Take a DIY retailer. Kitchens in January. Bathrooms in March. A brand campaign about competence and choice on TV, out of home, occasionally social. Standard brand-building media, planned, tracked, defended in the usual way.
Then look at the paid search account. Sitting alongside “kitchen renovation ideas” and “bathroom trends 2026” are thousands of long-tail terms: bedroom paint, chrome bath taps, white porcelain sink. The reported return on ad spend (ROAS) on these campaigns is often the best line in the whole media plan, sometimes by a wide margin. The temptation is to read that as proof the marketing team has found something the brand campaign hasn’t: a channel that converts.
It hasn’t. Chrome taps aren’t where a DIY brand is built. Nobody forms an opinion of a retailer because of its bid on “white porcelain sink”. These searches capture the last moment before purchase – the instant a consumer who has already decided what they want, and is close to buying it, types the product name into a search bar. The retailer isn’t creating demand at that point. It’s making sure it’s there when demand, created somewhere else entirely, arrives.
That isn’t marketing. It’s the digital equivalent of a well-lit shopfront on the right stretch of high street, or stock on the shelf when the customer walks in. Retailers already budget for that kind of thing – location, merchandising, stock availability – separately from the spend that builds the brand a customer had an opinion of before they ever went looking. Generic long-tail search deserves the same separation. It is a cost of trading, not a cost of building a brand.
The reported ROAS is real. Most of the incrementality isn’t. Last-click attribution – and the ROAS figure sitting in the platform’s own dashboard – gives the final click full credit for a sale that was, in most cases, already going to happen. The consumer who searches “white porcelain sink” and clicks the top paid result would very often have found the same retailer for free – through organic search, a bookmark, or typing the brand name they already had in mind. A well-specified Marketing Mix Model (MMM), run with a proper holdout or geo test, tends to tell a much less flattering story about this activity than the platform’s own reporting: high spend, high reported ROAS, and far less incremental volume than reported once you strip out what would have converted anyway.
This is worth calling out by name. It is, in effect, a Google tax, or a Meta tax – a toll paid to sit in front of demand a brand generated somewhere else, by other means. It is a real and growing cost of doing business online. It is not brand marketing, and holding it to marketing’s usual standards – return on investment (ROI), brand lift, long-term effect – flatters it for the wrong reasons and, worse, can quietly starve the campaigns that are actually building the brand of the budget and credit they deserve.
The fix is a budgeting question as much as a measurement one. Split the spend. Judge brand and category campaigns on what MMM is built to show: incremental volume, longer-term effects, contribution to the base. Judge generic, product-level search on what it actually is – a cost of sale, sitting closer to shipping, stocking or store rent than to advertising – and hold it to a cost-of-sale or margin standard instead. Some marketing directors go a step further and move ownership of that line to the sales or e-commerce team entirely, which tends to sharpen both budgets: marketing stops getting credit it hasn’t earned, and sales stops mistaking its cost of doing business for a marketing return.
None of this is an argument against the spend. Chrome taps still need to convert. It’s an argument for calling it what it is, so the model built to judge marketing effectiveness isn’t quietly grading a sales cost on a marketing curve.
The same misattribution shows up wherever a model is missing the right variable, not just at the last click – see The marketing ROI that was actually a price cut.
If your last-click numbers look suspiciously good and you’re not sure whether that’s brand building or a Google tax in disguise, drop a line to hello@themmmdoctor.com.