.svg){kind=logo}
Google Ads says it drove 180 conversions last month. LinkedIn claims 95. Your CRM shows 110 new opportunities in total, and your CFO wants to know which channel to cut. Every platform is reporting accurately by its own rules, and none of them can answer the question you've been asked.
That gap is the difference between measurement vs attribution, and most B2B marketing teams treat the two as the same thing. This post explains where each one belongs, why attribution on its own leads PPC budgets in the wrong direction, and what a measurement set-up that holds up in a board meeting looks like.
Measurement vs attribution: the short version
Attribution decides who gets credit for a conversion. It looks at the tracked touchpoints before a sale and shares the credit between them, using last click, data-driven attribution or whatever model your platform applies.
Measurement asks whether your marketing caused the result at all. Would those deals have closed anyway? What would pipeline look like if you switched off Microsoft Ads for a quarter? Measurement deals in cause and effect, while attribution deals in bookkeeping.
The two answer different questions. Attribution is useful for day-to-day optimisation inside a channel. Measurement is what you use to decide how much to spend and where. Problems start when attribution reports get used to make measurement decisions, which is what happens in most B2B accounts we audit.
Why PPC attribution breaks down in B2B
Attribution models can only credit what they can see, and in B2B they see less every year.
Start with the buying journey. 6sense's 2025 Buyer Experience Report found buyers complete around 60% of their journey before speaking to a vendor, and 94% of buying groups have already ranked their preferred vendors by the time they make contact. Much of that research happens in places no tracking pixel reaches: peer conversations, review sites, Slack communities and, increasingly, ChatGPT. The same report puts the average buying cycle at roughly ten months. Most click-based attribution windows top out at 90 days.
Then there's the tooling. In 2023 Google retired first-click, linear, time decay and position-based models across Google Ads and GA4, leaving data-driven attribution and last click. Data-driven attribution is a real improvement, but it still only redistributes credit among tracked clicks. Consent banners, cross-device journeys and six people on a buying committee each researching separately all fall outside its view.
The result is predictable. Attribution overcredits the channels that sit closest to the conversion and undercredits the ones that created the demand. Branded search looks like your best campaign. LinkedIn looks expensive. Cut LinkedIn on that evidence and branded search volume often drops a few months later, long after anyone connects the two.
What proper B2B measurement looks like
You don't need an enterprise data science team to measure properly. You need three things, used for different jobs.
Offline conversion import: measure what sales actually cares about
The first fix is making sure your ad platforms learn from real business outcomes rather than form fills. Offline conversion import sends CRM stages (MQL, SQL, opportunity, closed-won) back into Google Ads, Microsoft Ads and LinkedIn, so bidding works towards qualified pipeline.
This is the foundation of how we run every account, and it matters most in sales-led SaaS and FinTech, where a demo request from a student and one from a Head of Finance look identical to the platform. Without it, Smart Bidding optimises happily towards the cheapest leads it can find.
Incrementality testing: prove a channel caused the result
Incrementality tests split an audience into a group that sees your ads and a control group that doesn't, then compare outcomes. The difference is the lift your ads caused, not the credit they claimed.
Until recently these tests were out of reach for most B2B budgets. In November 2025 Google cut the minimum spend for Conversion Lift studies from around $100,000 to $5,000, using a Bayesian methodology that reaches conclusions with smaller samples. Video, Discovery and Demand Gen campaigns can run them directly. Search and Performance Max still need a Google rep to set up, though geo-split tests you design yourself are an alternative.
If you're spending on awareness channels and struggling to justify them, a lift test is the most direct evidence you'll get.
Marketing mix modelling: set the budget split
Marketing mix modelling (MMM) looks at spend and outcomes over time across every channel, including offline activity and seasonality, and estimates each channel's contribution. It doesn't rely on cookies or clicks at all.
Google made its MMM, Meridian, open source and generally available in January 2025, which removed the licence cost. It still needs two to three years of reasonably consistent weekly data to be reliable, so it suits businesses spending across several channels at meaningful scale. For smaller accounts, well-designed lift tests and clean CRM data will get you further.
How to use attribution without being led by it
None of this means switching attribution off. It's still the right tool for decisions inside a channel: which keywords, ads and audiences to push harder, and where to shift bids this week. It just shouldn't set the budget split between channels.
A simple rule we use: attribution optimises, measurement allocates. In practice that means giving each channel a job and judging it against that job.
- Google Ads and Microsoft Ads usually capture existing demand. Judge them on cost per opportunity and CAC from CRM data, not platform-reported conversions.
- LinkedIn Ads often creates demand that search later captures. Judge it on pipeline from target accounts and on lift, not last-click CPL.
- Meta is audience-led for B2B and works best when the audience genuinely lives there. Test incrementality before scaling.
- ChatGPT Ads is new enough that attribution windows and conversion models are still evolving. Treat early spend as a test with its own measurement plan.
The same thinking applies after the click. If your landing page converts poorly, every channel looks worse than it is, and no attribution model will tell you that.
What changes by business model
The right measurement set-up depends on how you sell.
Product-led SaaS gets volume quickly. Free trials and sign-ups give ad platforms plenty of data, but the real test is which trials turn into paying accounts. Import activation or paid conversion as the primary goal once you have the volume. For Uplisting, a property management SaaS client, that meant tracking growth in ARR (from $100,000 to $3,000,000) rather than stopping at a 5x rise in monthly free trials.
Sales-led SaaS and FinTech have fewer, bigger conversions and longer cycles. Offline conversion import is essential, and so is patience: judge channels on a quarter of pipeline, not a week of leads. In regulated sectors lead quality matters even more, because unqualified leads take up compliance and sales time.
eCommerce has the cleanest attribution because the purchase happens online, but it still overcredits branded search and retargeting. Revenue-based ROAS from your store data is the starting point. Sprintlaw's ROAS moving from 0.8 to 2.3 is the kind of number that matters, because it was measured on revenue rather than clicks.
Where to start
If you only do one thing this quarter, connect your CRM to your ad platforms and start reporting on cost per opportunity alongside platform conversions. The gap between the two numbers tells you how far your attribution has drifted from reality.
Next, pick the channel you're least sure about and run a lift test on it. Then, if you're spending across four or more channels with a couple of years of data behind you, look at MMM to set the overall split.
Your CFO isn't asking which platform deserves credit. They're asking what would happen to revenue if you spent more, or less. If you'd like a straight answer on how your current set-up measures up, get a free proposal and we'll show you where your attribution is telling a different story from your CRM.
Last reviewed: September 2026 by Michéal Breslin, Managing Director
