At Pearmill, we've had the opportunity to work with companies from the early stages of their growth and advertising journeys all the way to IPO and beyond! One of the common questions we're asked is, when should companies consider building an internal attribution model or buying an attribution tool?
Our Data Engineering team at Pearmill has built quite a few attribution models over the past few years. They listed their top 5 signs that you should look into internal attribution.
An attribution model, either built internally or externally, is a "data model" on how to attribute conversions to specific channels, campaigns, or ads that are involved in the conversion journey of a user.
For example: if a Google user clicks on an ad, and a few days later clicks on an Instagram ad before they decide to convert, which of these channels should count the conversion?
The work goes behind ensuring that you have the right tracking in place, as well as the decision behind how to weight the different channels, is what we're defining as "building an attribution model" in this post.
I'm also limiting this post to Web attribution, as mobile apps are a different world altogether!
If you're spending on Facebook, YouTube, TikTok, or generally modern paid social advertising channels, you may have noticed that they report on view-through conversions.
These are conversions that the network tracks based on someone viewing an ad and then converting on your site. If the percentage of conversions that the channel reports as view-through conversions is high, then it's a good time to look into building an attribution model.
This usually happens because you've either saturated your audience, have a lot of brand affinity, or have high frequency on your creative. When view-throughs are higher than 20-30%, we usually raise a flag to consider building an attribution model, or at the very least only trust click-based conversions in that channel.
While view-throughs should still be used as a signal to allocate budget, you should only build assumptions on view-throughs if you have an internal attribution model (or later on a marketing mix model) that supports those assumptions.
If you have a more niche audience, or if you're nearing market saturation with your ad spend, you may have a lot of overlap between audiences on the different ad networks you're spending your budget on.
For example, if you're running ads on Google and Facebook at the same time, someone may click on a Google Ad first, and a few days later click on a Facebook Ad before converting. In this scenario, both of the ad networks have influenced the conversion – which one of them should you reward the conversion to? Where should you reallocate your budget?
Understanding the conversion journey that users take before converting can help you make better micro decisions on putting your budget behind campaigns and ads that bring the most value.
Building an attribution model in these scenarios can help you optimize your budget toward both campaigns that are better at introducing potential customers to your brand, as well as campaigns that are great at converting those people into customers!
Most ad networks have a short attribution window – usually somewhere between 7-30days. This can be limiting if you have a very long sales cycle.
If you're in B2B, or high-touch consumer products where the time it takes for the company or person to make a purchasing decision is very long, then you may want to consider building an internal attribution model.
It's paramount to make budget allocation decisions, campaign decisions, and creative optimization decisions based on down-funnel metrics. The closer you are to revenue when making these decisions, the better!
If you want to have a more trustworthy set of numbers to look at when it comes to unit economics and the lifetime value of your customers on a per-channel basis, then it's time to build an attribution model.
Channels, and most analytics tools, are limited by the data you feed them and by the attribution windows that they enforce on the data. They're solving a different problem for you, which is how to optimize your ad spend. Your attribution model's job is to help you understand the whole picture – how does the money you spend advertising turn into revenue and eventually profits?
You can build an attribution model in conjunction to cohorts of new customers to analyze the lifetime value of customers of each channel, campaign, or ad. This information, unsurprisingly, can be extremely important when making fundraising decisions or general planning decisions.
I saved the most obvious reason for last! If you're adding up conversions that each channel is reporting, and they're higher than the actual conversions you're seeing in your database or backend tool, then:
For either scenario, it may be a good idea to first run an audit of your tracking infrastructure to ensure you're not double counting, and then consider building an attribution model to help you make decisions on which channels to optimize!
If you need help building an internal attribution model or with other data engineering challenges, reach out! Let’s grab a few minutes together and see if there is potential to unlock growth.