Your first ten customers are three different jobs, not one

The gap nobody warns you about

Ask a deeptech founder with one customer how it happened and you get a great story. A chance meeting, a professor's introduction, a corporate innovation scout with a thesis to prove. Then ask how customer two is going, and the story stops. Because customer one was an event, and events do not repeat on schedule.

I have watched this gap swallow companies whole. The technology is proven, the first pilot converted, the logo is on the website, and then: fourteen months of nothing. Not because the market rejected them, but because nobody in the company understood that the road from one to ten is not ten copies of the road from zero to one. It is three different jobs wearing a trench coat, pretending to be one.

Here is how the ten actually break down.

Customer one is a favor

Let us be honest about the first deal. However it arrived, through a network contact or a desperate buyer or a design partner program, it was not produced by a process. It was produced by an exception: someone took a risk on you that they had no rational obligation to take.

This matters because of what it fails to teach you. Customer one teaches you that your technology can survive a real environment. It does not teach you how to find the next buyer, because you did not find this one. Companies that do not admit this spend a year waiting for lightning to strike twice, in the same place, on schedule.

Customers two through four are the proof-of-process job

The second through fourth customers are the hardest of the ten, and almost nobody writes about them, because there is no glamour in the description. This is the phase where you must do, on purpose, what customer one did by accident. Same industry. Same problem. Same buying map. Deliberately.

The temptation is to diversify. You have one chemical plant, so why not try food processing, and maybe that OEM who called? Resist it. Two through four exist to prove one thing: that the first sale was a pattern, not an accident. Three customers in the same segment, with the same problem, bought through the same sequence, is a machine. Three customers in three different segments is a collection of miracles, and investors, procurement teams and your own future sales hire can all tell the difference.

There is a practical reason too. Your reference story only compounds within a segment. When the second chemical plant asks "who else uses this", the answer "a plant like yours, twenty minutes from here" closes the meeting. The answer "a bakery and a satellite company" opens a discussion.

The referral machine you are sitting on

Here is the part founders systematically underuse. Industrial people talk. Plant managers visit each other's lines. Process engineers change employers and carry their supplier preferences with them. The maintenance chief of your first customer plays cards with the maintenance chief of your fourth.

Which means every delivered customer is not one logo. It is a broadcast station. The question is whether you operate it. After every successful delivery, three asks should happen as a matter of process: who else do you know with this problem, would you take a reference call from them, and may we say your name when we call? Most founders make zero of these asks because it feels like imposing. Their customers, having just spent money and been proven right, are at the exact moment of maximum willingness to impose.

Customers two through four come disproportionately from this channel if you run it. A warm introduction inside an industry converts at a rate cold outreach cannot approach, and it arrives pre-armed with the sentence every industrial buyer wants to hear: someone like me already took this risk and survived.

Customers five through seven are the discipline job

Somewhere around the fifth customer, a new failure mode appears: the pipeline has enough in it that things start falling out the bottom. Follow-ups slip. A proposal waits three weeks. The champion at company six goes quiet and nobody notices for a month, because the founder was at the commissioning of company five.

This is the phase where the work stops being persuasion and starts being administration, and technical founders hate that sentence. But the companies that get through five to seven cleanly are the ones that wrote things down during two to four: the qualification criteria, the stage definitions, the pilot template, the response-time discipline. The pipeline is now too big for memory. This is also the moment where extra commercial capacity pays for itself fastest, because an hour of follow-up work here directly saves a deal that took four months to create.

Customers eight through ten are the decision job

By eight, something pleasant happens: the segment knows you exist. You get invited to meetings you did not chase. And right there, at the moment of arrival, sits the trap. Because now adjacent segments start calling too, and the diversification temptation returns, this time wearing a suit and carrying a purchase order.

The decision of customers eight through ten is not how to get them. It is what they are for. Are they the consolidation of segment one, making you the default supplier of this application? Or are they the bridge into segment two, chosen deliberately because the same problem lives there with the same buying map? Both are legitimate. What is not legitimate is letting inbound decide, because inbound is random, and your second beachhead is too important to be chosen by whoever happened to email.

Ten is not a number, it is a machine

Companies celebrate the tenth customer as a milestone. It is actually a diagnostic. If you can describe, in one page, how customers eight, nine and ten arrived, and the description matches how five, six and seven arrived, you have a machine. You can now pour resources into it, hire into it, or bring in a partner to run it faster. If every customer still arrived by a different miracle, you have ten customers and no company, and the next ten will take just as long as the first ten did.

The founders who get there fastest are not the ones with the best technology. They are the ones who treated customer two as the start of a process, rather than the continuation of a lucky streak.

Common questions

How long does it take to get ten customers in deeptech?

In industrial markets, eighteen months to three years is normal. The spread is explained less by technology than by whether customers two to four were pursued deliberately in one segment, or waited for opportunistically across several.

Should I take customers outside my beachhead segment?

Before customer five, almost never: you are proving a pattern, and scattered references do not compound. After eight, yes, but deliberately, choosing the second segment because its problem and buying map match, not because an inbound inquiry arrived.

What is the fastest source of customers two through four?

Referrals from customer one, run as a process: after every successful delivery, ask who else has this problem, ask for the reference call, ask to use their name. Warm introductions inside an industry outconvert every cold channel.

If you have customer one and a quiet calendar, the problem is usually not the market. It is that nobody is running the two-to-four job. That job is what we do.