Metrics
Most B2B demand is won before a buyer fills a form, then credited to the last searchable touch. A brand demand ledger combines share of search, direct and homepage behavior, self-report, selection signals, and pipeline so operators can see preference forming before CRM attribution catches up.
Your Monday report says paid search created the quarter. Sales says the strongest opportunities arrived already convinced. The content team points to a podcast nobody clicked. Finance points to the CRM and asks why the podcast has zero pipeline.
All four can be reading their instruments correctly and still misread the room. B2B demand often forms before an identifiable visit. A buyer hears your name in a peer group, sees an operator quote you on LinkedIn, reads three untracked comparisons, then types your brand into Google and requests a demo. Search receives the receipt. It did not sew the garment.
The answer is not a more ornate attribution model. It is a brand demand ledger: a compact set of leading, middle, and commercial indicators that shows whether more of the right market knows you, seeks you, prefers you, and becomes qualified pipeline. No single row gets to claim causality. Together, the rows reveal whether preference is being cut before revenue appears.
What follows is the operating system: the evidence behind it, five measurements, a worked B2B example, failure modes, a Monday cadence, and the checklist to install it.
The form fill is not where demand begins. It is where private preference finally leaves a measurement trace.
— THE SCALE MANIFESTO, 1924 (REV. 2024)
The strongest recent evidence comes from 6sense's 2024 Buyer Experience Report, based on 2,509 recent B2B buyers. The study found that 69 percent of the purchase process happened before buyers engaged sellers. Eighty-one percent had selected a preferred vendor before speaking with sales. The first seventy percent of the journey was not casual awareness. It was the selection phase, where the buying group set requirements, assembled a shortlist, and chose a favorite.
That changes the operator's question. You are not merely asking which channel caused a conversion. You are asking whether your company entered the buying group's memory and shortlist while its members were still anonymous. By the time a lead becomes legible in the CRM, much of the competitive work is finished.
The referral trail is no rescue. SparkToro's 2024 examination of dying clicks and attribution explains why influence on social feeds, private communities, video, audio, email, and messaging often resolves into a later search or a direct visit. Those channels create attention but do not reliably pass referral strings. The channel that receives the visit looks productive. The channel that created the preference looks decorative.
This is the same measurement seam explored in our note on growth when nobody clicks. The practical response is not to award invented credit. Keep causal measurement for decisions where treatment can be withheld. Use a disciplined demand ledger to read the spaces where withholding is impossible or commercially foolish.
A good ledger crosses the full cloth. It includes market attention, owned behavior, declared influence, preference, and commercial consequence. If every metric comes from the same platform, you have a dashboard, not triangulation.
In a 2025 study of roughly 150 million United States keywords, Ahrefs found that 36.9 percent of queries were branded and branded queries represented 45.7 percent of search volume. The useful signal for a B2B operator is not raw searches for your name. It is your branded search volume divided by the branded search volume of you plus a fixed competitive set.
Build the query set once: company name, common misspellings, flagship product, founder when relevant, and brand plus category or comparison terms. Remove customer login and support queries if they swamp prospect demand. Report a three-month rolling share and its direction. Search volume is modeled and navigational behavior is real, so treat the level as approximate. The trend and competitive movement carry the signal.
Track direct sessions and organic entries to the homepage, pricing, product, security, integration, and customer story pages. Segment new users where identity allows, and track engaged account domains for larger contracts. Seer's 2024 argument for replacing old SEO KPIs is useful here: brand-leaning direct and homepage traffic can move with leads and revenue even while generic organic traffic falls. Do not call every direct session brand. Bots, bookmarks, employees, and missing referrers live there too. Read the qualified page mix and movement together.
Ask two separate questions. At conversion: How did you first hear about us? After qualification: What influenced your shortlist? The first captures recall of discovery. The second captures the buying group's evidence. Keep the first open text, then code responses monthly into stable categories. Let sales capture the second in natural conversation, with exact language preserved.
This is research, not attribution. Reforge's 2024 decision-first research guidance starts with what decision you need to make, who has the right experience to answer, and what evidence would change your confidence. Apply that discipline. Memory is incomplete, recent touches crowd out earlier ones, and respondents cannot name sources they never encountered. Use the answers to find language and patterns, never to allocate exact revenue credit.
Add three fields to opportunity discovery: Were we on the initial shortlist? Were we the preferred vendor at first substantive meeting? Had the account used, evaluated, or bought from us before? The point is not to make sales perform anthropology. It is to distinguish demand arriving ready-made from demand the selling process had to manufacture.
Report preferred-at-first-meeting rate by segment and source cohort. A rising rate with stable win rate often means brand preference is strengthening. A rising demo count with a falling preference rate usually means you are buying more curiosity, not more demand.
The ledger must eventually touch money. Use qualified pipeline created, win rate, sales cycle, average contract value, and closed revenue by quarter. Keep these lagging measures beside the leading ones, not merged into a synthetic score. An index can hide the exact seam that needs repair.
When the ledger suggests a channel changed demand, test that claim where you can. Our field guide to incrementality testing with holdouts shows how to move from directional evidence to a causal budget decision. The ledger tells you where to cut a muslin. It is not the muslin.
Consider Loomline, a fictional B2B workflow platform at eighteen million dollars in annual recurring revenue. It sells contracts averaging forty-eight thousand dollars through a ninety-day cycle. The team invests sixty thousand dollars a month in an operator newsletter, customer workshops, founder-led posts, and two specialist podcasts. Last-click reporting assigns the program eleven opportunities and two hundred ten thousand dollars in pipeline after six months. Finance calls the cost indefensible.
The operator builds a ledger with a six-month baseline and fixes five competitors before looking at the result. At baseline, Loomline owns 8.4 percent of competitive branded search. Direct and homepage visits to commercial pages average 1,900 a month. Twenty-two percent of qualified opportunities name Loomline as their preferred vendor at first meeting. Quarterly qualified pipeline is 3.2 million dollars.
After two quarters, paid media spend and category demand are roughly flat. The ledger reads:
None of this proves the program caused nine hundred thousand dollars in added pipeline. A product release, sales hiring, competitor weakness, or economic change may share the credit. But the pattern is coherent across independent instruments: more category search belongs to Loomline, more people arrive through brand-leaning routes, more qualified buyers name the program's surfaces, more accounts prefer Loomline before sales, and more pipeline follows.
The decision is not to declare victory. Renew the program for one quarter, hold audience and creative themes steady, and run a matched-market workshop test where geography permits. Keep one podcast, pause the weaker one, and watch whether branded demand or preferred-vendor rate bends. The ledger has earned the next measurement cut.
A demand metric earns trust when it can change a decision, survive a skeptical question, and admit what it cannot prove.
— THE SCALE MANIFESTO, 1924 (REV. 2024)
A weighted brand index looks board-ready and destroys diagnosis. If search rises while preference falls, the business has a different problem than when preference rises and pipeline stalls. Keep the rows separate. Add a short written reading of the pattern.
Removing a fast-growing competitor flatters your share of search. Adding one mid-quarter ruins comparability. Fix the set for twelve months, document mergers and renames, and show both the continuing set and the market event when reality forces a change.
Direct is a bucket for known visits and lost information. Filter staff, customers, login routes, bots, and obvious repeat behavior. Pair it with commercial page depth, new visitor share, and account quality. An unexplained spike is an investigation, not a trophy.
A buyer who writes podcast is not assigning the podcast one hundred percent of the deal. They are recalling one thread from a group decision. Preserve the wording, code consistently, and compare patterns by segment. Never force the answer into a revenue pie chart.
B2B volumes are lumpy. A conference, renewal cycle, press mention, or five large accounts can move a week. Use rolling three-month trends for leading indicators and quarterly cohorts for commercial outcomes. Annotate launches and market shocks rather than explaining them from memory.
The ledger becomes useful when it changes work. Give it thirty minutes on the first Monday of each month, with marketing, sales operations, and finance in the room.
Quarterly, bring the longer cut to the planning table. Compare segments, not only totals. Enterprise demand may strengthen while small-business demand collapses. Report ranges and caveats beside every modeled metric. Archive each quarter's definitions so the company cannot quietly rewrite history.
A good tailor reads more than the final button. The grain, shoulder, drape, and movement all reveal whether the piece will hold. Measure B2B demand the same way. The CRM is the final fitting, not the first cut. Build the ledger upstream, and you can see preference taking shape while there is still time to alter the pattern.