Who it is for, and how long it takes
| Who it is for | How long | What you need | What you end with |
|---|
| Founders and sales leads who want an ICP built from deals they have already won and lost. | About 30 days: two weeks of reading, then two weeks of testing. | Sales history in email, chat, calendar, meeting notes and the CRM, connected read-only, and one person to own the month. | A two-line ICP stored in a shared doc, tested on live requests, with a note of who does not fit. |
Key takeaways
- Use the deals you already have. The pattern is in your sales history.
- Write the ICP in one or two lines and store it where GeniOS reads it, so it becomes company context.
- Answer Not relevant whenever a fit call is wrong. That is how the ICP gets tested.
An ICP written in a workshop tends to describe the customer you hope for. One read from your own deals describes the customer who actually buys. You can run this playbook with a spreadsheet and some patience. GeniOS shortens the reading, because the pattern shows up in the intelligence your sales lead already gets each morning.
Worked example. The examples use Vaultex, a safety-footwear manufacturer in Greater Noida and GeniOS's first design partner. The buyers, orders and dates are for illustration only.
PrepBefore you start
Gather these before the month starts.
- A list of the last twelve months of deals, marked won, stalled or lost, even if it is rough.
- Read-only access to the sales team's email, chat, calendar, meeting notes and CRM.
- One owner for the month, usually the founder or the sales lead.
- The current revenue goal in one line, to check against the goals GeniOS forms.
- A shared doc where the ICP will live, with a version number and date at the top.
- Three questions to ask of every deal: who asked, what they needed first, and what made it move or stall.
- Time in the calendar: 20 minutes a day for two weeks, then a review on day 30.
StepsThe steps
30 days · 7 steps
- Connect the sources that hold your sales history
Start with email, chat, calendar, the CRM and meeting notes, connected read-only so nothing in them changes. Include older threads, because lost deals teach as much as won ones. Without GeniOS, export the last twelve months of deals into one sheet.
- Check the revenue goal
GeniOS forms the company's goals from its own emails, decks and meetings. Read the revenue goal it found with your sales lead. If it is wrong or out of date, fix it before you go further, because every fit call depends on it.
- Read the pipeline for two weeks
Each intelligence names a situation in a deal, why it matters and who owns it. Note which buyers keep showing up in orders that move, and which keep stalling. At Vaultex, Harsh might see that buyers who ask for IS 15298 test reports early move faster than buyers who open with payment terms.
- Write the pattern in one or two lines
For example: "Footwear brands and industrial buyers ordering IS 15298 tested safety shoes in bulk, with a named sourcing lead." A pattern you cannot say in a sentence is too loose to use. Write down who does not fit, too.
- Put the ICP where your team and GeniOS read it
Save it in a shared doc or sheet that GeniOS already reads. It becomes part of the company context next to your goals and every team's pending work. New hires can find it on their first day.
- Let new requests test it
When Bata asks for 12,000 pairs of S3 steel-toe shoes, the intelligence Harsh receives can now say whether Bata fits and what matters for the next step. That might be the 60-day payment terms Mayank has to clear. Answer Not relevant when it gets the fit wrong.
- Review after 30 days
Compare the orders that closed on proof with the intelligence you marked Not relevant. Tighten the ICP where the two disagree, and save the new version in the same place with the date.
TemplateCopy this ICP format
Keep the ICP to this shape, so anyone in sales can repeat it.
Our ideal customer, version [n], [date], owner [name]
Who: [type of company] buying [product] in [volume], with [a named role who owns the purchase].
Signal that they are serious: [what they ask for early, for example test reports or a sample].
Who does not fit: [type of buyer], because [the reason, taken from lost or stalled deals].
What we do first for a buyer who fits: [the first step, for example a quote within 24 hours].
Changed since the last version: [one line].
ExampleWorked example: Vaultex reads its own pipeline
How Harsh might write and test the first version.
Week one. Harsh reads the pipeline intelligence each morning and notes which buyers move. Brands and industrial buyers who ask for IS 15298 test reports in the first conversation tend to reach a quote and a sample quickly. Buyers who open with payment terms tend to stall after the quote.
Week two. He looks at the lost deals. Several were small one-off orders from traders who wanted the lowest price and no test paperwork. They go under "who does not fit".
The first version. "Footwear brands and industrial buyers ordering IS 15298 tested safety shoes in bulk, with a named sourcing lead. Serious buyers ask for test reports or a sample in the first conversation." He saves it in the shared sales doc that GeniOS reads.
The test. Bata then asks for 12,000 pairs of S3 steel-toe shoes, about $144,000. The intelligence Harsh receives says Bata fits, the sourcing lead has already asked for test reports, and the open item is Bata's request for 60-day payment terms, which Mayank owns. A trader's request the same week arrives marked as a poor fit. Harsh sends a polite reply drafted by a Reply Drafter, after one approval.
Day 30. Two fit calls were marked Not relevant. Both were distributors, so version two adds a line on how to treat them.
MistakesCommon mistakes
Five ways an ICP ends up describing nobody in particular.
- Writing the ICP in a workshop. It describes the customer you hope for. Read the deals first.
- Studying only won deals. Lost and stalled deals show who does not fit, and that saves the most time.
- Making it too broad. "Companies that need safety shoes" fits everyone and helps nobody. Add the volume and the signal.
- Keeping it in one person's notes. Put it where the team and GeniOS read it, or it will not shape daily work.
- Changing it every week. Hold a version for a month before you edit it.
What good looks like after
- Your ICP fits in two lines, and everyone in sales can repeat it.
- New requests arrive as intelligence that says whether the buyer fits, and why.
- Fewer items get marked Not relevant from one week to the next, because the ICP and the context agree.
- The sales lead spends time on buyers who fit. The rest get a quick, polite reply through a Reply Drafter and one approval.
- The ICP has a short history, one line per review, in the same doc.
If the ICP changes every week, the sample is too small. Hold it steady for another month before you tighten it again.
How to know it worked
Check these at the 30-day review. A sharp ICP also gives you two lines worth repeating in your monthly investor update, and it tells you which requests deserve the 24-hour quote. AI for founders and What is company intelligence? explain why context like this changes what reaches you, and the Vaultex case study shows the company behind the example.
| What to measure | How to check it | A good sign |
|---|
| The team can repeat the ICP | Ask three people in sales to say it from memory | All three give roughly the same two lines |
| Time spent on buyers who fit | Look at last month's quotes and meetings | Most of the team's time goes to buyers who fit |
| Fit calls you disagree with | Count items marked Not relevant for a wrong fit call | Falls from week to week |
| Quote-to-order rate | Compare buyers who fit with buyers who do not | Higher for buyers who fit |
| Version history | Open the shared doc | One dated line per review |