Define your ideal customer profile

Your ideal customer profile is the filter behind everything Embers ranks. It belongs to the workspace, so every watched profile and keyword in that workspace is matched against the same buyer definition. During onboarding Embers drafts it from your audience and business answers and shows it for you to confirm or edit. It decides which of the people engaging with your content become leads, and it moves the score of the ones who do.

What you give it

Embers accepts either of two inputs, and at least one is required:

• A company website. Embers researches the public site to work out what you sell, who you sell it to, your pricing signals, your use cases, your category, and your stage.
• A description of your ideal buyer. Free text, up to 4000 characters. Job titles, industries, company stage, and pain points all help.

Giving both is better than either alone. The website supplies product context and the description supplies the buyer you actually want, which is not always the buyer the website implies.

What Embers builds from it

The answer is not stored as your paragraph. Embers turns it into a structured profile, and that structure is what every later decision reads:

If Embers reads you as a service business, it builds an aggressive peer exclusion into that structure so your own competitors do not arrive as leads.

How it affects matching and scoring

Every person an agent finds is classified against the structured ICP, in small batches, with the way they engaged supplied as extra context. The classifier weighs seven dimensions:

• Title match carries the most weight, followed by industry, then department and buying signals, then company, then location.
• Negative titles and negative industries have veto power. A hit there overrides every positive.
• Peers and competitors are vetoed too. A matching seniority title does not rescue them.
• Missing data is treated as neutral. It never counts against someone.

After the classification, deterministic guardrails run over the result. They can only turn a match into a non match, never the other way round, and they catch things like a disqualifying title carried by a current side role.

The verdict then feeds the score. ICP fit is a multiplier on the engagement score rather than a separate number added to it, and a confidence factor discounts uncertain verdicts. That is why two people with identical engagement can rank far apart. How scoring works has the full breakdown.

Correcting it later

You will get the first version slightly wrong. That is expected, and there are two ways to fix it.

1. Edit the fields directly

   Open the ICP screen (/icp), which is headed “Ideal Customer Profile”. Every list is editable in place: job titles, industries, departments, locations, company size ranges, exclusions, and buying signals. You can also regenerate the whole profile from an updated website or description, and preview the result before you commit to it.

   Saving keeps your existing leads and engagement history. Embers re-evaluates every lead against the new criteria in the background, which can take a few minutes, so match labels may change while it runs.

2. Give feedback on individual leads

   Open a lead from the leads table and use the thumbs up or thumbs down in the detail sheet, with an optional reason. Feedback is an override: that person keeps your verdict regardless of what the classifier says next time.

   Once enough new feedback accumulates, Embers refines the ICP itself, adjusting titles, departments, exclusions, and buying signals to match the calls you have actually been making.

Related