B2B Content Marketing Metrics by Funnel Stage

Picture two pages on one site. A plain explainer introduces a problem to people who have never heard of you. A pricing page answers the last objection of people who are about to buy. Rate both on page views and the explainer looks busy while the pricing page looks neglected, so the wrong page gets cut. The remedy is to grade every page against its own job.

The intended reader reports to a CMO or a board. The guide says what to count for each job, how to collect it with tools already in place, which figures to doubt and where a conversation with sales beats a dashboard. Analytics and a CRM are assumed, however messy.

Three jobs, one sentence each

Shortlisting content gets your name onto a list the buyer is drafting. Comparing content helps the buyer weigh options fairly. Committing content clears away the final doubts. The distinction matters because 6sense's 2025 buyer experience report put the eventual winner on the buyer's day-one shortlist in 95% of cases, and recorded buyers first contacting a seller later in the journey, about 61% of the way through in 2025 against 69% the year before. Most selection happens before any CRM record exists.

Label every page in your inventory with one of the three jobs. Shortlisting pages will look feeble in a pipeline report by design, so never force one yardstick onto all three.

Measuring shortlisting pages

The question is whether people can find your name. Useful signs include impressions, searches for your brand, newsletter sign-ups, visits typed straight into a browser and, for AI answers, how often an assistant names or cites you across a fixed list of prompts, as the AI visibility metrics guide explains. Put one question on every form: where did you first hear about us?

Impressions deserve more weight than before. Pew Research Center's 2025 browsing study recorded regular-result clicks in 8% of visits showing an AI summary and 15% of visits without one, so a page can be seen without being clicked. Since June 2026, Search Console has helped by adding generative AI performance reports with impressions broken out by page, country, device and date. Branded searches and typed-in visits trail awareness by months, so review them quarterly, and favour returning readers over raw page views.

Measuring comparing pages

Evidence here is that buyers stay and return. Look at how long people engage with comparison and how-to pages, how many click through from nurture emails, how often checklists are downloaded and whether newsletters draw replies. Shares count only when target accounts make them. Then ask the sales team which pages they send and which they never touch.

A page that reps keep forwarding outweighs its traffic. A visitor who reads several different pages within a week is probably working through a decision, which beats several one-page visits. Distrust any tool that treats an abandoned browser tab as reading.

Measuring committing pages

This is CRM territory. Count opportunities where the buyer viewed a given page before the deal closed, compare win rates for deals that engaged with content and deals that did not, and track how long cycles run. Describe the output as influence and not cause, because readers were often interested already. Print the definition of "influenced" beside each number, compare groups cautiously and use win and loss interviews to learn what the buyer actually read.

Quality checks that no tool computes

Three counts belong beside every stage: factual corrections made each quarter, the fraction of top pages refreshed within a year, and the fraction of published pieces that sales or customer success actually used. They matter because volume says nothing about usefulness. The CMI and MarketingProfs survey of 1,015 marketers found that 33% named measuring content effectiveness as a top challenge, and its AI users reported improved productivity much more often than improved content performance.

Our view: put sales usage first

Typical dashboards lead with traffic and finish with influenced pipeline. We would flip the order. Influence depends on an attribution rule and often on tiny samples, so it invites argument. A tally of how often sales sent a given page is simple, auditable and attached to a real conversation. A team closing fifteen deals a quarter cannot show a dependable win-rate gap, and should say so, yet it can still name the five pages reps keep reaching for.

The price is human logging, and people forget, so the habit has to be light: a CRM field or a shared channel. This view would fail in a motion with little sales involvement, such as self-serve software, where behavioural analytics should lead. The case for it rests on what buyers value. In Edelman and LinkedIn's 2025 survey of 1,934 executives, 53 percent said the quality of a piece's insight matters more to them than the brand behind it, which sales conversations reveal faster than clickstreams do.

The scorecard in brief

For shortlisting pages, lead with impressions, brand searches, mention rate and the stated source, drawn from Search Console, the prompt list and forms, and distrust raw page views. For comparing pages, lead with return visits, email clicks, downloads and sales usage, drawn from analytics, the email tool and the sales team, and distrust idle-tab engagement. For committing pages, lead with influenced deals, win rate and cycle length from the CRM, and distrust last-click attribution. Across all three, track accuracy, freshness and usage from an editorial log, and distrust any count of pieces published.

Where it falls short

Conversations in private chats, communities and AI assistants never reach your analytics, and self-reported sources fill the hole only partly. Samples are small, so interviews beat percentages. Vendors redefine engagement without warning, so record the tool version when comparing quarters. And any metric that becomes a target gets gamed; rotate your emphasis and keep accuracy in the set.

Ranges, not targets

Report each figure as a range over a window, with the rule that produced it. A target such as "grow branded search by a fixed amount" invites arguments about the number and encourages the gaming described above. A range invites a discussion about direction, which is what most of these metrics can support.

When a figure moves, ask three questions before explaining it to leadership. Did the definition change? Did the page change? Did the market change? Only after ruling those out should the team talk about performance. This habit costs minutes and prevents most of the embarrassing explanations in quarterly reviews.

A first month

Week one: label your top fifty pages. Week two: add the source question to demo and contact forms and ask sales to log what they send. Week three: build the table in a shared sheet with monthly columns. Week four: produce the first report and record two decisions the numbers support. A good quarterly report is then a single table, three short stories of content helping a deal, two decisions and an honest statement of what it cannot show. For a common trap, read the content ROI metric most teams get wrong, and for the wider picture see the complete guide.