Turn Any Company URL Into a First-Call Account Brief

Write your reusable product-context block once

Before you research a single account, write a short, reusable description of what you sell and who you sell to. You will paste this into the top of every brief prompt so the AI produces a 'where we fit' section grounded in your actual product instead of generic observations. Without it, the model will tell you the company is 'growing fast' and 'investing in technology,' which helps no one.

Keep it to roughly 150 words: one line on what the product does, the three concrete problems it solves, your typical buyer titles, one or two proof points, and your deal size and motion. Be specific. 'We help teams be more efficient' is useless; 'we cut new-region ops planning from weeks to days for 200-2,000 person logistics and manufacturing firms' gives the model something to map against.

Save it somewhere you can paste in one keystroke: a text-expander snippet, a pinned note, or the system-instructions field of a saved Perplexity Space or a ChatGPT custom GPT. The whole point is that you never retype it.

OUR COMPANY CONTEXT:
We sell: {{ONE_LINE_DESCRIPTION_OF_PRODUCT}}
We help with: {{PROBLEM_1}}; {{PROBLEM_2}}; {{PROBLEM_3}}
Typical buyers (titles): {{TITLE_1}}, {{TITLE_2}}, {{TITLE_3}}
Who we are NOT a fit for: {{ANTI_PERSONA, e.g. companies under 50 people, pure services firms}}
Proof points: {{CUSTOMER_OR_METRIC_1}}; {{CUSTOMER_OR_METRIC_2}}
Deal size / motion: {{e.g. $30-80k ACV, 2-call cycle, champion-led}}

Gather current, cited facts in Perplexity

Open Perplexity at perplexity.ai. If you have Pro, click the model selector near the prompt box and pick a strong reasoning model; the default is fine for this. Paste the research prompt below with the company name and URL filled in. Perplexity returns its answer with numbered inline citations, and that citation trail is the entire reason you use it instead of asking Claude directly: it keeps the brief honest and clickable.

Copy the full answer including the source list. If you do not have Perplexity, use ChatGPT with web search toggled on, or Google's Gemini. The non-negotiable is that the facts come from the live web, not the model's training data, which can be a year or more stale on funding and headcount and will state last year's numbers with total confidence.

Read the citation dates as they come back. For the 'recent news' section, discard anything older than about 12 months. A 'recent' funding round from three years ago on a first call makes you sound like you skimmed an old article, which is worse than saying nothing.

Research {{COMPANY_NAME}} ({{COMPANY_URL}}). Give me the following, with a source and a date for each point:
1. What they do, in one sentence, and who their customers are.
2. Company size (employee count) and HQ location.
3. Most recent funding round, valuation, or public financial note, with the date.
4. News from the last 6 months only: launches, leadership changes, layoffs, expansions, partnerships, acquisitions.
5. Their stated strategic priorities, drawn from recent blog posts, exec interviews, earnings calls, or recurring themes on their careers/jobs page.
6. Names and titles of leaders relevant to {{DEPARTMENT_I_SELL_TO}}.
7. What roles they are hiring for right now (this signals where they are investing).
Keep every point factual and dated. If you cannot find something, say 'not found' rather than guessing. Do not speculate.

Generate the structured brief in Claude

Open Claude at claude.ai. In the message box, paste three things in order: your product-context block from step one, then the full Perplexity output (facts and citations), then the brief prompt below. Order matters because the prompt instructs Claude to use only the facts above it.

Claude turns the raw, messy research into a consistent brief with the same sections every single time, including the connect-the-dots reasoning a junior rep would miss: why a usage-based pricing launch changes who their AEs prioritize, or why twelve open AE roles implies an onboarding-ramp pain you can speak to. The structure is deliberately scannable; you are building something a rep reads in ninety seconds, not a research paper.

Notice the prompt forces a [verify] tag on anything the model is unsure about and an 'unknown' on anything missing. This is what makes the output safe to act on: instead of confidently smoothing over a gap, it tells you exactly where to look before you trust it on a call.

You are prepping a sales rep for a first call. Use ONLY the facts I pasted above. Do not invent anything. If a fact is missing, write 'unknown'. Where you are inferring rather than citing, mark it [inference].

Produce this brief for {{COMPANY_NAME}}:

## Snapshot
- What they do / who they serve
- Size, HQ, stage
- Most recent material event (with date)

## Likely priorities right now
- 3 bullets, each tied to a specific cited fact above

## Where we likely fit
- 2-3 bullets mapping their likely pains to OUR product context. If we are clearly not a fit, say so honestly.

## People to know
- Name, title, why they matter (only people actually named in the facts above)

## Three opener angles
- 3 specific, non-generic conversation starters, each referencing a real dated fact

## Smart discovery questions
- 4 questions tailored to THIS account, not generic

Flag any claim you are unsure about with [verify]. Keep the whole brief under 350 words.

Verify the facts you'll say out loud in thirty seconds

Skim the brief for anything marked [verify], [inference], or 'unknown.' For the facts that carry the call, the ones you would say out loud, click back into the original Perplexity citations: the funding round, the headcount, and above all any leader's name you plan to mention. Models invent plausible executive names, and getting a name wrong on a first call costs more credibility than the entire brief earns.

This is a thirty-second check, not a research project. You are not re-verifying everything; you are confirming the three or four facts that, if wrong, would embarrass you. Everything tagged [inference] is your own reasoning surfaced honestly, so treat it as a hypothesis to test on the call, not a fact to assert.

Save the brief where it is searchable and reusable

Paste the finished brief into the account or contact record in HubSpot or Salesforce as a note, or into a row in a Google Sheet keyed by company domain. The point is that the brief stops being a disposable scratchpad: it becomes searchable, your manager can see your prep, the next rep who touches the account inherits it, and you can diff it against a refreshed version later.

In a Google Sheet, a clean structure is one row per account with columns for company, domain, date generated, the brief text, and a 'next refresh' date. This doubles as a log of which accounts you have prepped and when, which is genuinely useful for territory reviews.

Refresh only the research step the morning of the call

The morning of the call, re-run only the Perplexity step to catch anything that broke in the news overnight: a layoff, a funding announcement, an exec departure. For the large majority of accounts nothing material will have changed, and you skip straight to the call with the brief you already have.

If something did change, regenerate the brief in Claude with the fresh facts. The cost of this refresh is two minutes; the upside is never walking into a call unaware that the company announced layoffs that morning, which would make your cheerful 'how is the growth going' opener land very badly.