Build the profile from evidence you can observe
A merchant selling subscription food has almost nothing in common, operationally, with one selling custom furniture. A two-person team buying a simple utility app goes through a different process from a larger operation replacing a system that holds years of customer data.
If you already have customers, start there. The pattern sits across five things at once: who activates quickly, who pays, who stays, who gets a result and who needs a level of support you can afford.
Write down characteristics you can observe, not ones you can only feel. "Ambitious brands" cannot be verified by anyone, including you. "Subscription merchants running a particular app and processing enough orders to justify our price" can be checked before you send a single email. Fill in one line for each field below.
| Profile field |
|---|
| Merchant type or vertical |
| Problem being solved |
| Operational size or volume |
| Existing app or workaround |
| Required technical fit |
| Likely buyer and user |
| Expected monthly revenue to you |
| Evidence that this merchant needs the product |
| Reasons to exclude an account |
Use a fit gate before a priority score
Scoring has a specific failure mode. Strong marks on three dimensions quietly cancel out a fatal weakness on the fourth, and a bad account arrives at the top of your list looking respectable. Ask the binary question first: can the product serve this account at all?
Three things fail an account outright: a required capability you do not have, expected revenue that cannot cover what serving them costs, and requirements that pull against where the product is going.
Only once an account clears that gate does ranking make sense. Three fields scored 0 to 2 are enough: fit, relevant evidence and access to the buyer. The totals order your list, and the order is a working judgment you revise the moment results disagree with it.
Make the account list usable
Deduplicate companies running several stores before you count anything, so the account total reflects buying entities rather than storefronts.
You research 40 accounts. Fifteen fail the fit gate and come out, leaving 25 relevant merchants. That is a starting batch, not a benchmark.
Separate buyer fit from store size. Pick the operating variable your app depends on: return volume, booking complexity, catalog size or team workload. GMV is a proxy for all of them and a direct measure of none, so it earns its place only where it approximates the variable you care about. Where the real variable is unavailable, label the proxy as a proxy. A large merchant without the problem is a worse prospect than a small one feeling an urgent version of it, and size alone should never trigger a high-touch sales process.
Learn from the problem and proof interview
Early conversations are for improving the profile, not for closing. Ask how the workflow runs today, what goes wrong in it, how often, who deals with it and what fixing it would be worth.
Listen for episodes. "Reporting is a pain" gives you nothing to act on. "Every Monday our operations manager combines three files before we can make a decision" tells you who feels the problem, how often and where a product could slot in.
Then ask what they have already tried: a competing app, a spreadsheet, a contractor, a manual process everyone tolerates. Each shows what the problem is already worth to them. An account that has tried nothing is ambiguous rather than promising, because the problem may be new, unrecognized or simply unimportant.
Record the words they use. When several buyers describe the same outcome in similar language, that language belongs in your listing and outreach. Remember that what people say they want and what paying customers adopt are two different data sets.
Compare segments on the same research effort
Two segments can look similar until you count what each one costs to serve.
You test two groups of 20 accounts. Both pay $200 MRR per customer.
| Outcome per 20 accounts researched | Segment A | Segment B |
|---|---|---|
| Demos | 4 | 6 |
| Payers | 2 | 1 |
| New MRR | $400 | $200 |
| Setup hours | 6 | 12 |
Group A needs three hours of setup per payer. Group B needs 12 hours for its single payer.
Segment B generates more meetings and less revenue, and costs four times the delivery work per paying customer. That is enough reason to find out why the buying and delivery process differs so sharply.
Start narrow without getting trapped
Start where your advantage is easiest to explain and verify. Once customers begin repeating, test an adjacent segment with a label and a budget of its own. A workflow proven with subscription food merchants may transfer to another subscription category, and until you have run it, that is a hypothesis rather than an expansion plan.
Record four things each time: the assumption, the sample, the result and the decision. Keep the proven segment visible while you test the next one, because the most common way to lose a working niche is to become interested in a new one.
Map the buying group and keep a rejection log
The person who replies is not necessarily the person who will use the app, and neither is necessarily the person who approves it. At a small merchant all three are the same founder. At a larger one, an operator evaluates, finance approves and an agency implements. Ask who needs to be comfortable with this change, which works better than demanding "the decision maker." Record whoever owns the outcome, whoever operates it daily and whoever can authorize the spend.
Use a fixed set of rejection reasons so the log can be counted: technical mismatch, economics, no relevant need, missing capability, service burden, timing. Review repeats monthly, because each cause points somewhere different. Missing capabilities are a product decision. Inaccurate inputs are a research problem. Unsuitable stores mean your filters need work.
Do this week
- Write the merchant profile and a list of exclusion reasons using the nine fields.
- Fit-check 25 accounts, explain each selection and flag repeated custom exceptions.
- Deduplicate multi-store companies so you count buying entities, not storefronts.
- Track accounts researched, eligible accounts, disqualification reasons, then paid conversion and retention by segment.
Questions
How do I define an ideal customer profile for a Shopify app?
Start from customers you already have: who activates quickly, pays, stays, gets a result and needs support you can afford. Then write down characteristics you can observe before you send any message, such as vertical, order volume, existing app and required technical fit.
Is GMV a good way to qualify Shopify merchants?
Only as a proxy. Pick the operating variable your app depends on, such as return volume, booking complexity, catalog size or team workload. GMV approximates those but measures none directly, so label it as a proxy and do not let size alone trigger a high-touch sales process.
Should I score accounts or disqualify them first?
Disqualify first. A fit gate asks whether the product can serve the account at all. Scoring before the gate lets strong marks on three dimensions hide a fatal weakness on the fourth. Rank only the accounts that clear the gate.
Related guides
Chapter 9: How to find Shopify merchants with a reason to listenChapter 3: Self-serve, assisted or hybrid: how should Shopify merchants buy your app?Numbers marked as planning examples are arithmetic illustrations, not Shopify benchmarks or promised results. Founder-reported figures are not independently audited. Platform rules change: confirm current requirements with Shopify before you act.