Marketing pixels and A/B testing tools increase the efficiency of your advertising spend by increasing your store's conversion rate. To do so, both depend on tracking purchases accurately—marketing pixels to “learn” who’s purchasing and A/B testing tools to see what website changes lead to more purchases.
Small errors in tracking can lead to large costs and missed opportunities. For example, suppose we’re advertising with Facebook, and we spend $100 to get 100 visitors to our site, three of which make a purchase. Our actual customer acquisition cost is $33.33.
Customer acquisition cost = $100 / 3 customers = $33.33
But, if we under-report our purchases to Facebook by just two, Facebook thinks our customer acquisition cost is $100. That changes who Facebook shows our ads to and what it thinks it can spend efficiently. That could be the difference between having a profitable store and a store that's losing money on each purchase.
However, because we can't send our actual purchase data from Shopify directly to these tools, the data is never going to be perfect. We need a coping strategy to make the data good enough with the limited time we have.
Use our AI Powered chatbot to answer questions using your stores data
We need to set up conversion tracking for an advertising platform before we start spending money on that platform. Using our example, we need to set up the Facebook pixel to track purchases before spending money on Facebook ads.
If we don't, Facebook will optimize our campaigns for clicks because it doesn't have any other information. And the people who click aren’t always the people who purchase.
That doesn’t mean we should set up every marketing pixel we might conceivably use before launching our store. Nor does it mean every pixel must have perfect and complete data for advertising (which is impossible, anyway). It just means we need reasonably accurate tracking with a pixel before spending money on its associated platform. Anything is better than nothing.
There are literally thousands of different marketing pixels and platforms available. Some only let us track simple conversions (who made a purchase). Some, like Facebook, let us track the complete purchase data (revenue, products, prices, tax and shipping costs, shipping city, etc.). Others, like Google Ads, let us track revenue along with purchases, but nothing else.
We want to send as much data as the platform will accept. When we pump more data into these machine learning algorithms, they do a better job of finding potential customers that are more likely to purchase our products.
As an overly simplified example, if we tell Facebook who purchased skinny jeans and who purchased baggy jeans, its algorithm will learn to show skinny jeans to hipsters and baggy jeans to Gen Xers (or whoever wears baggy jeans these days).
I've never seen anyone set up a marketing pixel or A/B testing tool perfectly the first time. Something always goes wrong. What's more, things change constantly in our store and in the marketing platforms. As such, we need to continuously monitor the accuracy of our tracking tools. Don’t expect to set it and forget it.
To do so, we should set up charts that make spot-checking easy, like the one below.
To test and monitor our pixel data, we need to compare it to our actual purchase data, which comes from Shopify.
If we compare Facebook data to Google Analytics and they don’t match, which one is wrong? There’s no way to tell. In fact, they could both be wrong. That’s why we need to compare them both to our actual purchase data from Shopify.
Again, the pixel data (unlike Shopify data) is never going to be perfect. Some people use ad blockers, and other weird stuff can happen. Our goal should be to keep our pixel data within 10% of our actual data. If it’s outside that, something might be wrong, and we need to check it.
Since it takes time to get each advertising platform’s conversion tracking right, start with just one or two of them. Add more as your store becomes more sophisticated and you start to leverage more platforms.
All these marketing pixels will automatically create reports that show you how wonderful they're doing at finding new customers. But they all use different attribution models.
For instance, if Facebook just shows an ad to someone (and they don’t click), then a week later that person makes a purchase, Facebook will take credit for that purchase in its reporting.
That’s why we need to track and model our own marketing performance separately from what the platforms tell us. To do so, we need to use clear, consistent UTM parameters. Here's a spreadsheet for that purpose.
As I mentioned, there are literally thousands of different marketing and A/B testing tools available. By far, the most common are Facebook, Google Ads, and Google Optimize. Here are how they work and some tips for using them.
When we install the Facebook pixel, it tracks all the purchases (and more) that happen on our site. It doesn't matter if a visitor comes from a Facebook ad or not. For this reason, we should set up the pixel as soon as we can because we’re almost certainly going to use Facebook ads eventually.
After we’ve set up the pixel, Facebook combines its data specific to our site with its massive user dataset to find potential customers. The results can be counterintuitive. We may think our product is a great fit for 32-year-old dads who like to rock climb, but Facebook may discover that it appeals to 45-year-old moms obsessed with Christmas lights. The internet is a weird place.
The Google Ads pixel (formerly Adwords) tracks data for both search and display ads. Even though there’s a way to track conversion for Google Ads using Google Analytics, it’s best to track them separately.
When doing so, you’ll see that Google Ads and Google Analytics report different conversions. That’s not a mistake. Google Ads uses a different attribution model than Google Analytics, so the results are different. Compare them both to your Shopify order data to see if anything is wrong.
Of the various A/B testing tools—Google Optimize, VWO, Optimizely—Google Optimize is the most popular because it's free and integrates with the Google Marketing Platform suite of tools, including Google Analytics.
Google Optimize lets us create randomized control trials called experiences. For each experience, we define an objective, i.e., the conversion we want to optimize for. The conversion is defined in Google Analytics as either a traditional Goal or an Enhanced Ecommerce event.
For Shopify stores, we almost always want to optimize for purchases or revenue. As such, we should use the Enhanced Ecommerce purchase event (called transactions in Google Analytics) and not a traditional Goal, even if we have one defined for purchases. Doing so gives us both more accurate and more complete data.
A/B testing is powerful but can be tricky to execute correctly. Outside of thorny technical issues, it can be difficult to know what’s worth testing in the first place. Check out Thesis Testing, an A/B testing company, if you need help.
Both marketing pixels and A/B testing tools depend on accurate and complete purchase data to improve our site's conversion rate. Small changes make a big difference. Improving our conversion rate from 1% to 2% cuts our customer acquisition costs in half. That’s oftentimes the difference between being a profitable ecommerce store and losing money on each transaction.
Spend the time to get the most out of these tools. Create systems to make sure their data remains accurate, and add and improve your tracking as you go.