A CFO asks a fair question: which marketing dollars produced last quarter’s revenue?
Marketing teams can answer part of it. Form fills are visible. A gated spec sheet can be tied to a sample request. Trade show badge scans can become opportunities. What is harder to show is the architect who specified a product eighteen months ago because a rep spent twenty minutes explaining it, or the contractor who asked a branch manager for a brand by name after seeing it on jobsites for a decade.
So the part that can be proven gets presented. And the part that can be proven is often the part that gets funded next year.
That is the real trouble with building products marketing attribution. It is not necessarily inaccurate; it is selective. It measures a narrow slice of how a product actually gets sold, then quietly reallocates budget toward that slice.
The tracking got better. The picture got worse.
In April 2025, Google confirmed it would keep third-party cookies in Chrome after all. No deprecation, no standalone prompt, status quo maintained. Five years of cookieless contingency planning, and the industry landed roughly where it started.
Here is what we find interesting: for building products manufacturers, almost nothing changed. The measurement gap many teams feel every quarter did not close when the cookie was rescued, because the cookie was never the thing that was broken.
The gap is structural. A product gets specified by an architect, stocked by a distributor, sold by a dealer, and installed by a contractor. That can mean three or four handoffs between marketing and revenue, and not one of them necessarily writes back to the CRM. No amount of pixel fidelity fixes a chain where the person who chose the product never visited the website and the person who sold it never reported why the order came in.
We spent five years bracing for a tracking problem. The channel handed us a plumbing problem.
What your model is actually measuring
The most useful frame we have found for this comes from outside our industry. Professor John Dawes of the Ehrenberg-Bass Institute, working with the LinkedIn B2B Institute, popularized what is now called the 95-5 rule: at any given moment, roughly 5% of business buyers are actually in market. The other 95% are not buying anything today, from anyone.
That reframes the whole exercise.
An attribution model can only see behavior. Behavior happens when someone is in market. So the model becomes a very precise instrument pointed at 5% of the audience, while remaining structurally blind to the other 95%: the specifiers, dealers, and contractors who will buy eventually and are forming an opinion today about whether a company is credible, familiar, and worth considering.
Gartner’s B2B buying research reinforces the point from another angle. Buying groups typically run six to ten people, and buyers spend only about 17% of their total buying journey meeting with potential suppliers at all. Divide that 17% across every vendor under consideration and the sliver of the journey a manufacturer can directly observe gets very thin.
The model is not wrong. It is small. And we keep treating its output as if it were the whole picture. For the mechanics of how those touches actually chain together, our earlier look at mapping the customer journey is a useful companion to this piece.
An attribution model is a precise instrument pointed at 5% of the market. The danger is not the precision. It is mistaking it for coverage.
The doom loop this creates
Here is where a measurement quirk turns into a business problem.
What gets measured gets funded. Every marketing leader knows this, and in a tight year it stops being a maxim and becomes a budget meeting. Demand capture shows a cost per lead. Brand shows impressions. When leadership is choosing, the line with a number attached is easier to defend.
So brand gets trimmed. Not eliminated, just trimmed. The trade publication schedule gets thinner, architect lunch-and-learns get cut back, and jobsite photography disappears from the budget.
Nothing bad happens for about two quarters.
Then leads get more expensive. Paid search costs climb because fewer people are searching for the brand by name, which means the company is buying strangers at auction instead of capturing people who already had it in mind. Conversion rates soften because distributor reps have less familiarity with the brand. Sales cycles stretch because more conversations start from zero.
And because the attribution model still only sees capture, the diagnosis becomes: capture is underperforming. More money moves into capture. The loop tightens.
Binet and Field’s B2B research, commissioned by LinkedIn, put the effective allocation at roughly 46% brand building and 54% sales activation. That is not the 60/40 split they found in consumer categories, but it is nowhere near the 10/90 many building products budgets have quietly drifted toward. Their broader warning is the one worth remembering: activation-only strategies deliver in the short term and decline in the long term, while also increasing price sensitivity. In a tariff-volatile year, becoming the brand people only buy on price is an expensive place to end up. We have written before about balancing demand gen and lead gen rather than treating them as rivals, and this is the budget consequence of getting that balance wrong.
Stop trying to attribute brand. Start measuring it.
Our point of view can sound like a retreat before it sounds like a strategy: brand awareness is not going to be cleanly attributed to revenue in an indirect channel. We recommend stopping the chase for perfect attribution. Every dollar spent pursuing a unified model that finally credits an architect program is a dollar not spent on the architect program itself.
Brand does not need perfect attribution. It needs measurement. Those are different things, and conflating them is one of the more expensive myths about building product branding still circulating in our industry.
Attribution asks: which touch caused this order? Measurement asks: is the thing we are trying to build getting bigger? The second question is answerable. We use four ways to answer it:
Branded search volume
Branded search volume is one of the cleanest brand proxies available, and it is sitting in Search Console for free. When more people type a company name into Google month over month, awareness is moving. We track it as a trend line against brand spend, not as a conversion. It will not identify which billboard worked. It will indicate whether the broader program is gaining ground.
Share of spec
For companies selling into commercial or A&D markets, reps can log specifications the way sales logs opportunities. That means tracking every project where the product landed in the documents, not just won specs. Specs are a leading indicator of revenue twelve to twenty-four months out, and they are among the most direct evidence that brand work reached the person who decides.
What the channel actually hears
Distributor and dealer reps know something a dashboard does not: whether contractors are walking in and asking for a brand by name. Two questions on a quarterly channel survey can help capture that signal. Are customers requesting our brand unprompted? Has that gone up or down since last quarter? It is qualitative and imprecise, but it is still useful intelligence in a channel where the manufacturer does not control the last click.
Lead quality, not lead volume
If brand is working, the leads that do come in should arrive warmer. We look at the ratio of qualified to raw leads and average deal size over time. A brand-starved pipeline can get busier and worse at the same time, a pattern the volume metric hides completely.
None of these four measures will satisfy someone demanding a single ROI number. Together, they can show whether a brand is compounding or eroding, which is the question that actually matters.
Then fix the part attribution can see
While we stop chasing what cannot be cleanly attributed, we should get serious about the measurable part many manufacturers are still fumbling: the handoff.
A manufacturer generates a lead and routes it to a dealer. What happens next is, for many manufacturers we work with, a genuine mystery. Did anyone call? Did it close? Did it close for a competitor’s product because the dealer stocks both?
That is not a philosophical measurement problem. It is a process gap, and it is fixable this year without a data warehouse. We can route leads with a required disposition field, give partners something they actually value in return for closing the loop, such as co-op dollars tied to reported outcomes rather than activities, and follow up on leads thirty days later to learn what happened.
Manufacturers who close that loop typically discover two uncomfortable things: a meaningful share of leads were never worked at all, and the ones that did close took longer than anybody assumed. Both are worth knowing. Neither requires a new platform.
Where to start
A practical place to start is with three numbers this quarter: branded search volume, specs logged, and dealer-reported unprompted requests. Baseline them, then do nothing else with them for ninety days except track the trend.
At the next budget conversation, those three trend lines can sit alongside cost per lead. They do not replace attribution; they give it context. The argument is not that brand is unmeasurable and therefore beyond scrutiny. It is that brand is measurable on a different clock, and a company that only funds what reports monthly will slowly stop being a brand the market asks for.
That is a fixable problem. But we cannot fix what we have agreed not to count.
