Why Facebook’s attribution data is mostly guesswork

One of the biggest challenges when running ads is measuring accurate attribution data. This is difficult for a number of reasons, including the fact that different systems calculate metrics and attribution differently.

Differences in measurement approach mean that SegMetric’s will typically disagree with the sales according to facebook. They include:

1. Facebook measures thank-you page visits or web events, while SegMetrics looks at processed payments
2. Facebook only attributes conversions to Facebook clicks, while SegMetrics attributes across all of your ad platforms
3. View-through attributions will take credit for a sale, even if a customer barely glanced at the ad
4. Facebook has a 7-day attribution window by default, which will miss sales that take over a week to convert

Let’s look at these in more detail…

Web events vs Payment Events

Facebook really only attributes checkout events, not real revenue. 

In marketing analytics, it is important to define a Source of Truth. Ideally, your Source of Truth should be as close to reality as possible. When measuring sales this means processed payments, or when measuring leads it means contacts in your email marketing system.

Facebook’s Pixel can’t access either of these sources of truth. It looks instead at javascript events or thank-you page visits. That means, it will still report that a sale has occurred even if 

👎 the payment fails
👎 the purchase is refunded
👎 the order is canceled later on

These uncompleted purchases can quickly burn through your profit margin, so it is vital to use an attribution tool that helps you avoid fake “sales”.

Similarly, leads can be double-counted over multiple browsers and sessions, even if there’s only one contact in your email marketing system.

SegMetrics hooks directly into your payment provider and email platforms. That lets us only report on actual sales that turned into money in the bank or leads that made it into your subscriber lists.

Facebook Focused vs Holistic Attribution

Accurately defining which clicks are responsible for a conversion is the most challenging part of marketing attribution.

Facebook makes things easy for itself…by pretending other ad platforms don’t exist.

Facebook’s first and last touch attribution only looks at the attribution within their platform. They will still claim credit for a sale even if there were other clicks, such as on Google or AdRoll, since their Pixel ignores these other touchpoints. We call this greedy attribution.

SegMetrics instead looks at all possible touchpoints and attributes leads and purchases to the corresponding ads based on your selected attribution model. This creates a holistic attribution system, which gives you true attribution across all of your marketing tools.

7-Day vs Lifetime Attribution

Facebook is limited to a 7-day click attribution window.

This means the attribution data is missing anyone who took their time before making a purchase. You can check your own typical time in Google Analytics, using their Time Lag report. For SegMetrics, it looks like this:

For our own funnels, Google reports roughly half of signups occur after 7-days. For e-commerce, we have typically seen an average of around 14 days from first ad click to purchase.

Note:  this stat can be distorted by people who came back after 30 days, whose cookie will have reset so that signup will now be measured as same day.

The 7-day window is dangerous as it can mislead marketers. In an attempt to follow the data, advertisers will be biased towards impulse buyers, disregarding potential customers who take a slower approach to making a purchase.

Facebook’s Inflated Metrics

Facebook has a history of dodgy data…

In 2018, they paid $40 million in a class-action lawsuit for inflating viewership metrics of their new video ads service by 150% to 900%.

While it disputes the lawsuit, Facebook did admit in 2016 to inflating its average video viewing figures by only counting views that lasted longer than three seconds.

— Fortune

This focus on video, unfortunately, decimated the written journalism industry, as the high-value metrics convinced publishers to move to video.

Unfortunately, that’s not the only time Facebook has been caught muddling the attribution metrics — and were sued again for misrepresentation of their “estimated potential reach” data.

The lawsuit cites an example when Facebook told advertisers it had a “Potential Reach” of 230 million adults, even though census data showed only 170 million were using Facebook. – CNBC

Then they blundered again with the “Conversion Lift” tool. A bug was found that lead to over attributed view-through sales. In some instances, Facebook are compensating advertisers for vouchers for millions of dollars of ad credit.

It’s reasons like these that it’s difficult to implicitly trust Facebook’s black-box data, and why it’s important to use data in holistically attributed, accessible systems like SegMetrics.

How to Determine the Truth

One of the core tenets of SegMetrics’ reporting is that all data should be available and accessible. You can see exactly which leads, customers and purchases are making up the data in each report.

You can then click into a contact to see details of their journey, down to every ad click, tag and purchase they have made. It gives you the ability to see how Facebook ads truly contributed to individual sales, instead of taking them on good faith.

Step 0) Trust but verify

We don’t want blind faith from our users. You should first verify your SegMetrics setup, such as that your ad_ids are pulling in and that your Pixel is firing for accurate attribution.

Start by going through your custom setup guide and the troubleshooting SOP. If you’re confused about any step, you’re welcome to email support@segmetrics.io.

Once you’re set up, there are a couple of simple steps you can take to quickly understand where your conversions are coming from, and what your true ad attribution is. Check out these helpful articles we have on Facebook Lead Ad Tracking and Setting Up Facebook Ad Tracking.

Step 1) Filter to show only Facebook Ads

In the default report, SegMetrics will report across all of your ad platforms, and attribute leads and purchases correctly to the platform that brought them in. Facebook defaults to assigning all multi-touch attributions to Facebook.

To only show purchases that involved a Facebook ad click on their journey, go ahead and add a report filter.

Then make sure your date ranges are matching up.

Step 2) Flip the Script

The easiest way to get our true attribution numbers is to flip the script, and look at the leads or purchases for a single day, and then see where each one came from.

This can be a little time consuming, which is why we recommend looking at a day or two of day, but it’s the best way to understand the full customer journey of your ads, and verify what Facebook is reporting.

From the New Leads or Orders report, choose a specific 1 or 2 day timeframe that you can easily confirm with live data.

Next, choose the Contacts or Purchases table to get a list of all the contacts or customers that came in during that day. You’ll see their first-touch information in the report, which gives you a quick view if you are getting tracking or not.

Click the Open icon next to each of the names to open the contact’s full customer journey.

Here you can see all the clicks that a contact has made during their marketing lifecycle, and quickly verify which purchases and conversion events are attributed to which ads.

Conclusion

Multi-Touch Attribution will always be a challenge for digital marketers. That’s why the most critical tool a marketer can have is an accurate attribution platform that allows you to verify the metrics it reports on — by being open about the data it collects, how it attributes, and its limitations.

There will never be a perfect marketing tool. But SegMetrics is dedicated to being open with its attribution and data, so that you can trust the marketing decisions you make.

Because you can’t reliably improve your marketing if you can’t trust your reporting platform to have accurate attribution data.