Top-of-funnel content isn't dead. It just stopped being measured by clicks.
Top-of-funnel content traffic has fallen, but the job it does has changed rather than disappeared. How to measure, defend and rebuild TOFU content in 2027.
The argument doing the rounds is that top-of-funnel content is finished. AI Overviews have eaten the educational query, so the money should move to bottom-of-funnel pages that describe your product in forensic detail.
The first half of that is true. The second half does not follow from it.
What the data actually says
The numbers behind the pessimistic case are real. Organic search clicks have fallen around 42% since AI Overviews began expanding. Large language models cite plain educational content roughly 12% of the time. If your top-of-funnel content was built to acquire sessions from informational queries, that machine has broken and it is not coming back.
But the most interesting finding in Adam Gnuse's analysis for Search Engine Land — 30 major publishers across nine industries, year-on-year organic traffic — is the one being skipped over. Top-of-funnel and middle-of-funnel content tracked together within each industry. YMYL sectors (finance, healthcare, legal, consumer tech) fell hard across both. B2B tech grew across both.
That matters. If funnel stage were the cause, you would expect top-of-funnel to collapse while middle-of-funnel held, in every industry. It didn't. Industry predicted the outcome better than funnel position did. Which means "top-of-funnel is dead" is describing a real decline and naming the wrong variable.
The job it used to do, and the job it does now
The old job was simple and simply measured: acquire the session. Rank for a volume keyword, count the traffic, report the line going up.
The new job is three things, none of which appear on a sessions chart.
- It establishes the entity and the topical range that your bottom-of-funnel pages and your AI citations depend on. Models do not decide you are credible on a decision query in isolation.
- It gives something to corroborate against. A brand that only publishes pages about its own product is a brand with nothing to cross-reference.
- It serves the readers who do still click — a smaller group than before, and a considerably more motivated one.
The content did not stop working. The metric stopped measuring it. That distinction is the whole argument, and it is why the budget is under threat: a line item that loses its number loses its defender.
Why cutting it entirely is a trap
Three reasons, in ascending order of how much they will cost you.
First, the advice is self-defeating at scale. If everyone moves to bottom-of-funnel only, bottom-of-funnel gets crowded and the differentiation you moved there for evaporates. Advice that stops working when it is widely followed is a trend, not a strategy.
Second, your bottom-of-funnel pages rank on topical authority they did not build by themselves. Strip out the surrounding content and you degrade the thing you kept, on a delay long enough that nobody connects the two.
Third, and most directly: you will not be cited on a decision query if there is no basis for thinking you understand the category. Citation is a judgement about the source, not just the page.
What to keep and what to kill
Keep
- Expert-driven content where a named person with genuine experience says something a summary cannot reproduce. This is the E-E-A-T route and it is the one that survives, because it is the one an AI overview cannot flatten into three bullet points.
- Interactive tools — calculators, screeners, configurators, comparison builders. A model can summarise your article. It cannot summarise your calculator, which means it has to link to it.
- Local and geographically specific guidance, which has held up comparatively well.
- Content that answers the question your bottom-of-funnel pages assume the reader already understands. Every product page has one of these, and it is usually unwritten.
Kill
- Definitional explainers with no first-hand input. "What is X" was always a weak page
and is now a worthless one.
- Anything produced at scale with no editorial layer. Google confirmed three spam updates in 2026 — March, June and August — and the August one made a top-10 URL roughly 1.8 times more likely to drop out of the top 100 than in a normal week. That is not a climate for volume plays.
- Content written to a keyword with no view on who reads it or what they do next.
How to measure it instead
If you keep reporting top-of-funnel content on sessions, you will lose the budget, and you will deserve to. Report these instead.
Citation presence across AI surfaces at topic-cluster level rather than page level — the cluster is the unit that earns the citation.
- Assisted contribution: do users who touch educational content convert at a different rate later than those who don't?
- Bottom-of-funnel ranking movement after a top-of-funnel cluster ships. This is the cleanest causal read available to you and almost nobody runs it.
- Branded search volume, which is the crudest of the four and still more honest than a sessions chart.
None of these are as clean as the number they replace. That is the actual problem here. The measurement got harder, and "cut it" is what happens to a budget line when its metric stops making the case for it.
What to do this week
- Pull your top-of-funnel pages and segment them against a realistic expectation for your industry, not against a blanket decline everyone is quoting.
- Identify every page with no first-hand input, no named author and no original view. That is your cut list, and consolidating those pages beats deleting them.
- Find one topic where a tool would out-perform an article, and scope it.
- Ship one expert-led piece with a named author and something in it that could not have been written from a search results page.
- Change the reporting line before someone else changes the budget line. This is the
one with a deadline attached, and the deadline is your next planning meeting.
FAQ
Should I delete underperforming top-of-funnel pages?
Not reflexively. Consolidate before you delete — several thin pages on adjacent topics
usually make one page worth keeping, and you retain the links and the history.
Does this apply to B2B?
The available data suggests B2B tech has held up considerably better than YMYL sectors on both top and middle of funnel. Check your own numbers before importing anyone else's decline.
Is AI-assisted writing off the table?
No — unedited AI content at scale is. The distinction Google is enforcing is editorial, not tooling. The useful test is still whether the page deserved to be published.
23.png)
