Every person I interviewed was a friend of a friend, where do you find strangers who have the problem?

The option people dismiss and should not: pay for the interviews.

A modest amount for half an hour: a voucher, a transfer, whatever is proportionate: changes several things at once:

Reply rates go up a lot, which solves your access problem directly.

You get people who are not doing you a favour, which means they are willing to be blunt. Free interviews are given by nice people, and nice people are encouraging.

You can specify who you want, rather than taking whoever your network produces. This is the thing you actually lack.

It removes your guilt about asking, which is currently stopping you from asking at all.

There are services that recruit participants to a specification, and they are not cheap. For a small number of interviews, doing it yourself and paying individually is affordable.

One caution: paying does not buy you honest answers about whether they would buy: nobody is reliable about that, paid or not. It buys you access and candour about their current situation, which is what interviews are actually for.

And keep asking about the past rather than the future. Paid or free, would you use this produces the same useless yes.

Go to where the problem is already being complained about, and read before you speak.

The places that work:

Wherever people ask for help with the current bad solution. Forums, communities, support channels for the tools they use now. Somebody posting about a workaround is somebody with the problem, right now, describing it in their own words.

Reviews of competing products, especially the two and three star ones. Five stars tell you nothing; the middle ones are people who tried and were disappointed and explain exactly why.

Job listings, if your customer is a company. Roles being hired for tell you what somebody is trying to solve badly enough to pay a salary for.

Local or professional groups for the trade in question, if your customer is not a software person.

The important part is that you read for a while first. You will learn the vocabulary, and the single biggest reason cold outreach fails is using your words rather than theirs.

Then approach individuals, not the room. Posting can I interview people about X in a community gets nothing and often gets you moderated. Messaging one person about the specific thing they wrote gets a reply surprisingly often.

On the cold message itself, since intrusive is a function of how you write it.

What works, in about five lines:

Reference the specific thing they said or did. Not a compliment on their work generally - the actual post, the actual review.

Say who you are in one clause, honestly. Somebody looking into this, not a researcher and not a company.

Ask for something small and specific. Fifteen minutes, or even just two questions by message. Fifteen is a much easier yes than a call.

Ask for advice, not for their opinion of your idea. How do you currently handle X is a question people enjoy answering. Would you use my product is work you are asking them to do for free.

Make it easy to decline and mean it.

What kills it: any hint of a pitch, a calendar link before a yes, a long paragraph about your background, and the word revolutionise.

Expect a low reply rate and do not read that as a signal about the idea, it is a signal about cold messages. Ten replies out of a hundred is a good outcome and ten conversations with strangers will teach you more than your eleven friendly ones did.

And never ask a friend to introduce you to someone like them. That is how you get another eleven of the same person.