Stop reporting content by traffic. Report it by the deals it touches.
marketing November 21, 2026 · Mintec

Stop reporting content by traffic. Report it by the deals it touches.

Google told CMOs to plan on leads, not proxy metrics. That takes a page-to-deal join most teams never built. Here is how to build it, and what the audit finds.

If your content report ends at clicks, it tells you which pages get attention, not which pages produce revenue. The fix is a join: one key that connects a published URL to a CRM opportunity. Once that key exists, sort every page into four quadrants on attention against opportunities touched — and three of those quadrants get a different budget decision.

That is the whole method. What follows is the part that makes it survive contact with a real reporting stack.

What Google actually said, and the half it left to you

On November 19, Google's Think with Google account posted a line aimed at CMOs: "AI didn't make your brand invisible in Search. Your content did." It paired "Stop the 'Top 10' listicle" with "start leading with first-hand expertise." Search Engine Journal covered the reaction the next day, including the SEOs who read the first half as a ban on lists.

The article behind the post is more specific. Brendon Kraham, Google's VP of Search & Commerce Global Ads Solutions, published it on Think with Google on October 14. What to stop: a running store publishing a generic "Top 10 Things to Consider." What to do instead: a deep-dive video on why one customer's running shoes collapsed. Same store, same category, different evidentiary standard — one could have been written by anyone.

Then comes the sentence that matters more than the listicle debate. Kraham tells CMOs to keep planning "based on real business goals like leads or sales, not unproven proxy metrics." Google's Search documentation had already drawn the line in May, when its guide to optimizing for generative AI features used "7 Tips for First-Time Homebuyers" as commodity content — "often based on common knowledge" — against a first-hand piece about waiving a home inspection.

Read both halves together and you get an instruction with a missing step. Stop writing commodity pages, and stop planning on proxy metrics. Fine. But which of your existing pages are commodity, and which of them touched a deal? Nobody answers the second question from a traffic report, which is exactly why the proxy metrics survived this long.

Three systems, three timestamps, no shared key

Content reporting defaults to impressions for a structural reason: the three systems holding the answer do not talk to each other.

  • Search Console knows the page and the query behind the click. It knows nothing after the click, because that is the edge of its world.
  • Analytics knows what happened after the click, but a key event is a behavioral marker someone defined once — a form submit, a chat open. It is not an opportunity, and a B2B funnel has two to five human steps between the submit and the deal.
  • The CRM knows the deal, its stage and its value. It usually knows nothing about the page that started it, unless someone built a field for that on purpose.

So teams divide a number from one system by a number from another: sessions over total pipeline, key events over closed-won. The ratio is computed across two different populations, and it is stable enough to look credible — the worst property a wrong number can have.

The join, ranked by how much it can carry

You need one key, ideally three. They are not equivalent, and this is the order of trust.

Join keyWhat it attachesWhat it is blind toEffort
First-touch landing page, captured in a hidden form field and written to the contact recordThe exact URL that started the relationshipVisitors who never submit anythingOne form change, one CRM field, permanent discipline
Campaign parameters (utm_*, click IDs) stored on the recordPaid, outbound and affiliate pathsDirect, referrals, AI-assistant citations, anything stripped in transitLow, but only covers tagged traffic
Self-reported source ("how did you hear about us?")The paths no technical key can seeEverything, precisely — it is a recollectionOne optional field, plus a human who reads it

Three rules keep it honest.

  1. First touch is the primary denominator. It is written once, at contact creation, and never overwritten. A page that starts relationships does a different job from a page that assists them, and blending the two hides both.
  2. Assisted touches are a separate count. If the page appears anywhere in the sequence, record an assist, not a source. Two columns, never one.
  3. Never blend the three keys into a single "content ROI" figure. Different blind spots. Report the first-touch join, then use the self-reported field to explain the gap with reality, not to fill it.

The self-reported field is the weakest by precision and the only one that catches the traffic growing fastest: a reader who asked an AI assistant a question, got your name in the answer, and typed the brand a week later. No referrer, no parameter, no click path. A human asking "how did you find us" is the only instrument that sees it.

The four-quadrant audit

Pick a 90-day window. One axis, attention: clicks from search plus engaged sessions. The other, outcome: opportunities created where this page was the first touch. Then place every page that clears a minimum volume.

QuadrantWhat it isWhat you do with it
High attention, high outcomeThe assetProtect it. Refresh on a cadence, keep the URL, do not redesign it because the template changed.
Low attention, high outcomeThe under-read closerNext quarter's production budget goes here — into discoverability, not word count.
High attention, low outcomeA distribution surfaceKeep it, stop calling it an asset, add one route to the pipeline.
Low attention, low outcomeConsolidation candidateMerge, redirect or remove. Be honest about why it exists.

Rows two and three are where the money is, and they are exactly the two rows a traffic report cannot distinguish.

Low attention, high outcome is the most valuable and least funded page type in most portfolios. It looks unremarkable — a comparison page, a pricing explainer, a technical post with 900 lifetime clicks — and it converts the few people who reach it because it answers a decision-stage question with specifics. The instinct is to rewrite it for volume. Wrong repair. If it already touches deals, the constraint is that too few of the right people find it: internal links from pages that already rank, a title that names the decision instead of the topic, one more distribution surface. Do not rewrite the part that works.

High attention, low outcome owns the traffic slide and none of the pipeline. It is usually a list, a roundup, a broad explainer, or a comparison of things the reader will never buy from you. That is not worthless — it is a distribution surface, and distribution has value. But it has been funded as an asset, because it is the only page type that shows up in the report everybody looks at. If a quarter of giving it a route to the pipeline changes nothing, stop calling it lead generation and fund it as media, with a media budget and media expectations.

Google's other half, applied to pages you already published

Everyone applies the commodity test to plans, and plans always sound first-hand — that is what a plan is for. Apply it to published pages, with three questions.

  • Could someone with no access to your work have written this? Not "did they" — could they. If the page contains no observation, no measurement and no decision visible only from inside your operation, the answer is yes.
  • Does it name specifics or categories? "Several factors affect load time" is a category. "The 400 ms came from a font request that fired before the hero image" is a specific. Google's home-inspection example has the same shape: waiving an inspection is a decision, tips for buyers is a category.
  • Does it end in an outcome? Kraham's running store analyzed why a shoe collapsed — a result with a cause attached. A page that ends in a checklist has no result.

Run that over your top fifty pages by traffic and the pattern is usually uncomfortable: the pages that pass are rarely the ones at the top of the impressions column. In agency work that gap is consistent enough that we now treat a page's position on the traffic chart as a weak signal of anything except how broad its topic is.

What replaces the traffic slide

Three lines, once a month, per page group:

  1. Opportunities touched — first-touch and assisted, counted separately, per page.
  2. Cost per opportunity by page type — production and refresh cost divided by opportunities touched. This is the number that reallocates budget, because it is comparable across formats.
  3. Share of pipeline from pages with no paid distribution — the honest test of whether organic content does commercial work or just accumulates sessions.

Impressions, average position and total sessions are missing on purpose. They are diagnostic inputs, not outcomes, and putting them on the same slide as pipeline is what taught three years of content teams to optimize the wrong column.

None of this needs a new platform. It needs one field on a form, one field in the CRM, and the willingness to report a page as a distribution surface when that is what it is. If you want that join audited against your own stack — form mapping, CRM field, quadrant sort — it is where our SEO and GEO work starts, and you can book that review here.

Two related reads if you are fixing the reporting before the content: why your CTR benchmark no longer predicts clicks and how to tell which impressions were never people. If the audit found more gaps than you can write, the production system is the next conversation.

Frequently Asked Questions

How do I connect a published page to a deal without a developer?

Capture the first-touch landing page in a hidden form field and map it to a field on the contact record. Most form tools expose the page URL by default and most CRMs accept a text field from a native form mapping. The hard part is not the build, it is never letting that field get deleted during the next CRM cleanup.

Does a page-to-deal join still work when the click came from AI search?

It works for the clicks. A visit from an AI Mode or AI Overviews citation lands on a URL, so the hidden field captures it like any other. It cannot capture a citation nobody clicked, which is most of them, so the join under-counts AI search influence by design. Self-reported attribution is the only key that sees the rest.

What if our CRM has no landing page field today?

Start the field today and label the first 90 days as partial. Some form tools keep the submission payload, which allows a backfill; others do not. Do not model a backfill you cannot verify. An incomplete join reported honestly beats a complete-looking number assembled from guesses.

Related Articles