How to research a technical account before you call

You have 200 accounts and one week. By Wednesday you have forty tabs open: a careers page, a LinkedIn profile, an engineering post you can't find again. You know a lot about three companies and you have called none of them. Or you skipped the research and sent the same email as last week with a new company name.

Both end the same way. You had no quick way to tell which five accounts deserve your week, and nothing written down when you picked them. This article gives you a method for both problems. It takes four questions, answered in order. Each answer gets a source and a date. You stop when all four fit on one page.

The four questions

1. Is this account worth my week?
2. Who should I call?
3. Why would they care now?
4. What will I say?

The order matters. Fit is the cheapest question to answer, and a "no" saves you the other three. People come second because a reason to call means little without someone to call. The reason to call comes third because it only makes sense once you know who will hear it. What you say comes last, and by then it mostly writes itself.

Step 1: Check fit before anything else

Fit means two things. The people at the company have the problem your product solves, and they could plausibly use what you sell. For a company that builds software, you can read most of this from public pages in about ten minutes.

Then write fit as one sentence with its evidence. Here is an example about an invented company, Northwind Metrics:

If you can't find two separate signs of fit, move on. A fit claim with one source is a hunch.

This step has a limit you should accept. A company with no engineering blog, no public code and no job posts hides what you need. That is a reason to rank it lower this week, not a reason to guess.

Step 2: Find the people

Start with titles that own the problem. They change with your product. For an observability tool, look at platform and reliability leads and the director above them. For a security product, look at the head of security and the application security lead. For data infrastructure, look at whoever runs the data platform.

Try to pick two kinds of people. One feels the problem every day, like an engineer on the on-call rotation. One can act on it, like the leader of that team. The first tells you what is really wrong. The second can do something about it.

Then confirm each person still works there. A name from a contact list is a claim someone made on some past date. People change jobs. If you write to someone who left four months ago, you have also told the company you did not look. Three quick checks:

• Does the company's own site list them, or did they appear on its blog or in a talk recently?
• Is there recent public work under their name, such as a commit, a conference session or a post? Note the date.
• Does a current job post name them as the hiring manager or the team lead?

One current source beats three old ones. Write down the date next to the name.

Step 3: Find the reason to call now

A fit and a name tell you whether to call. They don't tell you why this week. For that you need something that changed:

• A new leader arrived, especially one who used your kind of product before.
• The company raised money or launched a product.
• Job posts show a migration, or a first hire for the role that owns the problem.
• The status page shows a run of incidents.

Not every change is a reason, and a few point away from calling. The next article covers which signals hold up and which mislead.

Two rules apply here. A reason needs a date. And it should fail a simple test: if the same sentence would work for any other company, it isn't a reason. "Congratulations on your growth" fails. "You posted your first staff reliability role on October 6" passes.

Step 4: Write the one-page brief

Put the answers on a single page. If it needs two, you are collecting facts instead of deciding. Here is a template, filled in for the same invented company:

The same page is your meeting preparation. Before a call, read it again and add one more line: what you still don't know. Missing information belongs on the page. "Not found yet: who owns the budget" tells you what to ask in the meeting. A page with no gaps usually means the gaps were hidden.

One habit that makes this work

Write down where each fact came from, and its date, as you find it. Don't plan to do it afterward. Three weeks from now you won't remember whether "they use Datadog" came from a job post or from your own assumption.

Mark your guesses as guesses. Use a plain word, like "Guess:", at the start of the line. Guesses are fine. You need them to make progress. Unmarked guesses are what turn into confident wrong statements on a call.

Where Repclose fits

Repclose does steps one through four for a whole territory. You upload your accounts and a sales profile that describes who you sell to. It ranks the accounts against that profile. For the top-ranked accounts (the top 10 by default) it lists about 10 people and gives the top 5 a profile check and a short card on why they might care. Plans include sections such as Top meeting plays, Why anything, Why now, Green flags and Supporting sources.

Each point links to the source it came from, with the publication date where one is available. When a source has no date, the plan says "undated". When Repclose could not find something, the section says "Not found yet" instead of filling the gap. Each researched point also carries a high, medium or low confidence word. That word comes from how many dated sources back the point, how recent they are, and whether the point is stated or inferred. It does not measure whether the point is correct. Open the links before you rely on a point.

Repclose gives you the context and the plays. You decide who to call and what to say. One run takes up to 1,000 accounts, and your first 5 plans are free.

You can see what a finished brief looks like in the sample brief, or run your own accounts.