500 free credits, no card. Try Smalt AI →

Sales

How do I build a list of 50 companies that fit my ideal customer profile

Turn your best customers into 4 to 6 filters, search wider than you need, score each company against them, and cut the list down to the best 50.

Smalt AI team ·

Write your ideal customer profile (ICP), the companies most likely to buy and stay, as four to six filters taken from your best customers. Search a database for matches, score each against the filters, and keep the top 50. Fifty is a working size, not a researched figure.

How to do it

1. Turn your best customers into filters

Start from the customers you most want more of, not from a persona. HubSpot’s ideal customer profile template suggests defining the profile from CRM data on your best customers (highest revenue, lowest churn, strongest adoption), then interviewing 10 or more of them. You are looking for patterns in industry, financial readiness, active need, growth trajectory, company size and geography.

Write the result as a short list of testable conditions. “Mid-sized logistics companies” is too loose. “Logistics, 200 to 1,000 employees, based in the UK or Ireland, hiring for operations roles” can be checked against a record.

Pattern-finding needs enough closed deals to mean something. The Starr Conspiracy’s ICP benchmarks put the minimum for closed-won pattern analysis at about 30 deals, and 100 or more for a mature dataset. That is practitioner guidance from a consultancy, not a study, but it is a fair warning. With fewer than 30 wins, treat your filters as a hypothesis and revisit them as deals close.

2. Search wider than your target

Run the filters in a company database and expect hundreds of results. LinkedIn’s Sales Navigator guide to prospecting describes account filters for industry, headcount, annual revenue, department headcount and growth, and job opportunities, and suggests using them to find companies similar to your customers. It also advises against over-filtering at this stage. Follow that: if a filter is a nice-to-have, leave it out of the search and use it in step 3.

Save the search so new matches show up as alerts, and save the accounts you want to a named list.

3. Score, then cut

Give each company a point value for every filter it meets, and weight the two or three that matter most. A simple version is a sheet with one row per company, one column per filter, and a total. The table below is illustrative only: the companies are fictional and the weights are an example, not a recommendation.

FilterWeightCompany ACompany BCompany C
Industry is logistics3PassPassFail
200 to 1,000 employees2PassFailPass
Based in UK or Ireland2PassPassPass
Hiring for operations roles2PassPassFail
Uses HubSpot1FailPassPass
Total (out of 10)985

Sort by total, then read the top 70 or so by hand and drop any that fail a check the database cannot make, such as a recent acquisition or a product you know competes with yours. Company C would fall below the line here, while A and B stay in for a closer look.

The final cut is where your own judgement comes in. Fifty is enough to spread across a few weeks of outreach and small enough to research each account properly. If you are working a larger territory, take that many per segment instead. Sales Navigator’s guidance is to segment lists by industry, growth stage or revenue band, which also lets you tailor messaging to each group.

4. Mark the list in your CRM

In HubSpot, you mark a company as a target by setting its “Target account” property to true, and rank fit with the “Ideal customer profile tier” property, where Tier 1 is a great fit. HubSpot’s account-based marketing setup guide also describes a workflow template, “Update company properties based on defined criteria”, that assigns tiers from revenue, industry and country. Use it once your filters have stabilised, so you are not automating a guess.

Review the list on a schedule. HubSpot’s template recommends refining the profile quarterly, and a quarterly pass is also a natural point to swap out accounts that went quiet.

Doing it with an AI coworker

If you want to hand the search and the sheet to Smalt, give it your filters in plain words: “Find companies in logistics with 200 to 1,000 employees in the UK and Ireland, score them against these five criteria, and give me the top 50 in a Google Sheet.” It searches and enriches the companies itself and puts the scored list in Google Sheets. The AI coworker for sales teams page covers the wider set of prospecting tasks, and the research use cases page shows how account research fits in.

You can then ask it to add the companies you approve to HubSpot, which is the only CRM it connects to. Company records written to HubSpot are not held for approval the way emails and Slack messages are, so say which rows to add rather than asking for the whole sheet. Smalt does not choose your filters or your weights; those stay yours.

Once the list exists, the next step is finding the right person at each account. Our walkthrough on finding the head of operations at target companies picks up from here.

What to check before you rely on it

Database coverage may be uneven. Headcount, revenue and industry fields in any provider are estimates. In my judgement, small companies and those outside the US are the likeliest to have gaps, though this varies by provider. Spot-check 10 companies on their own websites before trusting the filters.

Your profile may describe your past, not your future. Filters built from existing customers reproduce the customers you already have. If you are moving upmarket or into a new region, add a few deliberately different accounts and track how they perform.

A match is not a buyer. A company that fits on paper may have no budget or owner for the problem. Hiring signals and recent news help, but a person still has to decide whether the account is worth a call. If your sales cycle is long and deals are large, fewer accounts with deeper research may serve you better than a bigger list.

Check what you write to the CRM. Before importing or bulk-setting target accounts, confirm there are no duplicates of companies already in HubSpot, and that existing owners and lifecycle stages are not overwritten.

Smalt is an AI coworker that takes jobs like this end to end and hands back the finished file.