SBS #7 - Basics of Attribution & Campaign Valuation

As the saying goes, there’s lies, damn lies and then there’s statistics. 

While digital marketing channels are able to provide much more granularity in performance reporting than a billboard or a full page spread in a magazine, there's still much to be interpreted from the data. It’s really uncommon to simply have a single source of advertising  as you want to try and meet your customer in the many places that they may be, so understanding how much value each channel is providing is incredibly important in regards to determining where to allocate those ad dollars.

“Attribution” is just fancy marketing-speak for trying to determine how much credit you are giving a particular advertising source. AKA Someone sees a Facebook Ad, then searches for your product on Google and then buys something? How much credit do you apply?

This is a multi-million and possibly billion dollar question. Since hardly any of us have budgets that large or access to vast amounts of 3rd party data, we'll try to answer these questions in ways that are actionable for the small business with limited resources.

As the joke goes, the United States space program spent 2 million dollars developing a pen that can write in sub-zero temperatures, zero gravity and in blistering heat. The Russians used pencils.

We like pencils.

TL;DR: Google Analytics is the most conservative view of channel performance. Each ad platform will take as much credit as they can, so those views are the most generous view of that channel’s performance.

Google Analytics & On-Site Analytics Platforms

For nearly 2 decades, Google analytics has been the de-facto standard of on-site analytics. It doesn’t hurt that it’s free and integrates pretty seamlessly  to most website platforms. It’s the best way to analyze what incoming traffic does on your own website once they get there.

From an attribution perspective, Google Analytics is superior in the sense that it’s aware of all marketing channels that may be involved in the conversion process. Email, Paid Ads, Organic Search, Organic Social, Referral, Direct, etc….

You just have to make sure that your tracking parameters are set up and you're using correct UTM tagging in your inbound URL’s.

Because GA is aware of multiple sources, it’ll automatically divvy up conversion credit when multiple sources are involved in the conversion event. Google Analytics 4 uses a “Data Driven” model by default. This means that Google will use their machine learning secret sauce to divvy up the credit as they see fit. You can adjust this model if you like:

OK great, this makes GA4 the holy grail of attribution right? Well, not quite. GA4 is going to be unaware of anything that doesn’t happen on-site. There’s some impression data sharing from Google Ads to GA, however other marketing sources that don’t directly generate traffic to your site via a trackable link are not going to be represented in GA4.

An example of this would be if you had set up a television ad during the Super Bowl, you’d clearly drive awareness and potential traffic to your site, but since Google Analytics doesn’t have any way to identify that source of traffic, it’ll be mis-attributed to another source. QR codes are being used cleverly these days to try and bridge this gap, but most of it goes unidentified.

Generally speaking, we can call Google Analytics’ attribution the most conservative view of a channel’s performance. It only credits traffic that was clicked-thru onto your website and then makes sure to divvy up that performance to the other channels.

Meta Ads (Facebook & Instagram)

Meta gets it’s performance data from your Conversion Pixel and potentially the API connection between your website and the platform. Unfortunately, there’s an additional bit of funny business here and Meta straight up admits to faking some of your data:

This may be useful if you have hundreds of conversions a day or week, but for most smaller advertisers (ourselves included) there’ll definitely be some “conversions” tracked that don’t result in an actual sales. Try to use the API setup with your website platform to help push back good quality conversion data, but always take Facebook’s conversion reporting with a grain of salt.

Additionally, the default attribution window for Facebook is a 7 day click and 1 day view, meaning that if the user viewed the ad and then converted within 24 hours or clicked the ad and then convered within 7 days, Facebook will happily take the credit for that conversion..

Do be careful with this - especially with retargeting existing customers. The difference between re-engaging a customer and blasting an already-won customer with ads indiscriminately is a fine line. We recommend that you have a much higher expectation of return for returning customers than prospecting customers, in part for this reason.

We can consider the reporting out of the Meta Platform as the most generous credit that the channel can receive. Thru some more robust testing we’ve done with a few other clients, we’ve come to the conclusion that 60-80% of the revenue (as of 2023) that the platform reports can be incremental, assuming that there’s roughly a 85/15 split on the proportion of ad dollars being spent on prospecting vs retargeting.

Google Ads

Similar to Meta, the reporting within Google Ads is only aware of other Google Ads campaigns. Conversion performance will get divvied up between existing campaigns within Google ads (Search, PMax, YouTube, etc…) if a user engages with multiple campaigns in the conversion path. Unlike Google Analytics, however, it won’t be aware of other ad types (like Meta). In past test, there’s a fairly significant impact of advertising on paid social and it’s ability to generate search queries.

Do be careful that you’re measuring the right metric for success. Google loves to call anything it can a “conversion”. Make sure to review that portion in our previous blog post. 

Conclusion

Each reporting platform is going to have it’s own biases and perspectives when it comes to evaluating the performance of a particular campaign. While no single source of data is going to be perfect, understanding these biases will help give an informed view of the actual performance of each channel.

Tactically speaking, be careful of your overall spend levels. Measure your total inputs and your total outputs. We like to set up some reporting to measure the total % of revenue cost. Eg:

(Facebook Ads Cost + Google Ads Cost) / Revenue

Measure this at a weekly or monthly level and you should be able to see if you’re spending more without getting more in revenue overall. If you increase your ad spend and only your % ad spend increases, you know you’re not actually getting anything from your additional ad dollars, regardless of what platform is reporting.