Why doesn’t sales use our intent data?
Because intent data arrives as a score with no reason, no owner and no next action attached. Reps are asked to trust a number they cannot interrogate, about accounts they were not already working, with nothing stating why this moment matters. The data is usually fine. What is missing is the decision layer that converts an observation into a specific, explainable instruction.
What a rep actually receives
In most implementations, a rep opens a dashboard or a Slack channel and sees a company name and a number. Sometimes a topic. The implicit instruction is: this account is interested, go and do something.
Consider what that asks of them. To act, the rep must:
- Trust a score whose derivation they cannot inspect.
- Accept that an account they were not working is now urgent.
- Invent a reason for contact, because none was supplied.
- Set aside the accounts they were already progressing.
A rational rep declines. Not because they are resistant to data, but because the request carries no information they can act on. The dashboard sits unopened, and the eventual conclusion is that sales will not adopt the tool.
The missing layer
Between an observation and an action sits a decision. It is a real layer with real content, and in most companies nothing occupies it. Its job is to convert something happened into this person should do this specific thing, now, for this reason.
A decision layer that works states four things:
- What changed — the specific observation, dated. Not a score.
- Why it matters — the inference connecting the observation to a reason to act.
- Who owns it — a named person, not a queue.
- What is expected — the action, and the window it is expected in.
A rep given those four things does not need to be persuaded to use intent data. They have been handed something writable: a fact about the account, and a reason this week is different from last week.
"Sales doesn't use our intent data" is almost always a decision-layer failure being reported as an adoption problem. The tell is that the proposed fixes are enablement sessions and dashboards — interventions aimed at rep behaviour rather than at the thing producing it.
Why a score is the wrong output
A score compresses many observations into one number, and in doing so destroys the only part a rep can use. "Acme is an 82" cannot be put in an email. "Acme posted four roles reporting to a VP of Revenue Operations they hired in March" can be — it names something true, recent and specific, and the outreach writes itself.
Compression is useful for prioritising a list. It is useless for justifying a conversation. Most implementations deliver only the compressed form and then wonder why the conversation does not happen. The same mechanism drives why reps don't trust lead scoring.
The argument that never gets settled
Every quarter the same exchange: marketing says the signals are strong and sales is not working them; sales says the accounts are not real. Both cite anecdotes. Nobody can settle it, because no feedback layer exists to establish whether acting on a signal produced anything.
That absence is expensive twice over. It wastes the current spend, and it guarantees the disagreement recurs — because the only way to end it is evidence about causation, and nothing in the system is producing any.
What to change first
Not the vendor. Take one signal type, and for one month deliver it as a sentence rather than a score: what happened, when, why it matters, who owns it, what is expected. Then measure whether accounts touched that way move differently from accounts touched on the normal cadence.
That experiment costs almost nothing and settles the question the dashboard has been unable to settle for a year. Building the layer properly — detection through to feedback — is layers three, four and six of the GTM Architecture Audit.
Is the intent data itself wrong?
Sometimes, and it is worth checking — but it is rarely the binding constraint. Most teams that switch vendors get the same adoption problem with cleaner inputs, because the failure is downstream of the data. A signal that arrives without a reason, an owner and a next action is unusable regardless of how accurate it is.
Would routing it to the right rep fix it?
Routing helps and is not sufficient. A rep who receives a correctly routed account with a score and no explanation still has to invent a reason to reach out. What changes behaviour is arriving with the specific observation — what happened, when, and what it implies — because that is what makes the outreach writable.
Our reps say the accounts are not in-market. Are they right?
Often, partly. Most intent products infer interest from research behaviour that also occurs for reasons unrelated to buying. But reps also over-report this, because an account that produced an awkward call is remembered as a bad account. Without a feedback layer you cannot settle the disagreement, which is why it recurs quarterly.
What is the smallest useful fix?
Stop delivering scores and start delivering observations. One sentence stating what changed and when, attached to the account, with a named owner and a stated expectation of what happens next. That is a decision-layer change, not a data purchase, and it usually costs nothing but the argument about who owns the follow-up.