Why ICP Misalignment is Now a Revenue Risk
Four in five B2B marketing leaders say they are confident in their Ideal Customer Profile. Almost two-thirds have already seen ICP misalignment hurt pipeline or revenue. That gap between confidence and reality is what this webinar set out to unpack.
Hosted by Resonance and Agentcy, and joined by Emma Gair (Marketing Business Partner at Ekco) and Luke Farrugia (Founder of SignalForge and VP Global Marketing at Explori), the session walked through new research into ICP alignment, shared lived experiences of what drift actually costs a business, and laid out a practical, AI-native route back to a dynamic ICP that earns its keep.
Key Takeaways
- A segment list is not an ICP. Backward-looking workshops that describe your biggest customers produce a segment list with heavy survivorship bias. An ICP defines who you should be selling to, why they are ready to buy now, and — critically — who you should walk away from.
- The confidence gap is real. 82% of B2B marketing leaders are confident in their ICP, but 62% admit misalignment has already hit pipeline or revenue. Only 38% have reviewed their ICP in the past year, and almost a third are running on assumptions rather than evidence.
- The cost is compounding, not acute. Misalignment rarely shows up as one bad quarter. It manifests as wasted spend (47%), low conversion rates (33%), and mis-targeted marketing and PR (over 25%) — and, per Pavilion, roughly 60% of pipeline sits in segments with diminishing returns.
- A dynamic ICP has three layers. Quantitative foundation (win rates, velocity, retention, expansion by segment), situational layer (what is happening in the buyer's world right now), and voice-of-customer layer (the language buyers use, not the language your marketing team uses).
- ICP ownership is shifting towards the GTM engineer. A dynamic ICP requires a system that aggregates signal intelligence across CRM, conversational intelligence, product usage, ad platforms, and third-party signals — and that needs someone who can engineer and run it, sitting close to RevOps.
- PR depends on ICP more than most functions admit. Without a clear ICP, PR gets reduced to coverage-for-coverage's-sake. With one, it becomes the eyes and ears of the business — monitoring where buyers engage, feeding back on narrative, and earning its place in C-suite conversations.
- AI accelerates, but does not replace, the fundamentals. Governance, validation with SMEs and real customers, and iterative testing matter more than the tool choice. Feeding a stale ICP into AI just produces stale output faster.
- If you do one thing in 30 days, do the quantitative analysis. Pull deal data, retention and expansion numbers, and lay them forensically against your current ICP. If the picture matches your internal narrative, you have a foundation. If it doesn't, you have surfaced the most commercially important conversation your leadership team will have this year.
The Research: A Confidence Gap Hiding in Plain Sight
The session opened with fresh research commissioned among senior B2B marketing leaders across the UK and US. The headline finding was stark. Asked whether their ICP was strong and fit for purpose, 82% said yes. Asked whether misalignment had damaged pipeline or revenue, 62% said yes.
That gap is the story. It is entirely possible to feel confident in an ICP that is simultaneously costing you money — because the confidence is usually anchored to a document that was signed off at some point, not to the evidence the market is producing this quarter.
Supporting findings reinforced the pattern:
- Only 38% had reviewed their ICP in the past year, despite more than three quarters believing regular review was important.
- Almost a third were running an ICP built largely on assumptions rather than evidence.
- The operational costs showed up as wasted spend (47%), low conversion rates (33%), and misdirected marketing and PR — targeting the wrong journalists, the wrong publications, and the wrong audiences — for more than a quarter of respondents.
Why a Segment List Isn't an ICP
Luke Farrugia's framing cut through one of the most common sources of confusion in go-to-market strategy. Most organisations that say they have alignment on their ICP have alignment on a segment list. Those two documents do very different jobs.
A segment list is retrospective — it describes who you have sold to. An ICP is prospective — it describes who you should be selling to, why they are ready to buy, what makes them convert efficiently, and who you should actively deprioritise. The tell is in how it was built: if the definition came out of a workshop where smart people examined the largest existing customers and agreed on common traits, that is a segment list dressed up as strategy.
The survivorship bias is the kicker. Accounts that have been with you for a decade and grown into seven-figure relationships feel like ideal customers, because they stayed. That perception is shaped by retention, not by evidence that the profile represents your best future opportunity. The Pavilion Report puts a number on what this costs: around 60% of pipeline sits in segments with diminishing returns. The team is working hard, but in the wrong direction.
The Three Layers of a Modern ICP
Luke's operating model for ICP has three components. Without all three, you have a well-informed hypothesis rather than an ICP.
- The quantitative foundation. Win rates, deal velocity, revenue concentration, and — crucially — retention and expansion patterns by segment. Expansion and churn tell you who you keep winning with; win rates and velocity tell you where the market is actually pulling you. This is the most important foundational piece.
- The situational layer. Beyond firmographics, what is happening in the buyer's world that makes them ready to buy now? This might be a change in compliance legislation, a new exec hire, a shift in their own customer base — signals that move a dormant account into an active one.
- The voice of the customer layer. The language buyers use to describe their own problems, not the language your marketing team uses to describe your solution. This is where most messaging breaks down in practice.
The cost of running a go-to-market (including a PR programme) on an untested hypothesis is not theoretical. Benchmark data in the Pavilion Report attributes a 75% reduction in win rates to ICP misalignment. That is the difference between a business that compounds and one that grinds to a halt.
What ICP Misalignment Actually Costs PR
Claire Williamson's view from the PR side was candid. When a brief lands from a client, the ICP is hardly ever included in any meaningful form. The agency ends up second-guessing — and that second-guessing is what makes it impossible to answer the C-suite question everyone eventually asks: what impact is this PR activity having on the business?
Without an ICP, PR defaults to coverage for coverage's sake. With one, PR earns the role it should be playing: the eyes and ears of the organisation, a reputation and risk function, an active feedback loop on how buyers engage with messaging and narrative. That feedback is how you catch misalignment early — and it only works if the ICP is the building block.
This matters more than ever because, as Claire put it, there is no such thing as owning a narrative anymore. You have to jostle for it — on social, in the press, and increasingly across AI-driven discovery. Jostling without a clear idea of whose attention you are competing for is just noise.
When the Document and the Reality Diverge
Luke described the lived experience that defines misalignment for most operators: the ICP as a document and the ICP as the operating reality become two different things. A big logo comes in that doesn't fit the profile, but the business wants to land it. A seller with domain expertise in a different sector needs to hit their quota. Leadership gets excited about a different market. Each of those decisions is defensible in isolation; together, they quietly erode the ICP.
In businesses that have grown through acquisition the problem compounds. Multiple ICPs inherited from multiple acquired organisations, never reconciled into a single operating reality, and no one bold enough to force the conversation. The cost does not arrive as a bad quarter; it arrives as years of compounding misallocation — and the signals, from loss reasons to customer sales conversations, are usually there if you build the system to surface them.
Who Should Own the ICP?
As ICPs become dynamic — updating continuously based on new signal intelligence — ownership has to shift. Luke's view is that ownership is moving toward the GTM engineer persona, sitting close to RevOps. The rationale is simple: if you need a system to give you the intelligence to define the ICP dynamically, you need someone who can engineer and run that system across every intelligence layer.
The benefit flows everywhere. The CRO benefits from an AI-native system that surfaces which accounts to go after, and why, because buyer behaviours in a segment shifted overnight. Marketing benefits from content briefs that adapt on the basis of what is actually driving efficiency through the pipeline. PR benefits from a live view of which segments, stories, and narratives are earning engagement. None of this is owned by one function — but someone has to own the system.
Adopting AI Without Losing the Fundamentals
Emma Gair's lens on AI adoption was refreshingly practical. The process at Ekco has three parts:
- Tooling and platforms. Start with what is already in your governance envelope. For a Microsoft partner, that means Copilot first — but Claude might be better for specific tasks. HubSpot's AI. Agentcy for external visibility. The point is not the tool, it's bringing each capability in under the right governance and security framework. If you are not careful about AI governance, it will bite you.
- Validation. AI can produce extraordinary insight, but garbage in, garbage out still applies. Outputs need validating with internal SMEs, customers, partners, and real qualitative and quantitative research. Feeding a stale ICP into AI just produces stale output faster.
- Iteration. Ship campaigns, read the response, adjust. Don't reinvent the wheel; refine it based on evidence.
From the PR side, Claire made the same point about validation: AI outputs need to be sense-checked against external data. And the reason Agentcy exists is precisely this — surfacing the signals (what's being said about you, where, by whom, across media and AI) that make validation possible in the first place.
What AI-Native Go-to-Market Actually Looks Like
Luke's description of an AI-native go-to-market was the session's most concrete picture of where this is all heading. The foundational layer aggregates every signal across proprietary systems: CRM, conversational intelligence, product usage, outreach (Lemlist, Outreach, Salesloft), first-party data, second-party data (LinkedIn Ads, Google Ads), and a growing set of third-party signals — including niche ones like compliance legislation changes that materially shift the value of your solution overnight.
Pattern analysis across that aggregate is what identifies where to reallocate investment — which segments to lean into, which sub-segments perform better than others, which messages to sharpen. And crucially, it is a feedback loop, not a snapshot. Outbound performance, response rates, response sentiment, deal velocity, conversational intelligence — everything feeds back into the next cycle of account identification, messaging, and content.
Ideation stops being "someone had a good idea" and starts being "the market, the voice of the customer, and the efficiency of the pipeline told us what to prioritise." That is the shift.
If You Do One Thing in the Next 30 Days
Asked for the single most impactful move in the next 30 days, Luke's answer was unambiguous: do the quantitative analysis. Pull your deal data. Look at retention and expansion numbers. Lay them alongside your static ICP and be forensic about it.
If the picture that emerges matches your internal narrative, you have a solid foundation to build on. If it doesn't, you have just surfaced the most commercially important conversation your leadership team can have this year — and every downstream decision (targeting, messaging, PR, content, sales prioritisation) gets better from that point.
Then layer in the conversational intelligence — every sales call, every customer success conversation, every lost-deal debrief. Correlating the qualitative with the quantitative is where the real insight lives. Signals come third — the leading edge, but meaningless without the foundation underneath.
Where to Learn More
Panellists shared the places they turn to for sharper thinking on ICP and AI-native go-to-market:
- Pavilion — a community of GTM leaders experimenting with AI-native approaches in the open, and the source of much of the benchmark data referenced in the session.
- Clay — widely credited with crystallising the GTM engineer movement. A wave of sharp operators has emerged around it, publishing hands-on playbooks on LinkedIn and Substack.
- Vendor ecosystems — for organisations already partnered with hyperscalers (Microsoft, AWS, etc.), their enablement programmes are often underused sources of insight.
- Books and voices worth following — Jamie Bartlett's How to Talk to AI, Magnus Consulting, and journalist Greg Orme were name-checked as regular reading.
And the single best learning hack, in Luke's view: build, don't buy. Prompt engineering and vibe coding — with tools like Base44 or Lovable — accelerate learning faster than any course. Start low-stakes, fail fast, and systemise the inefficiencies you find along the way. Everyone has great ideas; the learning comes from articulating them in a prompt, seeing what the output tells you, and iterating.
Watch the Recording & Download the Slides
The full 55-minute recording is embedded at the top of this page, and the research slides used in the session are available to download as a PDF.