ON THE RECORD · NO. 15 · THE SHIFT

Marketing metrics were built for a world that’s disappearing

The buyer journey got longer, research moved out of view, and the dashboard describes less of what happens every month. The cookie reversal is the least of it.

Marketing metrics were built for a world that’s disappearing
"The attribution model, fully configured," 1923
WASSILY KANDINSKY · COMPOSITION VIII · SOLOMON R. GUGGENHEIM MUSEUM, NEW YORK

After half a decade of promising to phase out third-party cookies in Chrome, Google reversed course in April 2025 and kept them, then wound down the Privacy Sandbox project built to replace them by October. The cookiepocalypse marketing teams spent years preparing for got called off.

Keeping the cookie doesn’t mean keeping the data. Consent law reached the analytics layer while everyone was watching the browser, and it now reaches the pixel directly. A visitor who declines the banner, or arrives with Global Privacy Control switched on, is invisible to the dashboard with the cookie fully alive. Email tells the same story. Privacy controls now preload messages and mark them opened whether a human ever looked, so the open rate, marketing’s oldest report line, stopped meaning anything years ago. The click still counts. The open is a courtesy. The browser spared the cookie, and the instruments went quiet anyway.

The buyer journey kept changing underneath all of it, and the parts that got harder to track were never covered by a cookie in the first place.

The average B2B buyer journey now runs 272 days across 88 touchpoints and 10 stakeholders, every one of those numbers up on the year before, according to Dreamdata’s 2026 benchmarks, an attribution vendor’s read of more than 66 million sessions across 3.5 million customer journeys. Every added touchpoint is something a dashboard has to explain and usually can’t.

This series has already covered why that much complexity breaks a funnel built for one buyer. It breaks a dashboard built to log one buyer’s clicks the same way. Ten stakeholders moving through 88 touchpoints don’t leave 88 tracked events behind them. Most of those touchpoints are a colleague forwarding a link, a Slack thread, or a conversation nobody logged anywhere.

Marketing teams have started calling that untracked majority the dark funnel. The research, the peer conversations, and the internal debate that happen before anything shows up as a session, a form fill, or a lead score. It adds up to most of the buying process, hiding in plain sight, because the tools built to measure buying were built for a shorter and simpler journey than the one buyers run now.

The journey outgrew the dashboard

Longer, more crowded journeys mean more of the decision happens before anyone touches a system that logs it. Buyers now make first contact with a seller roughly 61% of the way through the buying process. Read the other direction, most of the decision gets made before a seller, or a CRM, ever sees it.

Some of that early research happens somewhere a dashboard was never built to see. 94% of B2B buyers now use generative AI somewhere in their purchase process, and twice as many call generative AI a more meaningful source of information than any other, ahead of vendor websites, product experts, and sales. The AI surfaces themselves send almost nothing back. Google AI Mode searches end without a single click 92 to 94% of the time, a pattern covered in full earlier in this series. An impression at least got counted, however loosely. A brand surfacing inside an AI answer comes with no counter attached at all, visibility happening with the meter unplugged. A buyer can read a complete answer, form an opinion about which vendor fits, and never generate the page view marketing’s reporting was built to count.

The CFO conversation

A CFO asking for ROI proof and a CMO staring at an incomplete dashboard are both looking at a real number that used to describe most of what happened and now describes a shrinking share of it. The dashboard remains accurate about what it sees. It just sees less of the journey every quarter.

Precise attribution was always an approximation, useful for a narrower, shorter journey than the one buyers run today. The gap between the dashboard and reality opened because the journey underneath kept getting longer and less visible, one stakeholder and one AI query at a time. The browser update everyone blamed never even shipped.

What fills the gap

None of this means throwing out existing measurement. It means treating dashboard data as one signal in a portfolio of evidence, and much of the rest of the portfolio comes from asking rather than tracking.

Ask buyers directly. A single question on a discovery call or a post-close survey, “how did you first hear about us,” captures a version of the journey no pixel ever will.

Ask the market directly. The periodic brand study comes back into the kit, awareness and consideration measured in the buyer population itself, because when the research phase is invisible, the only way to know you’re in it is to ask the people doing it.

And log what the humans already see. A competitor named on a call, a screenshot shared mid-demo, a question that quotes a page nobody tracked. No new software, just treating a salesperson’s notes as a measurement input instead of a CRM field nobody reads back.

Measurement is a portfolio now, and plenty of its instruments still read true. Email clicks, where opens lie. The ad platforms’ own metrics inside their own walls. Tradeshow badge scans, booked meetings, and webinar attendance, the signals from the moments a buyer chose to stand in front of you. Branded search volume rising in the weeks after a campaign, the second-dashboard move this series has covered. The skill is weighting each instrument by how much it can be trusted, with the click-path dashboard one instrument among several rather than the whole case.

Judge patterns, not paths

The path-level record is gone, and the pattern-level evidence is still sitting in systems most teams already own.

Read engagement for quality instead of volume. Three people from the same company reading the same comparison page in one week is a buying committee forming. A session counter reports it as three sessions.

Compare the deals marketing touched against the deals it didn’t. If marketing-influenced opportunities close faster, or win more often, the CRM can prove it without a single per-touch attribution model. Influence showing up in velocity and win rate is a cleaner answer to the CFO than any dashboard export.

And reset the clock a lead source is judged on. The old model attached an expectation to every lead, an opportunity inside the quarter. On a 272-day average journey, with complex deals running well past it, the opportunity a webinar sourced may arrive in two years, not two weeks. Judge sources on cohorts and vintages, the way finance judges investments, and stop retiring channels for missing a deadline the buyer never agreed to.

The same reset applies to the lead itself. On a journey this long, one buyer may cross the MQL threshold ten times before they’re ready, going quiet and coming back with every budget cycle, reorg, and renewal date. The old model reads that as ten conversions or as one dead lead, and both readings are wrong. It’s one person getting closer on their own clock. A scoring model built to expect re-entry, instead of punishing it, stops disqualifying the pipeline’s most likely future customers for the crime of not being ready the first time.

A dashboard built to explain the last 30 days of a 272-day journey was always going to undercount most of what drove the deal. The teams measuring this well gave up the perfect funnel and built the portfolio instead, self-reported, surveyed, cohort-compared, and patient. Directionally right beats precisely wrong.

The journey got longer. The dashboard didn’t.

Katie Clark signature
SIGNED & DATED · SEP 29, 2026 · WORK NO. 15
EXHIBITED SINCE SEP 2026 · REVISED OCT 2026
EVERY POST IS A SIGNED WORK
Get the next signed work first.
One piece a month. No filler, no reheated takes.
MORE FROM THE GALLERY
VIEW THE COLLECTION →
THANKS FOR VISITING. TIME FOR FIKA. THE SWEETSHOP IS OPEN