Is Your Attribution Model Holding Back Profitable Growth?
By Jerome Yip
Why It Matters
Attribution reports only credit what they can see, so a channel can look too weak to scale when it’s actually creating demand that gets credited elsewhere, while another can look great simply because it’s capturing sales that were already coming. That means budget decisions built on platform ROAS or GA4 data-driven attribution can quietly steer you away from your best growth opportunities. An MMM measures against total business results instead, so it can show which channels are truly underfunded, where spend is capturing existing demand, and where the next dollar will return the most. In the case study here, an MMM paired with a geo experiment gave a nine-figure retailer the evidence to invest further in Meta, a channel their attribution said to hold back on.
Who This Is For
Built for Founders members managing large ecommerce accounts across multiple paid channels, especially anyone whose customers move between ads, devices, and organic visits before buying. If your attribution report is telling you to cut a channel and you’re not sure whether the problem is performance or incomplete measurement, this one’s for you.
If you manage a large ecommerce account across several paid channels, with customers moving between ads, devices, and organic visits before buying, your attribution reports may be limiting where you’re willing to spend more.
There’s a pretty simple reason: a channel can look too weak to scale when some of the sales it helps generate are credited elsewhere (conversely, a channel can be overvalued when it takes credit for non-incremental sales).
A marketing mix model (MMM) can help identify those opportunities and give you evidence to use your budget more effectively.
1. The ads that create demand may receive too little credit
Someone sees a Meta ad, doesn’t click, then searches for your brand and buys through a Google ad. Standard GA4 website tracking sees the Google visit and purchase, but not the earlier Meta impression.
Someone else clicks a Meta ad on their phone, browses, and leaves. Later, they search for your brand on a desktop, click a Google ad, and buy. Without a login or shared identifier connecting the devices, GA4 may record the Meta visit without linking it to the purchase.
Neither of those scenarios tells us whether the customer would have bought without the advertising.
Data-driven attribution estimates how interactions contribute using the path data available to it. It doesn’t give you a complete record of the customer’s journey.
If those gaps understate a channel’s contribution, its reported return could discourage you from investing more.
2. High ROAS can include pre-existing demand
A customer who already intended to buy might click your ad instead of an organic result. The purchase improves reported ROAS but doesn’t necessarily represent an additional sale.
Paid media can also contribute to sales reported as organic. Someone might discover your brand through an ad and return later through organic search to buy.
The channel that records the purchase doesn’t tell you whether advertising created the demand or captured it.
3. Past ROAS doesn’t tell you what the next increase will return
A channel can perform well at its current budget and become less efficient as spending rises. Ranking channels by average ROAS doesn’t tell you where the next increase will add the most profit.
MMM response curves estimate how results change at different spending levels. That helps you choose where to test an increase and where returns may be flattening.
Smarter marketing, delivered.
Join thousands of marketers getting straight-talk strategies from Andrew Foxwell and team.
How an MMM helps you find room to scale
An MMM compares changes in channel spending with total business results over time, accounting for other influences such as seasonality and promotions.
Because it uses total sales or margin, it includes purchases that finish through organic channels. It doesn’t need to connect every ad interaction to an order.
I would use those estimates to identify underfunded channels, question spend that may be capturing existing demand, and decide where to test the next increase.
An MMM needs reliable business totals, enough history to observe seasonal patterns, and variation in channel spending. If every channel’s budget always moves together, separating their contributions becomes harder.
How MMM benefitted one retailer
I built an MMM for a nine-figure ecommerce retailer with substantial organic demand. GA4 data-driven attribution made Google look like the stronger investment. Meta’s attributed margin barely covered its ad spend.
The MMM results suggested Meta was contributing substantially more margin. I tested that finding with a geo experiment. The results gave the team evidence to invest further in Meta.
Your next opportunity to scale may be in a channel your attribution report tells you to cut. Before you write it off, test whether the problem is performance or incomplete measurement. For an account at this scale, answering that question could be some of the most valuable work you do.
Six years of Slack debate from 500+ paid-media operators, now queryable from inside Claude or Cursor. No seat in the community required.
