?",AP2&" ")-1),Table1[@[Unshort URL]])

You’ll need to name the column as “URL – Clean”. This will allow you to easily group your tweets by URL in the pivot table we will be creating later.

Column AR

The next column we need to create is one that will remove the URL from the tweet, leaving us with just the text of the tweet. Create a new column and in cell AR1 name it “Tweet No URL”. Then in cell AR2, enter this formula:

=IFERROR(TRIM(SUBSTITUTE(C2,MID(C2,FIND("http",C2),IFERROR(FIND(" ",C2,FIND("http",C2))-1,LEN(C2))-FIND("http",C2)+1),"[URL]")),"")

What this will do is remove the URL from the tweet and replace it with “[URL]”. For example, if the tweet originally was:

The formula will change it to:

  • 21 Ideas For Evergreen Social Media Content Updates [URL]

However, what happens when you are still left with a URL in this column? For example, if there is a link for an image? No worries, just one more helper column to fix this.

Column AS

Similar to the column above, this will remove the second URL (if there is one) from your tweet. In cell AS1, name it “Tweet No URLs” and then use this formula in cell AS2:

=TRIM(IFERROR(SUBSTITUTE(AR2,MID(AR2,FIND("http",AR2),IFERROR(FIND(" ",AR2,FIND("http",AR2))-1,LEN(AR2))-FIND("http",AR2)+1),"[URL]"),Table1[@[Tweet No URL]]))

This formula will take your tweet from looking like this:

To look like this:

  • 21 Ideas For Evergreen Social Media Content Updates [URL] [URL]

With this, you will be able to see which tweets contained an image.

Column AT

Next up is getting the date formatting to work. When you export your data from Twitter, it’s exported as something like this:

  • 2016-03-09 14:20 +0000

Long story made short, we need to get it in a more friendly format for Excel. To do this create a new column named “Date” in cell AT1 and in cell AT2 enter this formula:

=(IFERROR(DATEVALUE(LEFT(TRIM(TEXT(D2,"yyyyy-mm-dd hh:mm:ss")),FIND(" +0000",TRIM(TEXT(D2,"yyyyy-mm-dd hh:mm:ss"))&" ")-1)),""))*1

You will need to change the format of the column from “General” to “Short Date.” To do this, highlight Column AT and then in your “Home” tab of Excel, select “Short Date” from the dropdown.

Select Short Date Excel

This will magically create normal looking dates for your tweets. For example:

  • 03-09-2016

Download My Ready-To-Go Excel File  Click here to download my Microsoft Excel file so you can easily view your tweet analytics.

Create a Pivot Table to See Your Twitter Test Results

Once you have created your helper columns, you can get to the good stuff which is viewing your results.

Since you have already created a table, the next step is to create a pivot table with your data. I know what you are thinking, “A pivot table? I have no idea what that is!”

No worries, it’s really easy.

On your “Raw Twitter Data” tab, click anywhere inside your table. This will activate the “Table Tools” ribbon in Excel. In this ribbon, you will need to click on the icon that says “Summarize With Pivot Table.”

A dialog box will popup asking you where do you want to create this table. Just click “ok” and it will create a new tab for the table.

On this tab you will see the “Pivot Table Fields” menu.

In order to build out your table of Twitter testing results, you will need to drag/drop the pivot table fields to the following areas in this order:

Filters

  • Date

Rows

  • URL – Clean
  • Tweet No URLs

Values

  • Tweet No URLs
  • Impressions
  • Retweets
  • Replies
  • Likes
  • URL Clicks
  • Engagement Rate

You need to make some minor tweaks in the “Values” section so that the data is summarized properly.

For the “Tweet No URLs” data field, in the “Values” section you will see a down arrow. Click on it and then click on “Value Field Settings.” Next, you will need to click on “Count” and then click on “ok.”
change excel pivot table value setting

This will now count all the cells with data in a row with data for this field.

Following the same process, you need to change the “Value Field Settings” for the “Engagement Rate” to summarize the value field by “Average.” This will show an average engagement for each tweet.

Another thing you might want to do for your “Engagement Rate” column is to change the format of the column to show as a percentage. To do this, highlight the whole column and then right-click and select “format cells.” Then click on “percentage” and change the decimals to two places and click “ok.”

Reviewing Your Twitter Testing Data

Your pivot table is now ready to go. However, you will notice that there are probably a lot of tweets that you aren’t interested in.

For example,

  • Tweets to sites that are not yours
  • @ Replies
  • Generic text that could be a quote, thought, complaint, etc. – something you aren’t probably testing

The easiest way to clean up all the miscellaneous tweets is to click on the filter icon in the “Row Labels” cell A3. Then in the search box, type in the domain you want to show and click “ok.”

filter-urls-in-excel

Your filtered results will show you the URL(s) tweeted for that domain and the different tweets that were sent linking to that URL.

Viewing the columns to the right, you will be able to see the different type of engagement that specific tweet text generated. You will need to determine what your main KPI is.

Typically, I put more weight onto the average engagement rate in that I am looking to create social updates that drive more overall engagement.

best engagement rate on twitter

On occasion, I am more interested in driving clicks to a URL so I will look at the URL clicks. However, instead of showing the sum (total) clicks to a URL, I change the column to show the average clicks for the social media update.

more clicks on twitter

The beauty of this report is that it is pretty flexible. You can tweak it to show the data you want. You will notice that Twitter exports quite a bit of data which will allow you to manipulate it to your needs.

Some additional data you might be interested for your tweets are:

  • Which tweets generated the most hashtag clicks?
  • What are the average impressions per tweet?

If you are crafty with Excel, you can do even more – in fact I am in the middle of creating a full blown Twitter Analytics dashboard. You can definitely get creative with your use of the data.

Determining Statistical Significance For A/B Testing

There will be quite a few variables involved in regards to how long it will take you to reach a statistical significance with your results.

Among these variables are:

  • Twitter audience/reach
  • Typical engagement of your audience

If you have a rather large Twitter audience, you may not have to post your tweets as many times. This will be especially true if your tweets get a decent amount of impressions. With a good amount of impressions, your tweet should be able to generate some interactions.

These interactions will ultimately reflect on how engaged your following is with your tweets.

While you may not want to wait until you have enough data to truly reach statistical significance, more than likely you can still make an informed decision regarding what you are testing.

An easy way to see if you have reach statistical significance between two tweets is to use this free online tool. For the “The number of visitors on this page was” column you will enter your tweet impressions. Then in the “The number of overall conversions was” column you would enter the number of social actions you are monitoring. This could be URL clicks, likes, retweets, engagements, etc.

statistical significance

Once you plug in your numbers, this nifty tool will tell you if your results are statistically significant.

Even if your results are not statistically significant, chances are you can tell when something is a dud or not.

For most, the way you will be able to get to statistical significance is to be sure to post your tweets multiple times to increase the overall impressions of each.

Typically, my tests are not necessarily time sensitive in that the majority of my posts are evergreen in nature. This makes it easy to run my test tweets for an extended time.

Everyone’s situation will be different in regards to how long or how many times you should tweet your tests. I would say a good start is at least 10 times or so. However, it might be best to spot check your Twitter Analytics every few days to see how things are going.

Make Adjustments & Test Again

As I mentioned at the beginning of this post, I am always testing.

And you should be too.

Once you have run your initial tests, no matter what they may be, you should look at ways to improve upon what you found that works.

For example, let’s say you identified a tweet that generated the best engagement rate among the group. Why not test that against the same tweet, but with an image attached? See how that impacts your engagement rate.

Perhaps you didn’t get enough impressions for some of the tweets you were testing. Next time, schedule them more frequently during times that your Twitter audience is most active.

The key is to always look at ways to improve your results each time you test.

Will it always improve? Probably not.

But you won’t know until you test it.

Whether you are testing blog content titles or generating more clicks to your Chrome Extension (get Analytics for Chrome Extensions) – testing can have a dramatic impact on your efforts.

Be sure to grab the Excel file for Twitter testing if you haven’t already. Also, let me know in the comments how you plan to use this approach.

Download My Ready-To-Go Excel File Click here to download my Microsoft Excel file so you can easily view your tweet analytics.