Marketing measurement used to promise something fairly simple. Track what people do, connect those actions to revenue, and you should be able to work backwards to see what helped create the result. The better the data became, the clearer the picture was supposed to get. Except that hasn’t quite happened.
Revenue teams now have campaign dashboards, CRM records, pipeline reports, conversion data and increasingly sophisticated attribution models. There’s more information available at almost every stage of the commercial process. Yet when someone asks a deceptively simple question like “What actually drove this revenue?”, the answer can become surprisingly difficult.
Part of the problem is that we’ve been asking different types of data to explain things they were never designed to explain. A click can tell us someone clicked. A CRM can tell us an opportunity progressed. Attribution can connect recorded interactions with an eventual outcome. None of those statements is wrong. They’re just answering different questions.
The Interactive Advertising Bureau’s State of Data 2026 research shows how far this idea has already spread across marketing measurement. Among 430 US buy-side decision-makers, 76 per cent use incrementality tests, 73 per cent use attribution analysis and 67 per cent use marketing mix models. Yet only 39 per cent use all three together.
That gap is interesting because these approaches aren’t really competing versions of the same thing. They’re different ways of understanding performance. Once we stop expecting one model to provide every answer, the question becomes less about finding the perfect set of revenue metrics and more about understanding what each one can legitimately tell us.
Attribution Has A Job, But It Can’t Do Every Job
Attribution is useful because revenue doesn’t happen inside a marketing dashboard. Something has to connect the activity marketing can see with the commercial outcome that appears later. Attribution models do that by assigning credit to recorded interactions along the path towards conversion or revenue.
The problem starts when that connection is treated as a complete explanation. A multi-touch attribution model might show that a buyer attended a webinar, visited a product page and responded to a campaign before becoming an opportunity. That tells us something valuable about the observable journey.
It doesn’t necessarily prove which interaction changed the buyer’s mind, or whether the deal would have happened without any of them. This distinction between association and causation is easy to lose once the numbers reach a dashboard. A precise percentage can look much more certain than the reasoning behind it actually is.
The model may be precise about how it assigned credit without being equally certain about why the buyer acted. That doesn’t make marketing attribution useless. Anteriad and Ascend2’s 2026 B2B Marketing Edge research found that just 18 per cent of more than 630 surveyed B2B marketers could track marketing activity through to closed revenue using a complete attribution model and treated attribution as a core measurement approach.
Those “Attribution Leaders” were also more likely to report significantly exceeding their primary marketing goals, at 45 per cent compared with 24 per cent of other respondents. Attribution clearly has commercial value when organisations can connect marketing activity with what happens further down the revenue process.
The mistake, then, isn’t using attribution. It’s expecting attribution to answer questions that belong somewhere else.
Start With The Commercial Question, Not The Metric
Most measurement conversations start with the data already available. Teams look at the dashboard, choose the KPIs they trust most and try to build a story from there. But this reverses the more useful order. The first question should be: what are we actually trying to know? If a demand generation team wants to know whether people responded to a campaign, campaign analytics may provide enough information.
If revenue leadership wants to know whether opportunities are moving faster, the answer lives much closer to the CRM. If the question is whether an investment caused additional revenue, neither view is sufficient on its own. This is why a marketing measurement strategy needs more than a common dashboard.
Different decisions require different kinds of evidence, and those differences shouldn’t disappear simply because several datasets can be pulled into the same reporting platform. IAB’s response to this wider problem has been Project Eidos, an industry-wide initiative launched in 2026 to improve consistency across outcomes, attribution, incrementality and marketing mix modelling.
Its focus is deliberately structural, looking at how measurement approaches connect rather than proposing another isolated solution. That gives us a useful way to rethink revenue measurement. Instead of ranking metrics from weakest to strongest, we can organise them around the commercial questions they’re capable of answering.
Build Revenue Measurement In Layers
A layered measurement model doesn’t mean piling more KPIs onto an already crowded dashboard. Quite the opposite. It means giving different forms of evidence a specific job, so leadership can see what happened, what changed and how confidently the organisation can connect one to the other.
Activity shows what happened
The first layer is the most visible because it sits closest to marketing activity. Impressions, clicks, downloads, registrations, content engagement and responses tell teams how people interacted with something they put into the market. Those numbers aren’t automatically “vanity metrics”. A click is perfectly good evidence that someone clicked.
The problem appears when interaction is stretched into a conclusion it can’t support. A download doesn’t prove purchase intent. A webinar registration doesn’t prove pipeline contribution. But campaign metrics can still show whether an audience responded, which messages attracted attention and where marketing activity produced observable behaviour. That makes activity data useful for optimisation, provided it stays in its lane.
Progression shows what moved
Once marketing activity reaches the commercial process, another set of measures becomes useful. CRM and pipeline metrics can show whether accounts became opportunities, whether deals progressed, how quickly they moved and where they stalled. This gives revenue teams something activity data can’t: evidence of commercial movement.
But movement still isn’t explanation. If an opportunity moves from one stage to another after several marketing interactions, a sales conversation and months of internal buyer research, the CRM can record the progression without identifying every factor that produced it. That distinction is important because progression is valuable even when causation remains uncertain.
Revenue teams need to know whether opportunities are moving. They simply shouldn’t ask that information to explain more than it can.
Outcomes show what the business gained
Eventually, the process produces an outcome. Revenue was won or lost. A customer was acquired. An account expanded. A deal generated a certain margin or created a particular customer lifetime value. These are the measures closest to the commercial result itself.
For leadership, this is where marketing ROI starts becoming more meaningful because the conversation moves beyond activity and pipeline into what the business actually gained. Yet even here, an outcome tells us what happened without automatically telling us what caused it. A closed deal is strong evidence of revenue.
It’s not, by itself, evidence that one campaign created that revenue. Which is why outcomes and attribution still need one another.
Attribution connects observable activity with outcomes
Attribution sits between the activities an organisation can record and the commercial outcomes it eventually sees. Its job is to create a structured relationship between them. That relationship can be extremely useful.
It can reveal common journeys, show which channels repeatedly appear around successful opportunities and help marketers understand how recorded touchpoints contribute across a longer sales cycle. But every attribution model is working with what it can observe.
The resulting answer is therefore partly shaped by the available data, the attribution window and the rules or modelling assumptions used to distribute credit. That’s enough for many questions. It isn’t enough for all of them.
Incrementality tests what changed the outcome
Sometimes the question goes one step further. Rather than asking which marketing activity appeared around an outcome, leadership wants to know whether the activity created an outcome that otherwise wouldn’t have happened.
That’s where incrementality testing enters the picture. Incrementality uses a comparison, such as exposed and control groups, to estimate the additional effect created by an intervention. Instead of simply observing that someone converted after seeing a campaign, it asks what would probably have happened if the campaign hadn’t run.
LinkedIn made this distinction explicit when expanding its Conversion Lift Testing in June 2026. It noted that first-touch, last-touch and multi-touch attribution can credit conversions that may have happened anyway, while lift testing is designed to estimate the additional conversions caused by the campaign.
That doesn’t make incrementality a more advanced replacement for attribution. It gives it a different job. Attribution helps explain the observable path. Incrementality helps test whether an intervention changed the result.
Market-level measurement shows how investment works together
There’s another point where following individual journeys stops being the most useful way to look at performance. Leadership may want to understand what happens when spending changes across several channels, how marketing performs alongside other commercial factors, or where the next portion of budget is likely to work hardest.
Those are broader portfolio questions. Marketing mix modelling, usually shortened to MMM, approaches measurement at this more aggregated level. Rather than reconstructing individual buyer paths, it uses patterns across investment and outcomes to estimate how different factors contribute to overall performance.
That makes MMM useful for strategic allocation while giving up some of the individual journey visibility that attribution provides. Again, the trade-off isn’t a flaw. It reflects the question the method was designed to answer. Put these layers together and the measurement picture starts becoming more useful.
Activity tells us what people did. Pipeline shows what moved. Outcomes show what the business gained. Attribution connects observed activity with those outcomes. Incrementality tests causal impact. Market-level measurement helps leadership judge investment across the wider portfolio. No individual layer needs to pretend it’s the whole picture.
Measurement Confidence Matters As Much As Measurement Coverage
Once several measurement approaches sit together, there’s a temptation to think the gaps have finally disappeared. More evidence must mean more certainty. It doesn’t quite work like that. Different methods create different levels of confidence. Campaign analytics may tell us with considerable certainty that a tracked action occurred.
We can be far less certain that the action caused a later sale. A CRM can reliably show a stage change while saying relatively little about everything that influenced it. This is where measurement confidence becomes as useful as measurement coverage. Revenue reporting needs to distinguish between what the organisation knows directly, what it can reasonably infer and what has been estimated through a model.
That distinction isn’t always visible today. LinkedIn’s 2025 B2B measurement research found that 64 per cent of marketing leaders said their organisations didn’t trust their existing measurement for decision-making. Another 87 per cent struggled to measure the long-term impact of campaigns.
A measurement system can therefore become more sophisticated without becoming more trusted. If leadership can’t see where evidence ends and assumptions start, another model may simply produce another number to argue about. Better revenue measurement doesn’t remove uncertainty. It makes the uncertainty easier to understand.
Revenue Reporting Needs To Match The Decision Being Made
This becomes especially important when the same marketing data travels through different parts of an organisation. A campaign manager deciding where to adjust spend needs relatively fast operational feedback. A revenue leader looking at pipeline quality needs a wider view across opportunities and accounts.
A CMO defending next year’s budget needs evidence that can stand alongside the financial and strategic decisions being made elsewhere in the business. One universal dashboard will struggle to serve all three equally well.
The pressure on those decisions is also increasing. Gartner’s 2026 CMO Spend Survey found that marketing budgets averaged 7.8 per cent of company revenue. At the same time, 56 per cent of CMOs said they lacked the budget required to deliver their strategy, while 54 per cent reported insufficient resources.
When resources are tight, the purpose of revenue reporting changes. It can’t simply document marketing activity after the fact. It needs to help leaders decide where money should go next, which investments deserve more time and where the available evidence supports a change in direction.
That also means marketing, sales and finance need shared definitions without forcing every system to produce the same answer. Agreement on what pipeline, revenue, acquisition or contribution means is useful. Expecting campaign analytics, attribution and finance reporting to arrive at identical conclusions is something else entirely.
The aim isn’t one dashboard containing every number. It’s the right evidence reaching the right decision.
Influence Still Exists Beyond What The Architecture Can See
Even a well-designed measurement architecture has a boundary. Not every part of a B2B buying decision leaves a useful trail behind it. Buyers research privately, speak to colleagues, consult peers, compare vendors and encounter information across places the seller may never see.
Increasingly, some of that research is also taking place through AI systems that summarise information before the buyer reaches a vendor-controlled channel. We’ve become very good at recording observable behaviour. That doesn’t make everything else irrelevant. This is where B2B buyer journey measurement needs a little humility.
An organisation can improve its attribution, connect its CRM, run incrementality tests and model broader investment without creating a perfect reconstruction of how every decision formed. Nor does it need to. The objective is to gather enough reliable evidence to make better commercial decisions.
Trying to eliminate every unknown can quickly become expensive, complicated and ultimately impossible, particularly as the buying journey becomes more distributed. The useful question is therefore no longer:
- “Can we measure everything?”
- “Do we understand enough to make this decision well?”
A Better Measurement Architecture Starts With Better Questions
Once measurement is organised around decisions rather than dashboards, the questions asked in leadership meetings start to change too. Instead of looking for one number that settles every debate, revenue teams can test whether the evidence they have is actually suited to the conclusion they’re trying to reach. Four questions provide a useful starting point:
- What decision are we trying to make? Campaign optimisation, pipeline management, budget allocation and strategic planning need different evidence.
- What does this data actually prove? Separate observed behaviour from inferred influence and tested causal impact.
- Which part of the revenue picture are we missing? Look across activity, progression, outcomes, attribution, causality and broader performance rather than assuming one layer fills every gap.
- Are we combining evidence or forcing agreement? Measurement triangulation works when several methods contribute different perspectives, not when teams keep adjusting them until every number tells the same story.
That last distinction may be the most important. The goal of a measurement architecture isn’t to create several systems that all agree with one another. If attribution, incrementality and market-level modelling answer different questions, some differences between their results are inevitable.
Those differences can even be useful. They tell leadership where certainty is strong, where the organisation is making an inference and where another type of evidence may be needed before committing more budget. The architecture works when each measure helps answer the decision in front of it.
Final Thoughts: Better Revenue Measurement Doesn’t Need One Version Of The Truth
Revenue teams aren’t short of numbers. If anything, the opposite is true. The harder part is understanding what those numbers are allowed to mean. Attribution still has a valuable role in that picture. So do campaign analytics, CRM data, pipeline reporting, commercial outcomes, incrementality and broader market-level models.
Problems start when one of them is promoted from useful evidence into a complete explanation of why revenue happened. A stronger revenue measurement approach accepts that different questions need different evidence. It also accepts that some conclusions will be more certain than others.
That may sound less satisfying than finding one definitive source of truth, but it gives leadership something much more useful: a clearer basis for making decisions. And that may be where marketing measurement is heading next.
Not towards measuring every interaction more aggressively, but towards becoming much more deliberate about what needs to be known, what can reasonably be known and which method can provide the best answer.
As buying journeys become harder to observe and commercial decisions become harder to attribute neatly, EM360Tech will continue exploring how marketing, sales and revenue teams can build the evidence they need without mistaking visibility for certainty.