MEASUREMENT
How we measure.
Every agency reports the numbers that flatter it. This is the whole method instead: what we track, what we optimize to, what click attribution cannot see, and what we will never claim. Read it before you talk to us, not after.
What we track
Four things, set up in the first two weeks and never changed quietly afterwards.
The LinkedIn® Insight Tag
A free tag from LinkedIn on every page of your site. It does three jobs: conversion tracking, website retargeting audiences, and website demographics — the job titles, companies and industries of the people who visit. LinkedIn describes it as browser-based and cookie-based, which is also the reason it under-counts. That is a property of the tag, not an opinion about it.
Conversion definitions, written down
Every conversion has a written definition before it is created: what fires it, on which page or event, what value it carries, and which CRM stage it is meant to predict. You get that list. If a definition changes, it is in that week’s change log with the date it changed, because a conversion redefined quietly makes every earlier month unreadable.
One UTM convention
utm_source, utm_medium, utm_campaign and utm_content built from the same naming convention as the ad set and the ad. Lower case, hyphens, no spaces, no one-off exceptions. The convention is written into your account document so anyone on your side can read a URL and know exactly which ad produced it.
CRM stage mapping
Your stage names, mapped once onto four reporting stages: lead, sales-accepted, opportunity, closed won. We map to your stages; we do not ask you to rename them to suit our report. This mapping is what makes everything below it possible.
WHERE THE CONVERSION DATA COMES FROM
Insight Tag
Browser-based and cookie-based.
Conversions API
Server to server. Does not rely on cookies. LinkedIn rates its reliability high.
CSV upload from your CRM
Manual. LinkedIn rates its reliability medium.
Running the Insight Tag and the Conversions API together does not double-count. LinkedIn identifies an event sent through both and counts it once.
Source: LinkedIn Help. Last checked 23 September 2026.
What we optimize to
Sales-accepted leads and opportunities. Not MQLs, and not raw form fills.
A LinkedIn conversion is a form fill. A form fill tells you somebody read something. It does not tell you a buying committee is moving. Optimize toward form fills and LinkedIn’s delivery will do exactly what you asked: find the cheapest people who fill in forms. On LinkedIn those are rarely the people who buy.
So we send the answer back. Your CRM already knows which of last month’s leads your sales team accepted, which became opportunities and which closed. Those stages are imported into LinkedIn as offline conversions, and the campaigns are then optimized toward the stage that has money attached to it.
There are three ways to get that data back into LinkedIn: a file upload from your CRM, an automated partner transfer, or the Conversions API. A manual upload is not instant and appears in reporting after a processing delay; a partner-automated transfer is close to real time. We tell you which route your stack actually supports before we promise you a reporting cadence, not after.
Source: LinkedIn Help, offline conversion import. Last checked 23 September 2026.
Why we do not report MQLs
An MQL is a volume number defined by marketing, scored by marketing and reported by marketing. On a long B2B cycle it can climb all quarter while pipeline does not move. We still count them, as a leading indicator that delivery is working, and we will show them to you. They are not the result, and no campaign of yours will be optimized toward them.
The lag, stated up front
Pipeline is the right unit and it is also the slow one. On a long B2B cycle the signal from a campaign lands months after the spend that caused it, and on some cycles later than the initial term (90 days). We do not publish a figure for that lag, because it is a property of your sales cycle and not of our service: we measure it on your own closed-won records and tell you what it turns out to be. So the reporting promise is staged: leading indicators in weeks, sales-accepted leads and opportunities inside the first quarter, pipeline and CAC payback after that. Anyone who promises you pipeline reporting in month one is promising a number that does not exist yet.
What we cannot track, and we say so
Click attribution sees the last click. It does not see the message in someone’s Slack, the post a colleague forwarded, the peer who was asked who they use, or the ad that was seen and not clicked. That share is not small, and the honest position is that nobody has measured it cleanly.
THREE QUESTIONS ABOUT THE NUMBERS ON THIS PAGE
How much B2B pipeline is invisible to click attribution?
We do not publish a figure, because we could not find one that survives being checked. The ranges that circulate in B2B marketing are quoted from each other rather than from a dataset, and the ones we tried to trace did not lead back to primary research. Putting one on this page would be the same trick this page is asking you to watch for. What we will put here is measured, named and dated, which is the next answer.
Is it true that most of the buying journey happens before you hear from a buyer?
Less than the usual pitch claims. 6sense’s 2025 Buyer Experience Report put first vendor contact at 61 per cent of the way through the journey, down from 69 per cent the year before — so buyers are contacting vendors earlier, not later. We quote the number that moved against our own sales interest because it is the better-sourced one.
Then why advertise on LinkedIn at all, if you cannot prove the effect?
The same 6sense report found that 94 per cent of buying groups had already ranked a preferred vendor before their first vendor contact, and bought from that favourite 77 per cent of the time. The argument for LinkedIn is not cheap leads. It is being on the shortlist before anyone fills in a form. The part of that effect you can prove will always be smaller than the effect.
These are third-party figures about the B2B market, quoted with their source and the date we last checked them: 23 September 2026. They are not our results, they are not a forecast, and nobody should read them as a promise about your account.
Four things we will not pretend to know
Which person saw which ad.
LinkedIn reports at ad-set level, not person level, and we will not reverse-engineer a name out of a report.
How much of your direct and organic traffic we caused.
Direct traffic share is a proxy at best. We will never put it in a report as attribution.
Whether an expanded-audience impression influenced a deal.
LinkedIn turns audience expansion on automatically where it is available, and its activity is reported together with your target audience. We turn it off unless you ask for it, and the report says which ad sets have it on.
View-through effect, with any precision.
LinkedIn lets the conversion window be set and lets you credit every ad set or only the most recent one. We choose once, publish the setting in your account document, and do not change it mid-quarter.
How we handle it: we ask
One question, on every form, every time. How did you hear about us? It is required, it is free text, and the answer is read next to the platform data, never instead of it.
It is required.
An optional field is skipped by exactly the people whose answer matters most. Ours is required on every form we build, including the one on this website.
It is free text, not a dropdown.
A dropdown teaches people the answer you want. Free text gets you “a colleague sent me your teardown”, which is the sentence you actually needed.
It is written onto the CRM record, not counted on its own.
The answer travels to the opportunity, so at the end of a quarter you can read what your closed-won deals said about where they came from.
We publish its faults.
People misremember the first touch, they name the most recent one, and they say “your case study” when they mean “your ad”. Expect a large share of answers to be too short or too vague to use. We count the usable ones rather than estimating them, and the monthly report says how many there were. Self-reported answers and platform attribution measure different sets, and those sets no longer overlap. We run both and reconcile them. We do not pick whichever looks better that month.
What a monthly report actually contains
Six parts, the same six every month, in the same order. Nothing appears one month and quietly disappears the next because it looked bad.
EXAMPLE
EXAMPLE
$20,000
Spend
EXAMPLE
40
Sales-accepted leads
EXAMPLE
10
Opportunities created
EXAMPLE
$2,000
Cost per opportunity
Every number in that panel is made up. It shows the shape of the report, not a result.
01
The four numbers.
Spend, sales-accepted leads, opportunities created, cost per opportunity. Always these four, always at the top.
02
Platform view and CRM view, side by side.
What LinkedIn reported for the period, and what your CRM recorded for the same period, in two columns with the difference shown. We explain the gap. We do not remove it.
03
What we changed, and why.
The dated change log for the month: every audience, budget, bid and creative change, with the reason it was made and what it did.
04
What the self-reported answers said.
The free-text answers from the month, quoted, with the number of submissions that gave a usable answer alongside them.
05
What we do not know.
The measurement gaps that were open this month, written out rather than left for you to notice.
06
Next month’s test plan.
One or two named tests, and what would have to be true for each to count as a win, written before the test runs.
How often we walk through it with you on a call depends on your plan, and that is published on the pricing page.
What we will never do
Claim credit for pipeline we cannot evidence.
If the CRM cannot show it, we will not put it in a slide.
Report MQLs as the result.
They are a leading indicator. They go in the report, not on the front page of it.
Show platform-attributed conversions on their own.
The CRM view sits next to them, every time, including the months the CRM view is worse.
Move the goalposts mid-quarter.
Conversion windows, attribution setting and date ranges are set at the start and kept. Changing them to improve a chart is the oldest trick in this industry.
Publish a benchmark as a promise.
A third-party figure is quoted with its source and the date it was checked. We will never write “our clients see” in front of somebody else’s number.
Send a number without its definition.
Every metric in your report is defined in your account document, in words, before it ever appears in a chart.
See what your account is actually measuring.
The teardown starts with your tracking: what is firing, what is defined, what is double-counted and what is missing entirely. You get a recorded walkthrough and a one-page plan in 3 business days.
No call required. We reply to every message within 1 business day.