How to get your first 10 customers

Not a segment. A hundred rows in a spreadsheet, each one a person with a name, a place you found them, and a specific reason to believe they have this problem.

Where they come from, in order of how well they convert:

1. People who complained about the problem in public. A post, a review, a thread. They've already told you they have it, in their own words, which you can quote back.
2. People using the tool you're replacing. Findable from review sites, integration directories, community mentions.
3. People whose job description contains the problem. LinkedIn search, job boards, industry directories.
4. Your own network's second degree. Not your friends — their colleagues. Ask for one specific introduction rather than broadcasting.

The "specific reason" column is what makes this work. "Posted on 12 March about spending a day on invoice reconciliation" is a reason. "Freelance designer" is not — it produces a message that reads like a template, because it is one.

A hundred rows takes about a day. At realistic conversion rates that's roughly ten conversations and a handful of customers, which is exactly the goal.

Four lines. Under a hundred words. The test: if you could send this exact message to person number 47 as well, delete it and start again.

Line 1 — the specific thing you saw. "Saw your post about losing a day a quarter to invoice chasing." Line 2 — what you're doing about it, plainly. "I'm building something that does that reconciliation automatically." Line 3 — the ask, small. "Would you be up for 15 minutes to tell me if I'm solving the wrong problem?" Line 4 — the out. "Totally fine if not."

That's it. No deck, no calendar link in the first message, no "hope this finds you well", no paragraph about your background.

The ask is deliberately for their opinion rather than their money, because at this stage you genuinely need the opinion more — and because the conversation is where the sale actually happens, not in the inbox.

Send them by hand, twenty a day, and record every response including the silences. Twenty a day for a week is the whole list.

On the call, spend the first ten minutes not talking about your product.

Ask about the last time the problem happened. What they did. What it cost. What they've tried. Take notes in their words — this is the same interview from validation and it's still the highest-value thing in the call.

Then, only if what they described matches what you built, say: "That's exactly what I've built. Want to see it?"

Show it doing their task, with their kind of data if you can. Not a feature tour. A feature tour is a monologue; showing their job get done is a demonstration.

Then ask for the money directly and stop talking: "It's £X a month. Want to start?"

The silence after that question is uncomfortable and you must not fill it. Founders talk through the silence and negotiate against themselves — offering a discount nobody asked for, a longer trial, a free month. Let them answer.

If it's a no, ask one question: "What would have to be different?" That answer is worth more than the sale.

For the first ten, do the work manually. All of it.

• Onboard each one personally. A call where you set it up with them. You'll watch exactly where they get confused, which is worth more than any analytics dashboard you could build.
• Do the work by hand if the product can't yet. If it takes you two hours a week per customer, do it. Ten customers is twenty hours and you'll learn what to automate first — which is not what you'd have guessed.
• Message every customer weekly. A real message, not a newsletter. "How did last week go?"
• Fix their problem the same day. You will never again have the capacity to do this, and it buys a category of loyalty that turns a first customer into a reference.

This feels wrong to engineers because it doesn't scale. It doesn't need to. Ten customers served obsessively produce the testimonials, the case studies, the referrals and — most importantly — the precise understanding of what to build that no amount of automation would have given you.

After ten customers, stop selling for a week and look at what you have.

• Where did the ones who paid come from? Usually one source dominates. That's your channel.
• What did they have in common? Not demographics — situation. "Just took on a second client", "just got burned by their agency". That trigger is your targeting.
• What words did they use when they said yes? Those are your landing page headline, already tested.
• What did the ones who said no have in common? That's your exclusion criterion, and it saves you months of chasing the wrong people.

Now you have something worth scaling: a message that converted, a channel that produced, and a trigger that predicts. Automation applied to those is leverage. Automation applied before you have them just delivers an unproven message to more people, faster, which is how founders conclude that a good idea "doesn't work in this channel".

That's the point where the machine earns its place — and where kitty.build takes the proven angle and runs it continuously across your channels, with every outbound message still waiting for your yes.