How to write an ICP research brief your agency can actually use

Separate company fit from contact fit

“Find marketing directors in the US” leaves the researcher guessing about the company behind each person. A director at a multinational and a director at a ten-person consultancy may share a title while buying very different services.

HubSpot distinguishes the ideal customer profile, which describes the target company, from a buyer persona, which describes the individual inside it. Use that distinction to create two separate parts of your research brief. ¹ Start with your offer: what problem does your agency solve, and which businesses can reasonably use that work?

Write the rules someone must be able to check

Use observable criteria wherever possible. “Ambitious businesses” is difficult to research consistently. A named service category, operating location, and current website can be checked. If a criterion is an estimate, preserve that uncertainty instead of presenting it as a fact.

• Company: primary service or product, operating market, size range, and any required business characteristics.
• Contact: responsible function, preferred roles, acceptable alternatives, and maximum contacts per company.
• Exclusions: existing clients, active opportunities, competitors, unsuitable business models, and your suppression list.
• Evidence: source URLs, checked date, and a short reason the company fits.
• Delivery: required columns, file format, missing-value labels, and who approves the sample.

A worked example: an SEO agency targeting accounting firms

This is an illustrative brief, not client performance data. The agency sells local SEO to independent accounting practices and wants to test one market before expanding.

Approve the reasoning before increasing the volume

A small approval sample is our suggested workflow, not an industry benchmark. Review why each company was included. Separate account-fit errors from contact errors: a correct person at an unsuitable company still fails the brief.

Return specific corrections. “Too broad” is less useful than “exclude firms whose main service is personal tax returns.” Version the brief after feedback so the next batch follows the same decision rules. Do not quietly change the scope halfway through production.

Decide what happens when information is missing

Give the researcher permission to return “unknown.” Decide which missing fields hold a record back and which can be reviewed later. For example, an unconfirmed operating location might block inclusion, while an optional employee estimate might not.

Before delivery, check whether a second researcher could understand the inclusion decision from the exported record alone. If the explanation exists only in a chat message, the brief or delivery format needs another field.

Use the brief as your acceptance checklist

Download the blank CSV below and fill it in for one offer and one segment. Keep required criteria short enough to review, but specific enough to reject a bad match. Then compare a sample against every required rule.

Read the prospect-list quality checklist next to decide what evidence and handoff checks to request. If you need someone to carry out the research, share the completed brief with our prospect research team.