Your rented new-business pipeline

This is panel data on observed behaviour rather than survey response, covering both desktop and mobile across the first four months of 2026.

On the cause, the researchers are deliberately non-committal. They say it might relate to European rules on search engines promoting their own services. They also say it might be cultural, or about language, or simply about which features launched where first. Some write-ups have added a firmer reason than the study supports. We are not going to repeat it. The gap is measured. The cause is not.

So a UK plan built on US benchmarks will expect more organic traffic than it gets.

B2B is hit hardest. Business buyers mostly search for facts, and facts are what AI answers do well. In business technology, AI Overviews went from 36% to 82% of searches in twelve months.

Where a summary appears, position-one organic click-through falls to around 1.6%. Being cited inside the answer helps - cited brands see roughly 35% higher click-through than uncited ones on the same query - but the base is small enough that it is worth stating plainly: a 35% uplift on a 1% base is a 1.35% base. That is a useful signal, not a pipeline.

None of which is an argument against doing the optimisation work. It is an argument against treating it as the load-bearing part of a 2027 plan.

Separately, 73% of B2B websites recorded significant organic traffic losses, averaging 34% year on year. That is the supply side of the same equation - the clicks leaving organic are the clicks arriving in the auction as demand.

The planning implication is straightforward. A cost-per-acquisition assumption carried forward flat from this year is not neutral; it is an optimistic forecast, and worth labelling as one in the budget paper.

This is the mechanism most worth understanding, because it inverts the intuition that a small budget simply buys proportionally less. It does not. It buys less efficiently, and the efficiency gap widens as a better-funded competitor's model keeps learning while yours does not.

The entry cost is highest exactly where it hurts. Cross-industry average cost per click is around $5.26. Business and finance categories run $17 and above. So the sectors with the steepest prices are also the ones demanding the largest minimum commitment before the machinery works at all.

The honest counter, which you will hear and should be ready for: disciplined targeting, tight match types and good creative do recover part of this. That is true. The defensible claim is not that smaller advertisers cannot win the auction. It is that below the data floor the disadvantage compounds rather than remaining flat, and skill changes the intercept rather than the slope.

This is arithmetic, not a survey. If a firm changes supplier every five years, a fifth of the market is in play each year. On a 90-day sales cycle, that is about 5% in any quarter.

This is the strategic case for reaching people who are not looking. It is also why performance dashboards systematically understate the problem: they report on the 5% because that is the only group they can see. The 95% never appears as a line item, so the gap never appears as a risk.

The argument is about concentration, not channel loyalty. Paid and organic search should keep doing what they do well. But a pipeline that comes entirely from routes owned by three or four companies has a dependency no dashboard reports. And those companies have already shown how they revise the terms.

For comparison, the US Association of National Advertisers - formerly the Direct Marketing Association - puts direct mail response at 4.4% against 0.12% for email. Read that carefully: 4.4% is the upper end of the prospect-list range rather than a universal average, and the channels are measured by different methods. The order-of-magnitude difference is real; the precision is not. It is worth citing accurately or not at all.

Corpdata has supplied UK business contact data since 1992. Named decision-makers, picked by role, sector, size and location, for post, telephone and email. Continually verified.