What business-to-business measurement has to handle
Four properties of the market break the default measurement model. Lag: the revenue from this quarter's marketing arrives one or more sales cycles later, so a monthly report of spend against closed revenue compares unrelated things. Many touches on many people: an opportunity is the result of a committee's reading, listening and conversations over months, most of which left no click, so any single-touch rule is wrong by construction. Sales dependence: whether a qualified opportunity closes depends on the salesperson, the product and the competition as much as on marketing, so closed revenue is a noisy signal of marketing quality. And small numbers: a business that closes tens of deals a quarter cannot read statistical significance into anything, so the methods that depend on volume do not apply.
The response is a measurement plan with layers. An outcome layer: pipeline and revenue, sourced and influenced, read by cohortA group of customers who share a starting event and period, tracked together over time. over the cycle. A leading-indicator layer: the measures that move within weeks and are known to precede the outcome. A diagnostic layer: stage conversion, speed and quality by source. And an AttributionAssigning conversion credit to touchpoints. choice made deliberately, with its limits written down. The KPI trees guide covers the general method of hanging indicators beneath an outcome; this guide applies it to the business-to-business case.
Outcomes and leading indicators
| Layer | Measure | Moves within | What it tells you |
|---|---|---|---|
| Outcome | Revenue from marketing-sourced and marketing-influenced customers | One to two cycles | Whether marketing produced customers at an acceptable cost |
| Outcome | Pipeline created, by source and cohort | One cycle | The earliest revenue-shaped signal |
| Outcome | Customer acquisition cost by segment and source, over the cycle | One to two cycles | Whether the economics work |
| Leading | Qualified opportunities created | Weeks | Capture is reaching buyers |
| Leading | Engagement from target accounts and roles | Weeks | Demand creation is reaching the right organisations |
| Leading | Branded search volume and direct traffic | A quarter | The brand is being remembered |
| Leading | Self-reported attribution on demo and contact forms | Weeks | What buyers say got them here |
| Leading | Share of voice in the category | A quarter | Reach relative to competitors |
| Diagnostic | Stage conversion by source | Weeks | Where the chain is weak |
| Diagnostic | Time to first contact; service-level compliance | Days | Whether leads are being worked |
| Diagnostic | Win rate, deal size and cycle length by source | A cycle | Lead quality beyond the first stages |
The KPIA metric tied directly to an objective. at the top is revenue or pipeline, never leads, and the leading indicators are chosen because they have been seen to precede it in this business, not because a template lists them. Lifetime value (LTVTotal value a customer brings over time.) against acquisition cost (CACCost to acquire one new customer.) by segment is the outcome the economics finally rest on, and it takes the longest to read.
Lag and long cycles
The cycle length sets the clock for everything. A business with a nine-month cycle learns the revenue result of a quarter's marketing roughly a year later, and any decision made sooner is made on leading indicators and judgement. Three practices make that bearable. Report by cohort of first touch: the leads and accounts first reached in a month, followed through their stages as the months pass, so that each cohort's pipeline and revenue fill in over time and the latest cohorts are read on their early stages only. Keep the stage rates stable and reported, so that an early-stage read can be projected forward with known conversion rates to an expected pipeline, labelled as a projection. And keep spend and activity records that survive long enough to be matched with the revenue when it arrives, which means campaign naming and source fields that are consistent for years, not quarters. Turning those projections into a value per lead at each stage, from deal margin and the close rates of finished cohorts, is the subject of the guide to valuing a lead by pipeline stage.
The organisational practice is harder than the technical one: a leadership team has to agree that marketing will be judged on cohorts and leading indicators in the short term, and on revenue over the cycle, and then hold to that when a quarter looks thin. Writing the lag into the measurement plan, with the cycle length and the dates on which each cohort's result will be read, is the way to make the agreement stick. Where orders also pass through distributors, as the manufacturing industry guide describes, the plan should say which partner reports will be matched to each cohort.
Attribution options for business-to-business
The models are the same ones every marketer meets, and the attribution approach guide covers how to choose among them; the question here is which fit the market's properties. First touch at the account level answers how an organisation first found the vendor, and is the most useful single rule for judging demand creation, as long as the first touch is recorded on the account rather than the person. Last touch before opportunity creation answers what captured the intent, and is the fair rule for capture channels. The same split between creating and capturing demand runs through the comparison of LinkedIn Ads and Google Ads, which gives each platform a role and judges both on pipeline rather than cost per lead. Multi-touch across the account's contacts, with the touches of every committee member counted, is the closest to how buying actually happens and the hardest to build, because it needs the CRM connected to the marketing systems with the account as the key; the CRM connection guide covers that plumbing. Marketing mix modelling needs more volume than most business-to-business firms have, though it can work for large ones. And IncrementalityConversions caused by marketing. testing, holding out regions or account groups, is possible for reach advertising and account-based programmes and is the only method that answers whether the marketing caused anything.
The practical design for most teams is two rules reported side by side: sourced (first touch on the account) and influenced (any touch on the opportunity), with the gap between them understood rather than argued about, plus self-reported attribution as a third view. A single model that claims to settle the question will be wrong and, worse, will be believed. The guide to measuring webinars and events applies the sourced and influenced split to individual events, from registration source to the pipeline that follows. Both reports depend on the CRM holding one record per effort with its members and costs, which the guide to campaign records in the CRM sets out.
Self-reported attribution
The simplest addition to any business-to-business form is an open question: how did you hear about us? Buyers answer it with the podcast, the colleague, the conference talk, the post they saw months ago, the review site, the touches that left no click and that no model records. The answers are messy, biased towards the memorable and the recent, and incomplete; they are also the only direct evidence of demand creation that most teams will ever have. Collected as a free-text field, coded monthly into a short list of categories, and reported beside the click-based models, self-reported attribution changes budget decisions, usually towards the channels that create demand and away from the ones that only capture it.
Two rules keep it useful. Keep the question open rather than a dropdown, because a dropdown lists what the team already believes. And compare it with the tracked source on the same lead, since the difference between what the system recorded and what the buyer said is itself the measure of how much the tracked models miss.
The full method, from wording and placing the question to a stable coding scheme and the biases that shape the answers, is in the guide to self-reported attribution.
Cohorts and the quarterly view
The monthly report is for diagnostics; the quarterly report is where business-to-business marketing is actually judged, and it is built on cohorts. Each quarter's cohort of first-touched accounts and leads is a row; the columns are the stages, filling in as the quarters pass; the latest rows show only early stages, the oldest rows show revenue. Read down a column to see whether a stage rate is changing across cohorts; read along a row to see a cohort's result. Spend per cohort beside the row gives cost per opportunity and, eventually, acquisition cost, by the quarter the money was spent rather than the quarter the revenue closed.
The quarterly view also carries the leading indicators as trends, branded search, target-account engagement, self-reported mentions, with the question for each being whether the trend is consistent with the cohort rows that follow it. Over a few years the business learns which indicators lead by how long, and the measurement plan gets better at predicting. The metrics layer guide covers keeping the definitions behind these views consistent across tools, which matters more here than in most reporting, because the same cohort will be read for years.
The business-to-business dashboard
One page, read by marketing, sales and leadership, built from the CRM and the marketing systems with the definitions agreed in the alignment work. At the top, the outcome: pipeline created this quarter by source against target, and revenue from marketing-sourcedValue of sales opportunities whose original source was a marketing activity. and influenced customers over the trailing cycle. Beneath it, the cohort table. Beside it, the leading indicators as trends. Beneath those, the diagnostics by source: stage conversion, speed, win rateShare of closed sales opportunities that were won., deal size. And in a corner, the attribution views side by side with self-reported attribution, so that the disagreements are visible rather than hidden by picking one. The pipeline section sets this guide beside the stage model and the alignment work it depends on.
What the dashboard leaves off matters as much: lead counts in general, platform-reported conversions, engagement rates on content, and anything else that moves without moving revenue. Those live in the operating reports of the channels that produce them. For LinkedIn, the tutorial on building a monthly LinkedIn report builds that operating report, joining Campaign Manager leads to their CRM stages and marking recent months as incomplete.
A cohort table in its second year
A payments software vendor with a seven-month average cycle starts cohort reporting in January. By the following December, the first-quarter cohort reads: £180,000 spend, 1,100 first-touched leads, 310 marketing-qualified, 84 opportunities worth £1.9 million, 22 customers worth £560,000 in first-year revenue; cost per opportunity £2,140 and acquisition cost £8,180. The third-quarter cohort shows only £210,000 spend, 1,260 leads, 360 qualified and 70 opportunities so far, with revenue projected at £640,000 from the stage rates of earlier cohorts. Self-reported attribution across the year names the vendor's podcast for 27% of opportunities while tracked first touch credits it with 6%; branded search is up 38% year on year. Leadership reallocates £60,000 from paid search, whose cost per opportunity had risen to £3,400, into the podcast and reach advertising, and sets the date on which the decision will be judged: the following December.
The same reporting choices shape a driver tree: the KPI tree for lead generation shows how to root one in qualified pipeline and read each cohort of leads after the usual lag.
To set the spend that should fill the pipeline being measured, the lead generation guide to the budget calculator starts from the revenue target and each funnel rate.
When cycle length rises on the dashboard, the diagnosis of a business-to-business sales cycle getting longer reads median time in each stage to separate larger buying groups, early opportunities, missing proof, pricing and competition.
Where demand creation is being judged a failure on capture metrics, the page on demand generation not showing results sets out the horizon, leading indicators and self-reported evidence to agree before deciding.
Common mistakes
- Reporting spend against this quarter's revenue. The revenue came from an earlier quarter's marketing.
- Leads as the top line. The outcome is pipeline and revenue; leads are a diagnostic.
- One attribution model treated as truth. Report two rules and self-reported attribution side by side.
- Attribution at the person level. The buyer is a committee; the account is the unit.
- Dropping the free-text question. The dropdown only records what the team already believes.
- Changing campaign names and source fields each quarter. The cohort cannot be matched to its spend a year later.
- Judging demand creation on a capture metric. Branded search, target-account engagement and self-reported mentions are its leading indicators.
- Waiting for statistical significance in a business that closes tens of deals. Use cohorts, trends and holdouts where possible, and accept judgement elsewhere.
