Account profiling: a one-page B2B account research template

This page is for SDRs, account executives, account managers and founders who sell to other businesses and want to research a prospect before contacting it. It's about B2B sales, not user profiling in analytics, accounting or donor research. A full profile takes 20 to 30 minutes; a light one takes 5 to 10. Most of that time goes on checking sources, not writing.

What an account profile has to answer

A profile earns its place if it answers five questions. Anything else is decoration someone has to check.

The last row matters most: the unknowns are your first-call agenda. A profile with no unknowns usually means someone guessed.

The one-page account profile template

Copy this for accounts where a deal would matter. Leave a field blank rather than guessing.

ACCOUNT
Company (trading name):
Legal name and registry: [e.g. NC Secretary of State / Companies House no.], checked [date]
Website:
What they do (one line, in their words):
Size (their own published figure, or "unknown"):
Locations / markets:

FIT
Matches our ICP on:
Anything that rules them out:

WHAT'S CHANGED (last 12 months, each with source and date)
1.
2.
3.

PEOPLE (roles first; note where each name came from)
Owns the problem:
Signs off:
Others involved (IT, finance, procurement):

CURRENT APPROACH
What they use or do today, and how we know (or "unknown"):

UNKNOWNS (these become first-call questions)
1.
2.

HYPOTHESIS
Because [dated fact], [role] is probably dealing with [problem], which costs them [impact]. We'll test it by asking: [question]

FIRST MOVE
Channel, person, opening line, date:

Written by / date:

For accounts further down your list, a light profile is enough:

Company and website:
Registry match checked: [yes / no], [registry], [date]
Fits our ICP because:
One recent change (source and date):
One role to contact (and where you found it):
One question to open with:

If an AI assistant with web search drafts the profile, give it rules that make guessing harder:

I'm preparing for a first sales conversation. Research one company using public sources only.

Company: [name]
Website: [domain]
Location: [state and country]
What I sell: [one line]
Why I think they might need it: [one line]

Return:
1. Identity: the registered legal name, the registry it appears in, and any similarly named companies you found that are NOT this one.
2. What they do: up to 3 bullets, in the company's own words.
3. Changes in the last 12 months: up to 5 bullets, each with a date.
4. Roles likely to be involved in buying [my category]. Roles only, no names.
5. Possible fit: up to 2 bullets, each tied to a fact above.
6. Unknowns: what you could not confirm.

Rules:
- Put the source URL after every bullet. No source, no bullet.
- Don't estimate revenue, headcount, customers or tools. If the company publishes a figure, quote it and say where.
- If sources disagree, show both.
- Ignore results about any company with a different website domain.

Then open every link yourself. Delete any bullet whose source doesn't say what the bullet says.

Where to verify company and buyer facts in the US and UK

Start with registries and filings, because they tell you which legal entity you're dealing with. Then use what the company says about itself for what's changing.

Registries and filings

What the company says about itself (same in both countries)

Two habits make these tables work. Match the website domain, registry entry and address before trusting a fact. And apply a cut-off: changes from the last 12 months count as "what's changed"; anything older is background.

Worked example: a completed profile with sources and dates

Illustrative example. The companies, people, sources and facts are fictional, to show the method.

The seller: Trussline (trussline.example), project-management software for commercial contractors. The account: Calder Finch Construction (calderfinch.example), a commercial general contractor in Raleigh, North Carolina. Why research it: Trussline's best customers are commercial contractors with 100 to 500 staff that are taking on larger, document-heavy projects.

What 25 minutes of research turned up:

ACCOUNT
Company: Calder Finch Construction
Legal name and registry: Calder Finch Construction, LLC, NC Secretary of State, active, checked 20 Oct 2026
Website: calderfinch.example
What they do: "Commercial general contracting for healthcare, education and office projects across the Triangle"
Size: "more than 175 team members" (About page, checked 20 Oct 2026)
Locations: Raleigh HQ, projects across central North Carolina

FIT
Matches our ICP on: commercial GC, 100 to 500 staff, taking on larger projects
Rules them out: nothing found
Note: not Calder Construction Group (residential, SC). Don't merge.

WHAT'S CHANGED
1. Selected as GC for a four-story medical office building in Cary (news page, 8 Sep 2026)
2. Hiring 3 assistant PMs and a project controls lead; ad mentions "RFI and submittal logs in Excel" (careers page, checked 21 Oct 2026)
3. COO quoted wanting "fewer change-order disputes" (business journal interview, 12 Aug 2026)

PEOPLE
Owns the problem: VP Operations (leadership page, checked 20 Oct 2026)
Signs off: COO (leadership page; quoted in the interview)
Others involved: IT manager (integrations), the new project controls lead once hired

CURRENT APPROACH
Spreadsheets for RFI and submittal logs, per the job ad. Main PM platform: unknown.

UNKNOWNS
1. Which PM platform they use today, if any
2. How many projects run at once

HYPOTHESIS
Because Calder Finch has taken on a medical office project and is hiring APMs who will keep RFI logs in Excel, we think the VP Operations is losing sight of late RFIs across jobs, and that late RFIs feed the change-order disputes the COO wants to cut. We'll test it by asking: "When an RFI runs late on one job, how does anyone outside that job find out?"

FIRST MOVE
Email the VP Operations on Tuesday, opening with the Cary project and the APM hiring. Call Thursday. Save the interview quote for the call.

Written by: J. Ortiz, 21 Oct 2026

What makes this profile good:

• The registry check caught a name collision before it reached an email. An opener congratulating a commercial contractor on its custom homes would have ended the conversation.
• Every change is dated and inside the 12-month cut-off, and every source was opened.
• Headcount is the company's own figure, labelled as such. There is no revenue estimate, because none was published.
• People appear as roles, each with the source and the date it was checked.
• The two unknowns became the first two discovery questions, and the hypothesis can be proved wrong in one answer.

Turn the research into an outreach hypothesis

A hypothesis is your best guess at the prospect's problem, written so they can tell you you're wrong. Being corrected starts a real conversation.

Use this shape: Because [dated fact], [role] is probably dealing with [problem], which costs them [impact]. Test it by asking [question].

A few rules keep hypotheses honest:

• Pick facts that point at a problem you solve. A new office only matters if it creates the problem your product fixes.
• Two facts that point the same way beat one. The project win plus the spreadsheet job ad is more convincing than either alone.
• Address the role that owns the problem. The same fact means different things to a COO and an IT manager.
• Keep personal details out of the opener, even public ones. Lead with the company's news, not the person's life.

From here, the hypothesis becomes a message. For email, use one line on the change, one on what it usually causes and one question: see the cold email templates. For a call, open with the change and ask the test question: see the outbound call scripts.

How much research each account deserves

Not every account needs a full profile. Tier your list by fit and by how much a deal would matter:

Illustrative arithmetic: 30 tier-1 accounts at 25 minutes is 12.5 hours. Add 150 tier-2 accounts at 8 minutes (20 hours) and one person has spent most of a week researching before writing a message, which is where teams cut corners. Bebop researches each account it finds and writes a playbook (who to talk to, what's happening at the company, the angle to open with), so you start from a draft to check. Verify it with the checklist below.

Where account research goes wrong

• Identity errors and name collisions. Two companies with similar names get merged, a subsidiary gets confused with its parent, or a rebrand hides the legal name. Fix: match the domain, the registry entry and the address before you trust anything. People collide too: confirm a contact's employer and role on the company's own site.
• Stale news. A two-year-old expansion story becomes "they're expanding". Fix: date every change and enforce your cut-off.
• Invented specifics. Revenue, headcount, customer names and tech stacks are the details most often estimated, by people and AI tools alike. Fix: use the company's own figures or a filing, or write "unknown".
• Departed people. The leader you're about to congratulate left in the spring. Fix: confirm the person is still in the role, and note the date you checked.
• Volume without checking. GTM-Bench, a June 2026 benchmark published by Blackpearl Group (Bebop's parent company), ran AI agents through 72 prospecting tasks and scored every returned record +1 if strong, 0 if borderline and -1 if weak, after an audit of identity resolution, whether claims were supported and consistency with the source record. Several general-purpose AI agents finished with negative net scores, and the authors trace failures to identity errors, unsupported claims and poor decisions about when to stop. The same arithmetic applies to a profile: three checked facts are worth more than ten unchecked ones.
• Research that's too personal. Family details and personal social media don't help you sell, make openers feel invasive and raise data-protection questions (see below).

Researching named people: the data-protection basics

This is general information, not legal advice. Rules change and depend on where you and your prospects are, so check with counsel before you scale.

Facts about a company are not personal data. Details about a named person at that company are, even in their work role. The UK Information Commissioner's Office (ICO) puts it plainly: if you process personal data for direct marketing, "even in a business context, the UK GDPR applies".

UK and EU

• You need a lawful basis. For prospecting this is usually legitimate interests. The ICO says direct marketing may be a legitimate interest, but not automatically: you must pass its three-part test (a legitimate purpose, processing that's necessary for it, and a balance against the person's interests, including whether they'd reasonably expect it). The ICO recommends recording this as a legitimate interests assessment before you start. The European Commission also lists direct marketing as an example of a legitimate interest under the EU GDPR, provided people's rights and freedoms aren't seriously affected.
• Tell people where you got their details. If you collect someone's details from anywhere other than them (a website, a database, a colleague), UK GDPR Article 14 requires you to tell them within a reasonable period and at the latest within one month, or at your first communication if you use the details to contact them. The European Commission describes the same rule for the EU, including telling people the source of the data.
• Objections are final. The ICO says an individual business contact's right to object to direct marketing is absolute. Record it and stop.
• The ICO notes its business-to-business marketing guidance is under review after the Data (Use and Access) Act, so check the current version.

United States

• California covers B2B contacts. The California Consumer Privacy Act's exemption for business-to-business information expired on 31 December 2022, so since 2023 California residents' work details are covered like other personal information. The law applies to for-profit businesses doing business in California that meet any of three tests: gross annual revenue of $26.625 million or more (the inflation-adjusted figure from 1 January 2025), buying, selling or sharing the personal information of 100,000 or more California residents or households, or earning at least half their annual revenue from selling California residents' personal information. Covered businesses must honor requests to know, delete, correct and opt out of selling or sharing.
• Other states differ. Virginia's law, for example, excludes a person "acting in a commercial or employment context", so work contacts generally fall outside it. Check the states where your prospects live.

In practice: collect roles before names, add a name only when you're about to make contact, record where and when you found each person, keep personal numbers and social accounts out of your CRM, and check your do-not-contact list first. Emailing and calling have their own rules, covered on the pages linked above.

Verification checklist

Run this before any profile reaches your CRM or an opener:

• The legal name, registry entry and website domain all refer to the same company.
• Every fact has a source you opened, and the source says what the profile says.
• Every "what's changed" item is dated and inside your 12-month cut-off.
• Every figure is the company's own or from a filing, or it's marked "unknown".
• Each named person was confirmed in the role recently, with the source and date recorded.
• At least two unknowns are written as questions.
• The hypothesis could be wrong, and you know which answer would show it.
• If the prospect read the profile, nothing in it would be wrong or intrusive.

Next step: turn your hypothesis into a first message with the research-driven cold email templates, or plan the first call with the sales call planning template.