Attribution Breaks in Four Layers, and Each Needs Different Help
"Our attribution is a mess" usually means one of four things. They sit on top of each other, and a problem in a lower layer cannot be fixed from a higher one.
- Tracking. Whether the conversion events fire, and whether each one means something. A "conversion" that counts a page view is not a lead, and an ad account optimising toward a broken conversion spends confidently in the wrong direction.
- The CRM. Whether every lead lands in one record with its source attached, and whether that source survives the handoff to sales. If the CRM cannot say where a closed deal came from, nothing above it can either.
- The report. Whether the numbers are labelled honestly: which platform reported them, over which attribution window, and whether anyone has summed figures from different platforms and called the total revenue.
- Judgement. Whether the business has agreed how to judge the spend that no tool can trace, which in B2B is often the spend that matters most.
Each symptom points to a layer, and the layer tells you who to call.
| What you are seeing | Layer | Who can fix it |
|---|---|---|
| Google, Meta and LinkedIn each claim the same deals; their totals add up to more than you sold | Report | Whoever builds the report, with a rule that platform numbers are labelled and never summed |
| Platform conversions are up, sales says the leads are worse | Tracking | Whoever runs the ad accounts: redefine the conversion, import closed deals back |
| Nobody can say which campaign a closed deal came from | CRM | A CRM build or clean-up, before anything else |
| The brand and content budget gets cut every tight quarter | Judgement | Leadership, with a separate scorecard for that spend |
| Every dashboard tells a different story | Report | One report, one source of truth, the window stated on every view |
Notice what is missing from the right-hand column: "a better attribution model". A model sits on top of the first three layers. Give it untested events and sourceless leads and it will distribute credit very precisely across data that is wrong.
Where Can I Get Help With Marketing Attribution? Five Places, Five Different Jobs
Once you know the layer, the options sort themselves. Each of these is the right call for something and the wrong call for something else.
| Kind of help | Good for | Not good for |
|---|---|---|
| Your own analyst or RevOps lead | Owning the definitions and the report month after month, knowing your sales process | Fixing tracking inside ad platforms they do not run |
| A CRM build or admin partner | Getting every lead into one record with its source, stages that reflect how you actually sell | Deciding how to judge spend the CRM cannot see |
| The team running your paid media | Conversion tracking, offline conversion import, labelling platform numbers honestly | Grading its own homework without a report you can check |
| Attribution software | Modelling credit across touches once the data underneath is sound | Repairing that data. It inherits every gap below it |
| An independent consultant | An outside audit of all four layers, a second opinion before a big purchase | Running it every month afterwards |
The row people skip is the third. If an agency runs your Google, Meta or LinkedIn accounts, most of the tracking layer is already in their hands, and it is fair to ask them to fix it before you pay anyone else. The condition is in the last column: they should hand you a report you can verify, not a number you have to trust.
Where Can I Get Help With Marketing Attribution If We Have No Real CRM?
Then that is where the help goes first. Without a CRM, nothing can be measured past the form fill, so there is no way to tell whether any of the spend worked. Every other option on the list assumes the CRM exists.
The minimum is less than people expect. Every lead from every source routes into one record. The campaign source is attached when the lead arrives, so it survives the handoff to sales. The ad platforms' tags and analytics are wired into it. Then closed-won deals are pushed back out to LinkedIn and Google as offline conversions, which is what lets those platforms optimise toward revenue rather than form fills. That last step is what separates a serious B2B setup from a basic one, and it is the part of our CRM builds that attribution depends on most.
Buy Attribution Software Last, Not First
Software is the most visible answer because it is the one with a marketing budget behind it. It is also the one that can do least on its own, for three reasons.
- It inherits the layers below it. A tool reading conversion events that count page views reports page views, with more decimal places.
- Your own data is under-used. Website behaviour, CRM history, email and sales-engagement data are sitting unread at almost every company, and they cost nothing extra. We exhaust those before recommending anyone buy a platform.
- No model sees the first touch that mattered. Multi-touch models mostly reallocate credit among channels that were all late. The work that made a buyer aware of the problem months earlier is not in any of their data, which is the subject of the last section.
None of that makes tools bad. It makes them the last step. We are not resellers and we take no vendor commissions, so when a tool genuinely is the constraint we will name it and say what it costs. More often the recommendation is less software, configured properly.
One cheap source deserves a mention because tools ignore it: ask. A required "how did you hear about us?" field on the form, or the same question on the first sales call, captures what arriving customers say in their own words. It is softer than a tracked click. It is also the only data source that can name a podcast, a conference conversation or a colleague's recommendation.
Where Can I Get Help With Marketing Attribution I Can Trust? Six Questions to Ask
Whoever you are talking to, an agency, a consultant or a software vendor, these separate help from a sales pitch.
- "What counts as a conversion, and who tested it?" Tracking should be installed, tested and verified firing before a dollar is spent. If they cannot say how it was tested, it probably was not.
- "Does the source survive into the CRM?" Ask to see one closed deal and the campaign it came from. One real example is worth a demo.
- "Do you import closed deals back into the ad platforms?" Otherwise the platforms keep optimising toward form fills while the revenue data that should steer them sits unused.
- "Which attribution window is this?" Every number should carry its window and the platform that reported it. A dashboard without one cannot survive a board question.
- "Will you add the platforms together?" The right answer is no. Each platform counts the conversions it can claim, and the same deal gets claimed more than once.
- "How will you measure what cannot be attributed?" Anyone promising clean attribution for brand and demand creation is selling the other half of your budget with a different label on the invoice.

What a Good Attribution Report Looks Like
Whoever helps you, this is the thing they should leave behind. Here is what we report on the accounts we run, and what we refuse to present as a result.
| What we report | What we refuse to report as a result |
|---|---|
| Pipeline created and influenced, in dollars | Platform numbers summed across channels and called revenue |
| Cost per opportunity, not cost per lead | Clicks and impressions as an outcome |
| Cost per qualified lead: leads sales accepted, not leads the form produced | Conversions with no definition behind them |
| Win rate by source, and sales acceptance rate | Platform ROAS presented as profit |
| The demand creation and capture split, judged separately | Early-stage data presented as a trend |
| The platform named and the attribution window stated on every view | Any dashboard without its attribution window |
The ROAS line catches people out. Platform ROAS is the platform's own reported conversion value divided by spend. It is not audited revenue, it is not de-duplicated against your other channels, and it does not know your margins. It is directionally useful, so we report it, labelled as the platform's figure. LinkedIn reports what LinkedIn attributes, and the report should say so in those words.
There is also a discipline that makes attribution possible inside a single platform: change one thing at a time. We space structural changes three to five days apart, because a week with five changes in it produces a result nobody can attribute to any of them.
The Half of Your Marketing No Model Can See
In a long B2B cycle, the enquiry form records the end of a process it cannot see. Somebody reads for weeks, asks two people they trust, looks at three sites and often decides before they ever fill anything in. A campaign that produced the first visit in March gets no credit for the contract signed in June, and a last-click report will quietly recommend cutting the thing that started it.
That is not a tracking failure to be fixed with better software. A search click that becomes an enquiry the same week is easy to attribute. Work that made a buyer aware of a problem eleven months before they searched is invisible to every reporting tool in use. So the half with clean measurement keeps its budget, the half without it gets cut in a hard quarter, and two years later the pipeline stops growing while cost per lead climbs.
The fix is administrative. Split the budget on the report, with different targets, timescales and evidence for each half:
- Capture (search, retargeting, lead-gen ads) is judged on cost per opportunity and win rate by source, inside a stated window.
- Creation (content, video, brand, early account engagement) is judged on leading indicators read over a quarter: branded search volume, direct arrivals, and the share of new enquiries that already knew your name.
In an account-based programme the same logic puts coverage before revenue. On a nine-month cycle, revenue in month two is noise. How many target accounts are engaged at all, and how many people on each buying committee you have reached, move first and are the honest early evidence. Our account-based marketing guide sets out that report, and the demand generation guide covers why the two halves have to be budgeted apart.


