The short answer
Measure LinkedIn Ads on two things at once. Direct is the pipeline you can trace on the record: someone clicked, landed, converted, and the tracking held up. Influenced is the pipeline where the buying company was served your ads before it arrived by some other route, which you prove by matching company-level ad exposure against the accounts that entered your CRM.
Direct is the number your finance director wants. Influenced is the number that explains why the direct one looks small in a nine-month sales cycle. Report both, every month, as a range rather than a single figure you'd have to defend line by line.
Why the same deal gets counted three times, or not at all
Plenty of people do click B2B ads. We build campaigns to drive traffic and we expect to see it. The trouble starts after the click, when four different systems try to describe the same journey and none of them can see the whole thing.
LinkedIn counts a conversion if it saw an impression within the view-through window. Google counts the last click before the form. Meta counts something else again. Add all three up and you have more conversions than you have customers.
Original source is set once and latest source keeps changing. Neither of them was designed to describe eleven touchpoints across seven months, so a deal that started with an ad in February is logged as Direct in September.
Reject the cookie banner and the click still happened, but nothing recorded it. In DACH markets the gap is wide enough that a campaign can look dead in the reporting while the sales team is fielding calls from those exact accounts.
In-app browsers strip parameters, lnkd.in redirects lose the referrer, someone copies the URL into a Teams chat, and the whole lot lands in your reports as Direct.
Nobody is lying to you. Every system is telling the truth about the small part of the journey it can see.
Direct and influenced: two columns, both real
A healthy account has a number in both columns. If influenced is empty you're either not reaching the right companies or you're not looking. If direct is empty for months on end with decent spend, something is broken and you should go and fix it rather than reach for the influence argument.
| Direct | Influenced | |
|---|---|---|
| What it is | Clicked the ad, landed, converted, tracking held | Company was served ads, then arrived some other way |
| Unit | The person and the session | The company |
| Evidence | Conversion tracking, UTMs, self-reported source | Company engagement export matched to CRM accounts |
| Survives consent rejection | No | Yes, it never relied on a cookie |
| Best used for | Optimising campaigns week to week | Deciding whether the channel earns its budget |
Neither column is the "real" one. Optimise on direct because it moves fast enough to act on. Judge the channel on both, because in a long cycle the direct column always reads low in month two and looks completely different by month eight.
Measuring the direct half properly
Before anyone talks about influence, earn the right to. Most accounts we're handed are underreporting direct results for boring technical reasons, and fixing those adds more to the picture than any clever modelling.
- Conversion tracking that fires and gets checked. Spent £500 and recorded nothing? Test the tag before you question the offer. It's the tag more often than it isn't.
- UTMs on every single ad, captured on first touch and kept. Not last touch. If someone lands from an ad in March and books in June, you want March's source still sitting in a hidden field on that form.
- A "how did you hear about us" field on the booking form. Unfashionable, and the only signal that survives everything. People tell you things no pixel will.
- Server-side conversions where your CRM supports it. Sending stage changes back from the CRM recovers a chunk of the conversions that ad blockers and consent refusal binned.
Not sure whether your reporting is wrong or your ads are?
Book a 30 minute call. Tell us how it is set up now and we will give you an honest read.
Measuring influence when the click never gets recorded
This is the part almost nobody does, and it's the part that changes the conversation with a finance director. LinkedIn will tell you which companies your ads reached. Your CRM will tell you which companies turned up. Put the two lists side by side once a month and you have evidence rather than a theory.
- Export the company engagement report from Campaign Manager for the period. You get the company, paid impressions, engagement level, and whether the engagement was paid or organic.
- Pull every new account, sign-up or deal your CRM created in the same period.
- Match on company name and domain. Clean off Ltd, GmbH, Inc and the punctuation, and match the obvious near-misses by hand. On a normal month that's twenty minutes of work.
- Count how many of the new accounts had already been served ads, and how many of those are genuinely good-fit companies rather than accidental reach.
- Record what your CRM logged each of them as. That column is the whole argument.
Impressions are the only signal a cookie banner cannot delete. That is why we count them.
Company-level exposure is the honest unit here. You aren't claiming that one specific engineer saw one specific ad and therefore the deal is yours. You're saying that the account was reached, repeatedly, before it walked through the door, and that this happened far more often than chance would allow across a target list of a few hundred companies.
Judging retargeting, which is most of the work
A large share of the campaigns we run exist to move people already in the funnel over the line. Someone read three pages in April, went quiet, and needs to see you four more times before the internal business case gets written. That work is genuinely valuable and it is also the easiest thing in the world to measure badly.
Retargeting catches people at the end, so it soaks up credit for demand that cold campaigns, SEO and the sales team created months earlier. A 20x ROAS on retargeting usually means the rest of the mix is doing the heavy lifting.
Score everything on first touch and retargeting looks like it contributes nothing, so somebody switches off the campaigns that were closing deals. Both models are wrong in opposite directions.
What we judge retargeting on instead: velocity and conversion rate of the exposed group against the unexposed one. Do accounts that got retargeted reach a call faster? Do more of them get there at all? Hold a small unexposed group back for a quarter if the volume allows it, and the difference between the two groups is the closest thing to a clean read you'll get without a lab.
What this looked like on a real account
A European construction-tech client, selling software into main contractors and specialist subcontractors across Norway, the UK, Germany and the US. Long cycles, technical buyers, a CRM full of accounts with "Offline Sources" in the original source field.
We ran the match on one month of new sign-ups. Roughly one in five had already been served paid impressions, several of them dozens of times, and every good-fit company among them was logged in the CRM as Direct, Organic Search or Offline. Not one said LinkedIn. The campaigns were producing direct conversions that same month on top of all that. The influence was always there. The reporting just had nowhere to put it.
The gap between what the CRM said and what had actually happened was large, it repeated every month, and it took under half an hour to produce.
The six-figure deals that had stopped moving
The same account gave us the better story a few months later, once we had proper visibility of the sales pipeline rather than just the form fills.
Multiple deals worth six figures each had been sat still for months. Not lost, not cold, just quiet in the way big deals go quiet when a buying committee is arguing with itself and nobody wants to be the one to email you. We rebuilt the targeting around the named accounts the sales team actually cared about, put real creative in front of them instead of the usual feature grid, and let it run on under £3,000 a month.
Inside about four weeks the pipeline started shifting. Fresh leads coming in from target accounts, new MQLs, new SQLs qualifying, first meetings getting booked. And at the top end, two of the biggest stalled deals in the business went to proposal. One climbed three internal stages from qualified to proposal sent. The other went from a first meeting to a proposal inside the month. Across that window both sat among the most heavily served companies in the entire pipeline, hundreds of impressions each against the exact job titles in the room.
Under £3,000 a month in media kept the ads in front of six-figure deals decided by people who never once filled in a form.
The wider read changed how the client reports internally. Hundreds of companies from their live pipeline had been served ads over a two month window, across every stage from early conversation to proposal, plus a long tail of dormant accounts being kept warm. Their CRM had credited almost none of it. A last-click field on a deal record has no way of describing an account that saw you eleven times before someone finally replied to a sales email.
Be careful with the claim. Exposure sitting alongside movement is correlation, and some of those deals would have moved anyway. What it does prove, and what marketing had never been able to prove before, is presence. You reached the right companies, at the right stage, while the decision was live. Put that table in front of a finance director and the conversation stops being about whether LinkedIn works.
Make it a monthly ritual, not an annual argument
Attribution turns into a row when it only gets discussed at renewal. Do it every month on the same day and it becomes boring, which is what you want from reporting.
- Export company engagement and new CRM accounts for the month.
- Match, and log the direct conversions and the influenced accounts separately.
- Add the self-reported answers from the form as a third, independent check.
- Report a range. "Between the tracked conversions and the matched accounts, LinkedIn is involved in somewhere between six and fourteen of this month's opportunities" is a sentence you can defend. A single decimal-pointed number is not.
- Put budget where influence is provable, and treat markets with heavy consent loss as directional rather than dead.
A range you can defend beats a precise number you invented.
Where this method falls over
Be honest about the edges, because someone in the room will find them anyway.
- Low impression volume. Below a few thousand impressions a month the company engagement export is too thin to say anything.
- Very small target lists. If you're reaching thirty companies and eight of them buy from you every year anyway, overlap proves less than you'd like.
- Sole traders and micro businesses. Company-level matching needs companies. It falls apart for one-person buyers.
- Exposure is not causation. Some of those accounts were coming regardless. Say so, and let the trend over several months carry the argument instead of one month's table.
Read next
Common questions
If you want your ads measured properly rather than argued about
- Conversion tracking and first-touch UTM capture rebuilt so the direct column stops underreporting.
- Monthly company-level influence match between ad exposure and new CRM accounts.
- Retargeting structured around genuine intent signals, with exposed against unexposed reporting.
- A reporting pack that shows direct, influenced and self-reported side by side, as a range.
- Budget recommendations based on where influence is provable, not where the dashboard is prettiest.
Starter £975/mo, Growth £1,725/mo, Scale £4,025+/mo. Most accounts start in Growth. Multi-market setups with heavy consent loss usually need Scale because the measurement work multiplies per region.