Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

B2B Marketing › Advanced

Designing Pipeline Measurement for an Industrial Pump Maker

The board of an invented industrial pump maker asks which campaigns create pipeline, and the only report it has credits the last touch. You list the decisions marketing must support, match each to a measurement method, design an account-level holdout for the channel the board doubts most while being honest about what so few opportunities can show, and specify how qualified pipeline flows back to Google Ads.
ExerciseAdvanced levelModule 2: Measuring influence over long sales cycles5 minUpdated

Why this matters

A board question about which campaigns create pipeline hides several decisions with different stakes. The guide to choosing an attribution approach starts from the decision and sets the standard of evidence by its cost and reversibility. The enhanced conversions guide shows how outcomes recorded weeks later in a CRM can reach bidding without a stored click identifier. Long cycles and few deals make both harder.

The maker and its board's question

Pellow Pumps is an invented manufacturer of industrial pumps for water treatment and process plants. Engineers specify, procurement buys, and a deal takes six to twelve months. Its CRM credits each opportunity to the last touch before it was created, and the board reads that report as the truth. Board members doubt LinkedIn most, because the report credits it with almost nothing.

Illustrative example — not a real result

Pellow Pumps: last year's channels and outcomes

ChannelSpendOpportunities credited by last touchShare of "how did you hear about us" answers
Google search£120,0004614%
LinkedIn£90,000322%
Trade shows£80,0001431%
Technical content and search£40,000818%
Distributor referrals£30,000915%
  • 80 opportunities, 24 won: a 30% win rate. Average order £60,000 at a 40% gross margin.
  • 480 enquiries accepted by sales, about one in six becoming an opportunity. Nearly all come from a list of 400 target plants and contractors, which LinkedIn reaches through a company list.
  • Google Ads optimises to 1,900 form submissions a year, including job seekers and students.
  • The CRM stores the form-filler's email on each lead; opportunities sit on the account and often name a procurement contact instead.

Matching methods to B2B decisions

List the decisions first, with who makes them and how often. Weekly bid and budget choices inside one platform can run on that platform's own Attribution, if it learns from the right outcome. Questions about each channel's role suit model comparison and buyers' own answers, read as hypotheses. A decision to keep or cut a large, contested line needs a test of Incrementality, as the incrementality testing guide explains, and the board report needs the CRM as its system of record, read by cohort as the business-to-business measurement guide describes.

Task

  1. List the decisions marketing must support, with owner and cadence, and choose a method for each.
  2. Design a holdout for LinkedIn: unit, groups, outcome, duration, break-even and decision rule.
  3. Estimate roughly what the holdout can detect, and say what a null result would mean.
  4. Specify enhanced conversions for leads: capture, stages, values, matching and checks.
  5. Write the one paragraph the board will read about what the measurement can and cannot tell it.

Worked answer

DecisionOwner and cadenceMethod
Bids, budgets and queries in Google AdsAgency, weeklyPlatform attribution on imported pipeline stages
What each channel does in the journeyMarketing, quarterlyAnalytics data-driven model against last click, with self-reported answers
Whether LinkedIn keeps its budgetBoard, annuallyAccount-level holdout
Trade show and distributor spendBoard, annuallySourced and influenced pipeline by cohort, with self-reported answers
The return the board seesBoard, quarterlyCRM pipeline and revenue by first-touch cohort

Trade shows and distributors leave no clicks and happen too rarely to test, and a mix model needs more history and volume than Pellow has, so they are judged on cohort pipeline and what buyers say.

Illustrative example — not a real result

Pellow Pumps: holdout arithmetic

Split the 400 target accounts into two groups of 200, matched on sector, size and last year's enquiries. Withhold LinkedIn from one group for twelve months; the other keeps its usual spend, about £45,000 for the year.

Baseline per group for the year: about 240 accepted enquiries and 40 opportunities. Gross margin per opportunity: 0.30 × £60,000 × 0.40 = £7,200; per accepted enquiry about £1,200. Break-even: £45,000 ÷ £1,200 ≈ 38 extra enquiries, about 16% above baseline.

Rough smallest detectable difference, treating counts as independent: enquiries 2.8 × √(240 + 240) ≈ 61, about 25%; opportunities 2.8 × √(40 + 40) ≈ 25, about 63%. Differences between accounts make both thresholds higher.

What the holdout can show. Enquiries are the primary outcome, because opportunities are too few to read in a year; opportunities are reported as a secondary outcome and followed for another cycle. The test can confirm LinkedIn if it lifts enquiries by about a quarter or more. It cannot tell a break-even effect from no effect, so a null result means "below about a quarter", not "zero". Write that into the decision rule before launch: a clear lift keeps LinkedIn and tests a higher level; a null result moves LinkedIn to an account-based programme on the top tier only, at lower spend, rather than cutting it on evidence the test could never provide.

Enhanced conversions for leads. Tag Manager pushes the enquiry form's email to the data layer, where the conversion tag hashes it, gated by consent. Three import actions follow the stages: accepted enquiry as primary, valued at its gross margin figure above, because it arrives within days and in volume; qualified opportunity and won order, valued the same way, stay secondary until opportunity volume allows bidding on them. The bid target becomes a CPA per accepted enquiry, with job seekers and students removed at acceptance. When an opportunity names a procurement contact, the CRM links it back to the originating lead, so each upload uses the form-filler's email once. Uploads run weekly; a monthly reconciliation compares imports with the CRM and tracks the unmatched share, which will never be zero because not every engineer was signed in when they clicked.

For the board. Last touch shows which channel captured an enquiry, not which created it. The plan reports pipeline by cohort beside buyers' own answers, tests LinkedIn directly and teaches Google Ads to buy enquiries sales accept.

Common mistakes

  • One model for every decision. The cost of being wrong sets the standard of evidence.
  • Testing on opportunities alone. Too few to read; use a frequent leading outcome.
  • Reading a null result as no effect. Say in advance what the test could not detect.
  • Importing every stage as primary. The same deal counted several times inflates bidding.

Next, take the long-cycle measurement checkpoint or revisit pipeline tracing at the intermediate business-to-business level.

Pages this lesson applies

Terms used on this page

Also related

Related