To build an ideal customer profile from HubSpot data, export your closed-won deals and the properties of their companies, join the two, check how many rows have each field filled in (the audit this post is about), count how your wins spread across industry, company size, country, deal value and lead source, and keep the fields where the wins pile up in a few values. The profile is whatever those counts support. You need at least 10 closed-won deals to get anything useful, and 30 or more for a sharper picture; below about 50, treat it as directional.
This post covers the audit that decides how far the rest goes: checking which HubSpot fields are filled in on your wins, with a worked example on an invented dataset. For the full method of weighting fields and handling small samples, read how to analyse your closed-won deals; for why the profile matters, what is an ideal customer profile.
Step 1: Export the closed-won deals
In HubSpot, open Deals, filter the deal stage to Closed won, and set the close date to the last 12 to 24 months. Before you export, edit the columns so the table includes the deal amount, create date, close date, lead source and the Associated Company column. Then export as CSV.
Industry, number of employees and country are company properties, so they are not Deals columns. Export your Companies as well, with those three columns, and join the two sheets in your spreadsheet on the company with a lookup. This works on any HubSpot plan. Those company columns decide how much of a profile you can build.
Step 2: Check what is filled in before you count anything
A count of industry is meaningless if half your wins have no industry.
Open the CSV and, for each column, count the rows that are not blank. Write the result in a small table.
Here is one for an invented company, Larkfield Systems, which sells compliance software. The numbers are made up for this post. Larkfield has 60 closed-won deals in its export.
| Field | Rows filled in (of 60) | Usable? |
|---|---|---|
| Deal amount | 60 | Yes |
| Close date | 60 | Yes |
| Company country | 58 | Yes |
| Company industry | 47 | Yes, with a note |
| Employee count | 41 | Thin |
| Lead source | 19 | No |
Larkfield can count industry and country with some confidence. Employee count is usable but only for two thirds of the deals, so the profile will say that. Lead source is filled in on 19 of 60 wins. Counting it would describe the habits of whoever fills in lead source, so Larkfield leaves it out and puts "fix lead source capture" on the RevOps list.
When you upload an export to Telepath's free report, the column check shows which columns it found and which it removed.
Step 3: Count the values
With the audit done, count the fields that passed. Put the field in the rows of a pivot table and the count of deals in the values.
Larkfield's industry pivot, again invented:
| Industry | Won deals |
|---|---|
| Financial services | 21 |
| Healthcare providers | 14 |
| Legal services | 7 |
| Manufacturing | 3 |
| Other (two values, one deal each) | 2 |
| Blank | 13 |
Financial services and healthcare account for 35 of the 47 deals with an industry. Put that in plain words: three quarters of the wins with a known industry sit in two industries. That is a concentration worth writing down. The 13 blanks go in the notes, because they could change the picture.
Repeat for country, employee band and deal value. For deal value, group into bands before you pivot. Counting exact amounts gives you a long list of single deals.
Step 4: Decide which fields earn a place in the profile
A field earns a place when your wins pile up in a few of its values. A field where wins are spread evenly across every value tells you nothing about who buys.
For Larkfield: industry earns a place (two values hold three quarters of the data). Country does too, since 52 of 58 are in the UK, although that mostly reflects where Larkfield sells. Employee count is thin and goes in as provisional. Deal value splits into two bands that do not overlap, which suggests two kinds of customer, so the profile has two entries instead of an average that describes neither.
Write the profile as a short list of fields with the counts attached: "Financial services or healthcare providers (35 of 47 with an industry), UK (52 of 58), deal value either £8K to £14K or £40K to £60K." That is a profile you can defend.
What this method cannot tell you
Counting wins shows where your wins sit. It does not show whether you win more often there than elsewhere, because that needs your losses as well. If most of your pipeline is in financial services, most of your wins will be too. Only the closed-lost deals tell you whether the rate is better or worse. If you want that comparison, keep your closed-lost deals in a separate file and run the same counts on it yourself. Do not add them to a Telepath upload: upload won rows only.
Telepath's report works from closed-won deals only, so it describes resemblance to your wins and does not compute win rates or predict a result. A low resemblance means a deal shares little with the deals you have won, and says nothing about whether it will close.
Doing it without the spreadsheet
Everything above is arithmetic, which is why it can be automated. The free ICP report runs the same checks on your closed-won export: it detects the columns, strips personal details before analysis, weights each attribute by how concentrated your wins are on it, and gives up to three target profiles plus a qualification checklist. When no attribute stands out, there is no checklist. When no group of wins stands out, there is no profile. Each is its own result. ICP intelligence is the paid version, which keeps the profile current and scores your open deals against it.
Frequently asked questions
How do I build an ideal customer profile from HubSpot data?
Export closed-won deals and their companies, join the two on company, check which fields are filled in, count how wins spread across each field, and keep the fields where wins concentrate in a few values. Write the profile as those values with their counts.
Which HubSpot fields do I need?
Industry, number of employees, country, deal amount, close date and lead source cover most of it. Add any custom property you trust, such as segment or product line.
How many closed-won deals do I need?
The free ICP report asks for at least 10 closed-won deals. It gives the sharpest results between 30 and 500 deals, and below roughly 50 the output should be read as directional.
What if my HubSpot data is incomplete?
Report the gaps and work with what you have. A field filled in on fewer than half your wins should not drive the profile. Fix the capture first, and recount in a quarter.
Can HubSpot build an ICP for me?
HubSpot provides the data, the export and the filters. The counting and the judgement are yours, or a tool's.
Do I need closed-lost deals as well?
For win rates, yes: keep them in a separate file and compare by hand. For a profile of what your wins have in common, no. Telepath's free report uses closed-won deals only, so upload won rows only.
Does the free report write anything to my HubSpot account?
No. The free report never writes to your CRM. You can upload a CSV, which connects nothing.
Run it on your own export
Once your audit shows which fields are solid, upload the export to the free ICP report and compare its profile with your pivot counts. Where they differ, check the fill rates first. ICP intelligence is the page for keeping that profile current.