%20png%20copy.avif)
B2B marketing has never had more data, more tooling, or more dashboards. It has rarely had less confidence in what any of it means.
Marketing budgets sit at 7.7% of company revenue, flat for the second year running, with martech taking roughly 22% of that spend. Yet Gartner's most recent martech research finds only 49% of stack capabilities are actively used, and just 15% of organisations qualify as high performers — those whose tools demonstrably hit goals and produce ROI. Spend on measurement keeps climbing while 75% of buy-side leaders now say their core measurement approaches — attribution, incrementality, MMM — underperform, according to the IAB's State of Data 2026.
The reflexive read is that this is a tooling problem. Buy more software. Hire a data engineer. Stand up a CDP. None of that is the actual problem. The problem is structural — four forces have moved the centre of gravity of B2B buying away from the things our data can see, and away from the things our metrics were designed to measure.
Most teams are still optimising the part of the iceberg above the waterline.
This piece is about the four forces: the 95:5 reality of long sales cycles, the dark funnel that has eaten visibility into the buying group, the wrong-instrument problem of attribution-led measurement, and the unglamorous hygiene gap that quietly costs more revenue than any of the above. None of them are new. What's new is how much they now compound.
The single most disruptive idea to land in B2B marketing measurement in the last five years is also the simplest. Professor John Dawes of the Ehrenberg-Bass Institute, working in partnership with the LinkedIn B2B Institute, formalised what is now called the 95:5 rule: at any given moment, only about 5% of potential B2B buyers are in-market for a given category, while the remaining 95% are out-of-market. The split is not a matter of taste. It's a structural property of how B2B categories work. Average contract lengths in many B2B categories sit around five years, which means roughly 20% of the market is in-market across a year, but only about 5% in any given moment.
This is the number that should sit on every CMO's wall, because almost every default measurement system contradicts it.
Lead generation captures a fraction of the 5%. CRO optimises the experience of the 5%. Attribution measures who closed the 5%. Pipeline reporting tracks the velocity of the 5%. Demo bookings, MQLs, SQLs, conversion rates, cost-per-lead — all of them are instruments calibrated to in-market behaviour. They are useful. They are not what describes the marketing function's actual job.
The job, as Dawes and the Ehrenberg-Bass team have articulated it, is to build mental availability — the probability that your brand comes to mind in the buying situations buyers face when they finally enter the market. Mental availability is a stock, not a flow. It accumulates from years of consistent presence with the 95%. It is invisible to last-touch attribution. It is invisible to MQL counts. It is invisible to most of what marketing reports up to the board.
When 6Sense surveyed more than 4,000 buyers across North America, EMEA and APAC, they found buyers don't engage with sellers until they're roughly two-thirds of the way through their journey, and that 94% of buying groups had ranked their preferred vendors before any first contact, ultimately purchasing from that preliminary favourite 77% of the time. 6Sense's data also shows the average B2B sales cycle ran 10.1 months in 2025, down from 11.3 months in 2024, with the point of first seller contact moving from 69% to 61% of the journey.
Read those numbers carefully. The window in which marketing can actually shift a buyer's decision sits before the data starts collecting. By the time someone fills out a demo form, the marketing job is largely done — for better or for worse. The vendor that became part of the consideration set six months ago by being present, credible, and easy to remember is the one that gets shortlisted today. The vendor that wasn't isn't.
This rewires what counts as a leading indicator. Pipeline isn't leading; it's a deeply lagging indicator of mental availability built years earlier. Branded search volume is leading. Share of search is leading. Unprompted recall in customer interviews is leading. The cadence and reach of your brand's appearance in the channels your future buyers occupy — LinkedIn, podcasts, industry publications, peer communities, AI-generated answers — is leading. Almost none of that lives in the marketing dashboard.
There is, fairly, an opposing perspective worth airing. Not everyone in the marketing science community accepts that B2B and B2C behave identically under Ehrenberg-Bass laws, and some practitioners argue that the 95:5 ratio is itself a category-level average that varies meaningfully — Peter Weinberg of the LinkedIn B2B Institute has himself published work showing the ratio shifts by category and contract length. For categories with shorter contracts, more of the market may be in-market at any given time. The framework is directionally right; the exact ratio for your category is an empirical question, not a constant.
But the directional point is what matters for measurement design. If even half of buyers are out-of-market at any point, then a measurement system that ignores them is missing half the marketing job. The data you have describes activation. The job is also creation. They require different instruments.
Mike Maynard's MarTech piece notes, accurately, that working out which company a website visitor works for is harder than ever. That sentence understates the scale of what's happening.
Anonymous browsing is not a marginal phenomenon. Across the major dark-funnel research bodies, estimates converge: Forrester places the share of the B2B evaluation that happens in channels marketing cannot capture at 70 to 80%, and 6Sense's 2025 work puts the share of the journey buyers complete before any seller contact at roughly two-thirds. On the website itself, only about 3 to 3.5% of unique visitors self-identify via a form fill, leaving roughly 97% of traffic anonymous, per joint research from Unbounce and 6Sense.
The infrastructure that used to convert that anonymity into intelligence is failing in three ways simultaneously.
Forms have collapsed as a willingness signal. Buyers know that filling a form means a sales sequence. They prefer to defer that contact for as long as possible. Gartner's most recent technology buyer research shows 61% of B2B buyers prefer a fully rep-free buying experience, and 73% actively avoid suppliers that send irrelevant outreach. HubSpot's own strategic pivot is the loudest signal of how broken the form-first playbook has become. The company that built itself on the inbound methodology renamed its flagship Inbound conference to Unbound in 2026, acknowledging that the original framework — built around content gating, form fills and lead scoring — no longer matches how buyers behave. When the company that defined inbound steps back from the brand, the broader category should pay attention.
Cookie consent has eroded the second-best signal. A long-running behavioural study by Advance Metrics tracking interaction with cookie banners over five years post-GDPR found total acceptance rose to 25.4% in their dataset, but 68.9% of users now close or ignore the banner outright, withholding consent and producing significant data loss for performance measurement. Etracker's benchmark research on legally-compliant German cookie banners reports an average 60% rejection rate for non-essential tracking. The number that matters for B2B specifically is grimmer — average B2B website acceptance rates sit around 80% but can drop to 40% in some categories. Add ad-blockers, third-party cookie deprecation in major browsers, and remote-work-driven IP fragmentation, and the addressable population for retargeting and behavioural attribution shrinks every quarter.
The buying group itself has expanded beyond what individual-level tracking can capture. Even when you can identify one anonymous visitor, you're still looking at one node in a network. Gartner's research now puts the average enterprise B2B buying group at five to 11 stakeholders representing roughly five distinct business functions, with buying groups ranging from five to 16 people across as many as four functions in their 2024 sales survey. Forrester's 2025 buyers' journey survey describes the typical purchase as involving 13 people inside the buyer's organisation and nine outside it, spanning three or more departments. Buyers also rely on between 15 and 27 information sources during evaluation, per Forrester. Each member is researching independently, often in different channels, often invisibly to the vendor.
Layer AI on top of all this and the dark funnel deepens. 94% of B2B buyers now use large language models during their purchasing process, and 89% ultimately purchase a solution with AI features. When a VP of Operations asks ChatGPT to compare three vendors in your category and walks away with a shortlist, no pixel fires, no cookie tracks, no UTM lands in your CRM. The recommendation that shaped the deal was made in a channel you cannot see, by a system you don't control, citing sources you didn't write. Forrester's analysis suggests this AI influence layer typically gets zero attribution credit — the buyer eventually Googles your brand directly and your "first touch" is recorded as branded organic search.
The implication for measurement is uncomfortable. Most of the visibility infrastructure built between 2015 and 2022 — marketing automation platforms, behaviour-based scoring, retargeting pixels, multi-touch attribution — was designed for a world where buyers signalled interest early and visibly. They no longer do. The signals haven't stopped existing; they've moved into channels where the measurement infrastructure isn't.
The teams making the most progress here aren't trying to recreate full visibility. They're accepting that the dark funnel is a permanent feature of the landscape and instrumenting around it. That means leaning on account-level signals — anonymous account engagement on the website, third-party intent data from providers like 6Sense or Bombora, review-site activity from G2 Buyer Intent, anonymous visitor identification through tools like RB2B — and treating those signals as influence inputs to account scoring rather than as conversion inputs to attribution.
It also means accepting a different reporting standard. 6Sense data also shows that even Google search now ends without a click 58.5% of the time, per Search Engine Land's analysis. The honest answer to "what brought this lead in?" is increasingly: many things, most of them invisible. Pretending otherwise is what produces dashboards full of false precision.
When marketing measurement broke, the response of most teams was to add more attribution. Move from last-touch to multi-touch. From multi-touch to data-driven. From rules-based to algorithmic. Each generation of the model promised more precision. Each one ran into the same wall: attribution was answering the wrong question.
Attribution answers a backward-looking question — given that this conversion happened, how much credit does each touch deserve? It assumes every recorded touchpoint contributed, which it cannot prove. As Directive's measurement framework lays out, attribution has no mechanism to ask whether the buyer would have converted anyway, which is the fundamental question that budget decisions actually require. It assumes a clean, trackable journey, which most B2B journeys aren't. And it inherits every blind spot the previous section described — anonymous research, dark social, AI-mediated discovery, offline conversations, peer recommendations.
Despite that, an estimated 67% of B2B marketing teams still default to last-touch attribution in 2026, crediting the final form fill while ignoring 26+ touchpoints that came before it. Forrester's research suggests modern B2B buyers traverse more than 27 touchpoints across the cycle. Single-touch attribution against that reality produces fiction in dashboard form.
Two alternative measurement frames have moved decisively from the periphery to the centre of the conversation in the last 18 months: incrementality testing and marketing mix modelling.
Incrementality testing asks the forward-looking question attribution refuses to answer: if we removed this campaign entirely, would these conversions still have happened? The answer is established by holding out a randomised cohort from a treatment — a paid campaign, an outbound sequence, a piece of personalisation — and comparing the control to the treated group. The standard implementation uses geo-holdouts or audience holdouts, with statistical synthetic controls when randomisation isn't practical. Adoption has accelerated sharply: 27.6% of US marketers in a recent EMARKETER and TransUnion survey now rate MMM as their most reliable measurement methodology, with multi-touch attribution at 19.4% and unified measurement at 18.9%.
Marketing mix modelling takes a different angle. Rather than tracking individual journeys, MMM uses statistical time-series analysis on aggregate spend, impressions, and outcome data — typically across two to three years of weekly history — to estimate the incremental contribution of each channel to outcomes like pipeline or revenue. As Improvado's primer notes, MMM requires no cookies, device IDs or consent signals, which makes it the default measurement framework in privacy-regulated industries in 2026. Crucially for B2B, MMM can be calibrated against intermediate outcomes — pipeline movement, opportunity creation, demo bookings — that occur within weeks of marketing exposure, rather than waiting for closed-won revenue eight months later.
Neither is a panacea, and the honest case for them includes their limits. A counter-perspective worth airing: CaliberMind argues that incrementality testing struggles in B2B specifically because tests typically require a 5%+ revenue impact to yield valid results, and the long-tail B2B influence cycles often exceed standard test windows. Asking a B2B company with a small number of large customers to deliberately withhold marketing exposure for a quarter — the time it takes for an average deal to close — is often impractical at the channel level. MMM has its own constraints. As Improvado notes, B2B MMM works only if the outcome variable matches the cycle length; modelling closed-won revenue against marketing spend with a nine-month lag is mathematically punishing, and most teams need to model leading-stage outcomes (MQL→SQL, SQL→Opportunity) instead. And as Gartner's CMO research confirms, MMM and incrementality both need clean, consistent two-to-three years of aggregated data — something most growth-stage companies haven't accumulated yet.
The most defensible posture is that none of these methods is sufficient on its own. The maturing view, sometimes called unified marketing impact analytics, organises measurement into three layers: a clean unified data layer at the foundation; an attribution layer for in-channel optimisation on a weekly cadence; and an MMM and incrementality layer at the top for quarterly budget allocation and annual planning. Each layer serves a different decision. Each layer fails when asked to answer the others' questions.
The strategic reframe worth internalising is this. Attribution belongs to the campaign manager — it informs which keyword, which creative, which audience, which bid. MMM belongs to the CMO — it informs how much goes to brand vs. demand, which channels are saturated, where the next dollar produces the next dollar of pipeline. Incrementality belongs to the experiment design — it validates whether anything in the model is causal or just correlated.
Most B2B teams have built one layer (attribution) and asked it to do the work of all three. That is the source of an enormous amount of the false precision marketers complain about. The fix isn't more attribution. It's a different organising principle for the measurement stack.
Maynard's piece ends on a deceptively simple point: even the things B2B marketers say they do well, they often don't. He cites a test his agency ran in the engineering sector — signing up for newsletters at a sample of companies — in which around half of those companies sent nothing afterwards. That anecdote is not a curio. It's a window into a category-wide hygiene gap that is almost certainly the single most expensive measurement failure in the room.
The single most studied operational metric in B2B is speed-to-lead. The research is unambiguous, and it has been for over a decade. The original MIT and InsideSales.com Lead Response Management study analysed 15,000 leads across multiple industries and established that companies contacting a lead within five minutes are 100x more likely to make contact than those waiting 30 minutes, and 21x more likely to qualify the lead. Velocify research found that contacting a lead within one minute boosts conversions by 391%. Harvard Business Review's analysis showed a 60x reduction in qualification likelihood for companies that waited 24 hours versus the first hour. 78% of customers buy from the first vendor that responds.
What B2B teams actually do is response in an average of 42 hours. InsideSales' 2021 review of 55 million sales activities across 5.7 million inbound leads found 57.1% of first call attempts occurred more than a week after the lead came in, and only 0.1% were engaged within five minutes. Workato's audit of 114 B2B companies found that just one of the 114 sent a personalised email within five minutes, and that nearly 1 in 5 didn't email at all. More than 30% of inbound leads are never contacted at any point, per Forbes research.
Frame the gap commercially. If a company is spending $500K a year on demand generation and 42-hour response is costing 21x in qualification rate at the front of the funnel, the dollar value of that hygiene gap exceeds the dollar value of nearly any model refinement, technology purchase, or attribution upgrade. It is, in bald financial terms, the largest leak in the system. And it sits in a budget line nobody fights for.
The hygiene problem extends well beyond speed-to-lead.
Strategy hygiene. Per Content Marketing Institute's 2026 B2B benchmarks, only 22% of B2B marketers describe their content marketing as extremely or very successful. The leading reason isn't poor writing. It's the absence of a documented strategy: 42% of B2B marketers with moderate or lower content success cite a lack of clear goals as a contributing factor, and 56% report difficulty attributing ROI to content efforts as a top measurement challenge.
Tooling hygiene. Gartner's martech research finding that only 49% of stack capabilities are actively used means the average enterprise is paying for half a stack that doesn't get used. Add the 22% of marketing budget that goes to martech and the implied annual waste at the category level runs into the billions.
Personalisation hygiene. Maynard makes the point cleanly: most B2B companies that run ABM campaigns deliver custom content only on campaign-specific landing pages, not on the core website, even when they know exactly who the visitor is. Personalisation lives in the campaign, not the experience. The result is a brand that promises tailored relevance in the demand-gen ad and delivers a generic homepage on the click-through.
Process hygiene. Gartner's 2024 sales research found 74% of B2B buyer teams demonstrate unhealthy conflict during the decision process, with buying groups that achieve consensus 2.5x more likely to close high-quality deals. Most marketing programs are built to convert individuals, not to facilitate group consensus — yet group consensus is the actual unit of decision. Sales enablement tools that help buying groups validate decisions internally are vastly under-developed compared to lead capture tools that already work fine.
There is a cultural reason hygiene loses every budget battle: it is not glamorous. No one gets promoted for cutting average lead response time from 42 hours to two minutes. No one writes a LinkedIn post about deduplicating their CRM. No one wins at a marketing awards show for a documented messaging framework. The unsexy work has no champion, which is exactly why it persists as the largest source of leakage in the system.
The teams that quietly outperform almost always have this part disproportionately right. They've fixed the basics first, then layered the sophistication on top. Most teams do it in the opposite order.
Each of these forces is meaningful on its own. The harder problem is that they multiply.
The 95:5 reality means most data describes the wrong audience. The dark funnel means even the data that does describe in-market buyers misses most of the journey. Attribution-led measurement assumes visibility that no longer exists, then assigns false precision to the fragments it can see. And operational hygiene gaps mean that even the in-market buyers who do come through the visible front door are mishandled at the moment of highest intent. Each force amplifies the others.
The losing posture is recognisable. It looks like more dashboards. More martech tools. More attribution refinement. More sophisticated lead scoring. The investment goes into squeezing more confidence out of measurement systems that were calibrated for a different buyer behaviour. The dashboard gets prettier; the underlying decisions don't get better.
The winning posture looks different. It accepts that visibility into the modern B2B journey is structurally lower than it was, and stops trying to recreate full transparency. It separates the layers of measurement so each one answers a question it can actually answer — attribution for tactical optimisation, MMM for strategic allocation, incrementality for causal validation, qualitative customer research for what none of them can see. It treats the 95% as a reach-and-memory problem, not a conversion problem. And it spends a disproportionate share of attention on the hygiene basics that no consultant, vendor, or AI agent will ever advocate for, because those basics are where the largest unrecovered revenue actually sits.
The companies that get this right are not better-instrumented. They are better-organised around the right questions.
Stop optimising for visibility you don't have. Start instrumenting for influence you can't fully see. Replace last-touch attribution as the budget defence and treat it as a tactical optimisation tool only. Build at least directional incrementality testing into recurring planning cycles. Where MMM is feasible — where 18-24 months of clean spend and outcome data exists — start on intermediate outcomes (pipeline stage movement) before attempting closed-won. Where it isn't feasible, accept that brand and out-of-market investment will be defended on share-of-voice, share-of-search, and unprompted recall research, not on revenue attribution. None of these are perfect. All of them are better than false precision.
Re-categorise marketing investment by audience temperature, not channel. The 95:5 frame implies a portfolio split that very few B2B marketing teams actually run. Most over-invest in the 5% (paid search, demo capture, retargeting, BDR outbound) and under-invest in the 95% (brand campaigns, thought leadership reach, podcast presence, category education, AI visibility). The Ehrenberg-Bass and LinkedIn B2B Institute work suggests a portfolio closer to 60/40 in favour of long-term brand-building for most B2B categories. Most marketing P&Ls invert that ratio.
Treat the website as a buying-group destination, not a lead-capture funnel. If 73% of buyers are using AI to research and 92% arrive with a vendor in mind, the homepage's job is no longer primarily to harvest a form fill from the 5%. Its job is to confirm the brand they remembered, validate the choice the buying group is converging on, and equip the internal champion to sell the decision inside their organisation. That's a different content brief, a different navigation structure, a different metrics stack. Bounce rate and form conversion are the wrong scorecard for a job that's increasingly about validation depth and time-on-decision-content, not capture.
Audit the unsexy stuff before buying anything new. Before the next martech purchase, run two diagnostics. First: what's the actual lead response time from inbound demo request to first human touch, measured in minutes, not days? Second: what percentage of the existing martech stack is used to capability — not to login? These numbers are usually worse than people expect. Both are typically a higher-leverage place to put a quarter's effort than any new tool, and neither requires another procurement cycle.
Build self-reported attribution into your reporting. The cheapest, most under-used signal in the B2B measurement stack is the question every customer can answer that no platform can: "How did you first hear about us?" Asking that question of every closed-won customer for a year produces a qualitative attribution layer that catches the dark funnel signals every quantitative model misses. It's not statistically rigorous. It's directionally honest, which is more than most attribution dashboards can claim.
The fundamentals of B2B marketing have not changed. Buyers still decide based on whether they trust a vendor, whether they believe the product fits, whether the case for change is stronger than the case for inaction, and whether they remember the brand at the moment they need it. None of that is new.
What has changed is the visibility of how those decisions get made. They get made earlier, in more places, by more people, with more independent inputs, in channels that none of the measurement systems built between 2010 and 2020 were designed to see. The pipeline shows up later than the marketing job ended. The attribution dashboard credits the touch that happened to land last. And the team chases the metric that the dashboard happens to surface.
The B2B marketing teams that win in 2026 will not have more data than the teams that don't. They will have better questions about which data matters for which decision. They will have separated the measurement layers, accepted permanent visibility limits at the brand level, and rebuilt their measurement stack around what they're actually trying to influence — which is increasingly the 95% they cannot see, not the 5% their CRM can.
More data was never the answer. Better questions are.
Explore Big Moves Marketing services and resources:
%20-%20white.svg)