GTM Club Join

GTMap · Plays

Intent-based outreach

Contacting an account because data says it is researching your category right now, and writing the message around that timing.

Intent-based outreach is signal-based prospecting where the signal is intent: evidence, usually from a third-party provider, that people at an account are reading about your category this week. The account was already in your ICP; intent moves it to the top of the list, and shapes what you say.

It is the play most teams buy a tool for and then run badly, because the tool is the easy part.

Why it matters now

Most of an ICP is not buying at any given moment. The old model worked the whole list anyway, on the theory that enough volume would find the few who were. That model is dying: inboxes are defended, LinkedIn is rate-limited, and the reply rate on untargeted volume has fallen every year.

Intent flips the order. Instead of asking a thousand accounts whether they are interested, you ask the fifty that just showed they are. Volume drops by an order of magnitude; reply rate and meeting rate climb; the domain stays healthy because you send less.

The second reason is AI. Intent used to be expensive to act on: someone had to read the signal, research the account and write something relevant. A model now does the research and the first draft in seconds, so acting on every intent spike the same day is finally cheaper than working the static list.

How it is actually run

  1. A tight ICP first. Intent on a wide list produces a wide list of spikes, most of them at companies you would never sell to. Narrow the ICP until intent has something to sort.
  2. One or two intent sources, not five. Third-party intent (topic-level research data from a provider), first-party intent (visits to your pricing page, opens across a sequence, product usage), or review-site intent (someone comparing you to a competitor). Start with the one that maps to your category best, and test it before paying for another.
  3. A threshold you trust. Every provider produces a score; the score is noise until you know what it means for you. Run one month, tag every account that spiked, and see which ones took a meeting. Then set the line.
  4. A sequence built for the signal. The message does not say "I saw you were researching X". It opens with the problem people research X to solve, in the reader's words, and offers something a person in that research would want: a comparison, a teardown, a number from a similar company. The intent is the timing, never the opener.
  5. Speed. An intent spike is a window of days. The account is talking to vendors this week; being the vendor who showed up with something useful on Tuesday is the whole play. Automate the flow from spike to draft, keep a human on the send.
  6. Close the loop. Every intent-sourced meeting gets tagged as such in the CRM, so the play has a win rate and a cost per meeting after a quarter. Otherwise the intent tool gets renewed on faith.

The one-signal test in the library is exactly step 3 for any signal, intent included: seven questions, one signal, scored against your audience before you build a campaign on it.

The numbers

What members report, and what lemlist campaign data shows for signal-scoped sends against the same copy on a static list:

  • Reply rate: 2 to 3 times the static list, on a fifth of the volume.
  • Positive reply rate: 4 to 8% on a well-thresholded intent list, against 1 to 2% on the wide list.
  • Meetings per hundred accounts: 8 to 15 when the message is built for the signal; 3 to 5 when the same accounts get the generic sequence.
  • Time from spike to first touch: the teams seeing the top of those ranges send within 48 hours.

The variance is in the threshold and the message, not the provider. Two teams with the same data and the same ICP can be a factor of three apart.

The mistakes

Buying intent before the ICP is narrow. The tool surfaces spikes everywhere; the team chases them; the list gets wider, not tighter.

Saying "I noticed you were researching". It is the one thing a prospect does not want to hear. The signal decides who and when; the message is about their problem.

Treating the score as truth. A spike means several people at the company read about the topic. It could be an intern writing a report. Threshold on outcomes, not on the provider's colour coding.

Acting slowly. An intent list worked over three weeks is a static list with a good excuse.

Ignoring first-party intent. The richest intent data a company has is its own: pricing page visits, three opens in a row, a trial signup from an ICP domain. It is free, and most teams never wire it into outbound.

Not tagging the source. Without it, nobody knows a year later whether the play worked, and the renewal argument becomes a feelings argument.

Related plays

Intent-based outreach is one member of a family. Trigger-based outreach fires on a public event at the account (a hire, a round, a job post); intent fires on research behaviour, which is quieter and earlier. Signal-based prospecting is the general case, and the discipline of running outbound only where a signal says to.

Most mature teams run all three on the same ICP: triggers for the loud moments, intent for the quiet ones, and a small static cadence for the strategic accounts that never show a signal but matter too much to wait for.

Go do it