Book customer interviews by cold email before you write code

Before writing code, find people who match your target customer, send each one a short personal note, and ask for an hour to talk about an idea built for them. The documented case is WP Engine. In February 2010 Jason Cohen did this for his WordPress hosting idea, and he reports that 40 of 50 people agreed to talk and 30 later became customers.

How it works

Cohen's own write-up of how he found people to interview gives the steps. The WP Engine case page collects them in one place.

1. Pick a segment you can search. Cohen used LinkedIn to find people by title and industry: web developers who worked with WordPress.
2. Research each person briefly. He checked who each business served, where it was based, and whether a recent blog post gave him something personal to mention.
3. Ask for an hour, framed around their expertise. He asked for an hour to talk about a new startup concept and told them they were its ideal customer.
4. Offer to pay. His message offered any fee the person felt was fair and made clear he was not asking for a favor. Only one of the 40 who agreed asked to be paid.
5. Test willingness to pay. In an account of the outreach on Marketing Examples, Cohen is quoted saying thirty people agreed to pay $50 a month. He also says he had those 30 willing customers before he had a company name, slides or a server.

The two sources give different counts. Cohen's essay says 50 requests, while Marketing Examples says he emailed forty consultants. Cohen founded WP Engine in 2010 and bootstrapped it at first, as his author page states.

What it takes

Our estimate: about 25 hours of work, and about 14 days before you can read a signal. The hours go into building a list of 50 named people, a few minutes of research on each, writing the messages one at a time, and running the first batch of hour-long calls. Replies tend to arrive within days. Calls take another week or two to schedule, though, and you cannot judge willingness to pay until you have finished enough conversations to compare.

What to measure

Agree-to-talk rate tells you whether your segment and message work. A low rate across a large batch usually means the segment is wrong or the note reads as a template. A high rate means you found people who recognize the problem.

Share asking for the fee shows whether people see the call as a favor or as paid consulting. If many ask for payment, your ask may read as a sales pitch, or your segment bills by the hour and expects it.

Price commitments before product is the reading that matters most. Agreeing to talk costs nothing. Agreeing to a specific monthly price is a stronger signal, and it is what the WP Engine case reports.

Commit-to-paid conversion at launch keeps you honest. A verbal yes is not a payment. Track how many committed interviewees actually pay when you open signups. Compare this with the free-to-paid conversion benchmarks only loosely, because these are hand-picked prospects.

Interviews per hour of effort tells you whether the personal research is paying for itself or whether you can trim it.

When it backfires

• The segment has no search handle. Cohen could filter by title and industry. If your buyer has no common job title, you will spend the hours on list building and get few replies.
• Messages go out as templates. The case relied on short personal research for each person. Mass sending drops the per-person research the case used.
• Polite interest gets counted as demand. People who agree to talk are not customers. Ask for a price and a start date, then track who follows through.
• The call turns into a pitch. You asked for their expertise. If you spend the hour selling, you learn little, and the next referral dries up.
• One case gets treated as a benchmark. The 40 of 50 figure comes from one founder in one niche. Your rates will differ.

Run it as an experiment

Hypothesis: personalized notes that offer a fee for an hour of expertise will book more interviews, and more price commitments, than a short generic ask.

Comparison: we recommend splitting one list into two arms of about 25 people each, sent in the same week. Arm A gets a personal line and a fee offer. Arm B gets a plain request. With samples this small, treat the result as directional. The sample size calculator will show how far you are from a reliable split.

Log: send date, segment, arm, reply date, agreed or declined, fee requested, interview held, price stated, commitment yes or no, and later whether they paid.

Decision rule: our suggestion is to adopt the winning message only if it books clearly more interviews and produces at least as many price commitments. If neither arm yields any commitments after a full batch, change the segment before you change the copy. Track the test in the experiment backlog template and write it up using how to run a growth experiment.

How strong the evidence is

The grade is case with numbers: one company documented the tactic and published counts, quoted from the founder by name. WP Engine carries it alone, through Cohen's own account. The numbers are self-reported, and the two sources disagree on whether he contacted 50 or forty people. There is no baseline or control, so we cannot separate the effect of the fee offer from the personal research. The sources do not say when the 30 became customers, or how long they stayed.