There is a number in your marketing reports that everyone trusts, and it is quietly steering your budget in the wrong direction. It is the one that says this channel drove that sale. Last-click attribution, or some lightly dressed-up version of it, still decides where a large share of spend goes in most established brands. It is comforting, it is simple, and it is wrong often enough to be dangerous.

The problem is not that attribution is hard, though it is. The problem is that the easiest number to produce is also the most flattering to the wrong channels, and finance teams make real budget decisions on it. The result is a slow, invisible misallocation: the channels that close sales get the credit, the channels that create demand get cut, and nobody notices until growth stalls.

Why last-click feels right and is wrong

Last-click gives full credit to whatever a customer touched immediately before buying. That is almost always a harvesting channel: branded search, retargeting, an email to someone already deep in the funnel. These channels look spectacular in a report because they are standing at the finish line collecting demand that something else created.

Meanwhile the work that actually built the intent, the brand campaign, the content, the upper-funnel prospecting, gets almost no credit, because it rarely sits in the final click. So your paid media reporting tells you to pour more into retargeting and brand keywords, and to trim the prospecting that fed them. Do that for a few quarters and you have optimised your way into a smaller market.

The branded search line is the clearest example. In a last-click report it looks like one of your best-performing channels, all the way to the bank. Then a brand runs a two-week holdout, pauses brand bidding in matched regions, and total branded revenue barely moves because most of those people were going to find you anyway. The spend was not winning customers, it was buying clicks you already owned. The dashboard never showed that, because the dashboard cannot.

The cost of believing it

This is not a theoretical concern. It shows up as three predictable mistakes. You over-invest in retargeting audiences that would have converted anyway. You bid hard on your own brand name when much of that traffic was already yours. And you defund the top of the funnel because it cannot prove itself in a last-click world, which shrinks the pool of future demand the rest of your spend depends on.

Each decision looks rational in isolation. The report said the harvesting channel returned better, so you backed it. The damage is only visible in aggregate, months later, when blended performance softens and no single channel looks to blame.

This is why the problem is so persistent in well-run businesses. It is not caused by careless marketers, it is caused by careful ones following an honest-looking number to its logical conclusion. The better your team is at hitting the targets the dashboard sets, the faster they will optimise toward the bottom of the funnel and away from the demand creation that keeps the machine fed. A measurement error that punishes your most disciplined people is far more dangerous than one that is obviously broken, because nobody questions a number that keeps appearing to work.

Incrementality is the only honest question

The question that actually matters is not "which channel touched the sale". It is "what would have happened if we had not spent this money". That is incrementality, and it is the number a finance director should be asking for, because it is the only one that maps to return on capital rather than to the accident of who got the last click.

Incrementality is measurable, just not through the attribution dashboard. Geo holdout tests, where you switch a channel off in matched regions and watch what happens to total revenue, are the cleanest read. Conversion-lift studies and properly designed audience experiments get you there in digital channels. None of it is exotic, and all of it tells you something the attribution model structurally cannot: the true lift of the spend.

Three lenses, not one verdict

No single method is complete, which is exactly why relying on one dashboard is the mistake. Mature measurement runs three lenses at once and looks for where they agree. Attribution gives you fast, granular signal for day-to-day optimisation, as long as you treat it as indicative rather than true. Experiments, the holdouts and lift tests, give you the causal read on what a channel actually adds. Media-mix modelling gives you the top-down view across everything, including the offline and brand effects no pixel ever sees.

When all three point the same way, you can move budget with confidence. When they disagree, that disagreement is the most valuable thing in your reporting, because it is pointing at exactly where your instincts are being misled. The brands that compound growth are not the ones with the cleverest attribution model. They are the ones disciplined enough to hold these three views side by side and trust the experiments when it counts.

What this changes on Monday

You do not need to throw out attribution. You need to demote it. Treat it as a directional input, not the verdict, and triangulate it with experiments and a simple media-mix view. The brands that get this right run a steady cadence of holdout tests rather than debating attribution windows, and they judge every major channel on incremental return at least once a year.

The single most useful habit is also the simplest. Before you cut or scale a channel, look at the change in total revenue, not the channel's own self-reported figure. If switching a channel off makes the whole number worse than its attribution suggested, it was doing more than the dashboard gave it credit for. It is the kind of measurement discipline we build into the marketing programmes we run for brands like Lake Country, where the goal is growth you can actually bank, not a tidy report.

The conversation to have with finance

The reason this persists is that last-click gives finance a clean, channel-by-channel story, and incrementality gives them a messier, more honest one. That trade feels uncomfortable until you frame it correctly: you are not asking them to trust a worse number, you are offering them the one number that survives scrutiny. Finance leaders are not wedded to attribution, they are wedded to certainty, and incrementality gives them a more durable version of it than a model that quietly rewards whichever channel happened to be standing closest to the sale.

Reframe the conversation away from attribution models and toward two things finance already understands: incremental return and payback period. Agree a north-star that everyone trusts, fund the tests that keep it honest, and stop relitigating which pixel fired last. The brands that make this shift stop starving their own growth, because they finally stop paying their harvesting channels for work the rest of the budget did.

None of this requires a heroic measurement overhaul, just the willingness to stop trusting the comfortable number. If you want a second pair of eyes on whether your reporting is flattering the wrong channels, that is exactly what our marketing consultation is for.