The audience definition you write at launch is almost never the audience definition you need a year later. That is not a failure — it is the predictable consequence of writing the definition before the market has met the product. The work is not to write a more accurate first definition. The work is to write one that can update without losing its shape, and to know which parts of it must hold and which were always negotiable.

Why most audience definitions drift

The standard ICP exercise produces a document that is several things at once: a sales-targeting tool, a marketing-message anchor, a product-prioritisation input, and a brand-positioning reference. The trouble is that these four uses tend to pull the definition in different directions as the company grows. Sales wants the definition to reflect who actually closes; marketing wants it to reflect who responds to messaging; product wants it to reflect who uses the thing most; brand wants it to reflect who the company is for, in the longer sense. By month nine, these versions have diverged. The sales team is closing a segment the original document barely mentioned. The product team is building for a slightly different user. The marketing team is still running messaging that pitches the original audience. And the founder is increasingly unable to say, in one sentence, who the company is actually for. The definition has not become wrong; it has become silently plural.

The audience definition has two layers, and most people only write one

A definition that survives a year holds two layers in tension. The first is the strategic layer: the kind of organisation, person, problem, and moment the brand is built to serve. This layer should change slowly. If it changes inside the first year, the company was either misdiagnosed at launch or has pivoted, and that needs to be named. The second is the operational layer: the specific segments, accounts, channels, and signals the team is hunting within the strategic layer at this stage of growth. This layer should change quickly — possibly every quarter — as the team learns what is responsive and what is not. The mistake is to treat these two layers as the same document. When the operational layer changes, the strategic layer feels obsolete. When the strategic layer is rewritten, the operational layer becomes orphaned. A serviceable definition writes the two layers separately, names them as separate documents, and ties them together with a versioned reference. The strategic layer carries the brand promise. The operational layer carries the quarter's testable bets.

What belongs in the strategic layer

The strategic layer answers four questions and stops. Who is the brand for, expressed as a class of person or organisation? What problem does it solve for them, expressed as a job the buyer is trying to do? What is the moment in their work or life at which they need this kind of help? And what makes them the right audience for this company specifically, rather than for any of the alternatives in the category? Each of these four answers should be one sentence. Together they should be readable in under a minute by anyone in the company and recognisable by anyone in the audience. The strategic layer is what a new hire reads on day one to understand the company. It is what the founder reads before a fundraise to remember what they were doing. It is what marketing aligns messaging to. It is what product uses to test whether a feature belongs. What does not belong in the strategic layer is industry vertical, company size band, geography, or tech stack. Those are operational. The strategic layer is allowed to be aspirational by twelve months but not by twelve years.

What belongs in the operational layer

The operational layer is granular and disposable. It names the segments the team is actively pursuing this quarter, the channels they are testing, the messaging they are running, the conversion benchmarks they are measuring against, and the experiments they will close out at the end of the quarter. Some segments will graduate into the strategic layer's definition; most will not. The operational layer should be revisited at the end of every quarter and rewritten where the data demands it. The point of the rewrite is not to chase a new shiny segment. The point is to absorb what the company has learned about which versions of the strategic audience are responsive at this stage of growth and at this price point. A startup whose strategic audience is "operations leaders at mid-market industrial businesses" will probably find, over a year, that the responsive operational sub-segments are narrower — perhaps two named verticals and a particular role profile — and that the operational layer needs to reflect this without the strategic layer collapsing into it.

How to handle audience drift inside the first year

Most companies experience some form of audience drift in the first year. The signals usually come from sales — deals close faster in a different vertical than expected, or a different buyer persona keeps appearing on calls, or a different price point keeps emerging from the negotiation. The instinct is either to ignore the drift (the original definition was right; the market needs educating) or to fully replace the definition (whatever closes is the new audience). Both are wrong. The correct response is to ask whether the drift is operational or strategic. If the drift is in the operational layer — a different vertical, a slightly different buyer title — update the operational definition and keep moving. If the drift is in the strategic layer — a different class of buyer, a different problem, a different moment of need — that is a positioning question, not an audience question, and it deserves a different kind of conversation involving the founder, not just the head of growth. The diagnostic question is: would the audience that is showing up be served well by the company as it is currently built, or would serving them well require a different product, a different price model, or a different operational footprint? If the answer is "different product", the drift is strategic and the company is in the early innings of a pivot.

The signals worth tracking

Tracking audience truth requires four kinds of signal, not just CRM data. Closed-won data tells you who actually buys. Closed-lost data tells you who almost bought and what stopped them. Customer-success data tells you who renews and expands versus who churns. And finally, a small set of open-ended customer conversations — five or six a quarter — tells you what the bought audience is actually trying to do with the product, in their own words. The fourth source is the one teams under-invest in. The CRM tells you what closed; the customer says why. The closed-lost interview tells you what stopped a deal; the customer says what would have unstopped it. A quarterly rhythm of customer conversations — short, narrative, not survey-shaped — is the cheapest possible audience-update mechanism and the one most likely to survive in a busy operating cadence.

Common errors to watch for

Persona inflation. The definition grows from one persona to three to seven, each named after a fictional buyer with a job title and a coffee preference. By month eight, the definition is a personas deck rather than a strategic document, and nobody refers to it. The strategic layer should remain one or at most two persona-shaped statements. Vertical capture. A single early-customer vertical closes faster than the rest and the audience definition gets quietly rewritten to make that vertical sound strategic. This is dangerous when the vertical was an accident of network rather than a sustainable market segment. The operational layer can name the vertical; the strategic layer should not, unless the founder has consciously chosen vertical specialisation. Channel-driven definition. The team finds traction on one acquisition channel and the audience definition slowly conforms to whoever responds on that channel. The strategic layer should anchor the company against this drift, not be displaced by it. Quiet pluralism. The biggest failure mode is the one named earlier — sales, marketing, product, and brand each operate on slightly different audience definitions, none of them written down. The strategic layer exists to resolve this; it has to be enforced.

What This Looks Like in Practice

In the work with BGR, the original audience definition at launch named a particular type of customer at a particular scale band. By month nine, the company was closing deals from organisations one band larger and from one adjacent industry the original document had not mentioned. The team's first instinct was to rewrite the audience definition wholesale to match the closing data. The exercise that surfaced from working through the strategic-versus-operational split was different: the strategic layer held — the class of buyer and the job to be done were unchanged — but the operational layer needed to widen by one band and add the new vertical as a deliberate experiment. The brand positioning stayed coherent. The marketing operation gained a quarter of clarity. The founder could still answer the "who is this for" question in one sentence at the next investor update.

Closing

An audience definition that survives the first year is one that has been built to update without losing its shape. Two layers, written separately. Strategic layer changes slowly, names the class of buyer and the problem and the moment. Operational layer changes quarterly, names the segments and channels and tests. Drift inside the strategic layer is a positioning conversation; drift inside the operational layer is a normal quarterly rewrite. Anyone in the company should be able to read the strategic layer in a minute and recognise it in the work. If your audience definition has gone silently plural inside the first year — different versions held by different functions — we are happy to walk through what bringing it back into one shape would look like for your business.