How Do You Find Product Market Fit Through Marketing Research?

Problem interviews return symptoms you cannot act on because you asked someone to describe their own working life and they obliged. A list of complaints holds no comparison inside it. Without one there is nothing to price, nothing to position against, and no way to tell a serious buyer from a sympathetic one.

Most product market fit frameworks open with a pain interview script. Ask about the workflow, ask what is hard, ask what they have tried. The answers come back in the customer's own technical vocabulary. Every one of them is true.

They are also unranked. Everybody has twenty irritations, and nobody has ever signed a contract for an irritation.

There is a second failure, and it gets worse the longer you have been selling. By the fourth call your customer has learned your vocabulary, so when you ask what is hard they answer inside your categories, and what comes back is your own thesis read aloud in a stranger's voice. That is the most expensive research there is: it feels like evidence and it moves nothing.

What I keep seeing is that the founder already knows the pains, in more detail than any interview will produce. The missing piece is the comparison, and the comparison is the only part of the picture the customer holds and the founder does not.

The replacement question is one sentence, put to customers who bought quickly, paid full price and defend you internally when something breaks. Its answer is the alternative your product is being measured against. That alternative decides what better means, and what the room expects to pay before you have opened your mouth.

Ask it in these words:

“If we disappeared tomorrow, what would you actually do instead? Not which vendor you would call. What you and your team would do next month.”

Two details carry the weight. Not which vendor. The vendor list is the one part of the market your customer has never had to live inside, and naming it is the reflex you are trying to interrupt. And next month, because a deadline moves the answer off principle and onto logistics: who would do the work, out of whose budget, in which week.

There is a reason this order matters. April Dunford's Obviously Awesome (2019) puts competitive alternatives first among the components of positioning, ahead of your own attributes, because whether an attribute is unique at all depends on what it is being held up against.

The question above, and the test that follows it, are the house exercise. I published the one hour version, with the follow up question and how to group the answers, as what your best customers would do if you did not exist. It is the same hour whether you file it as positioning research or as a product market fit test. What changes is what you do with the column of sentences at the end of it.

Here is what that column looks like for an invented company, one that files regulatory reports for mid sized insurers. The quotes are written for this article, not taken from anyone.

Bought in three weeks

“Nothing clever. We would go back to living with the March scramble.”

The status quo. Priced in risk and bad weeks, never in licences.

Paid full price

“Our two analysts would rebuild the sheet and work the weekends.”

Internal labour. You are weighed against two salaries and their other work.

Refers you to peers

“We would ask our auditor to take it back and pay them for it.”

A services budget, not a software one. Different shelf, different number.

Defends you internally

“We would take the module our core system vendor keeps offering.”

The only named vendor, and it arrives bundled with something already bought.

1

One of the four answers names a vendor. The other three are the market: a bad month, two analysts, a professional fee. Each one implies a different buyer, a different budget line and a different price, which is why grouping by what people would do beats grouping by what sector they are in.

You read the answers into segments by grouping customers by what they would do instead, not by who they are. Then you check whether each group can become a list of companies you could name and call this afternoon. Before the grouping is useful, write one line per bucket: what only you have against that specific alternative, and what that lets the customer reach. Against two analysts it may be the weekend they stop losing; against the auditor, a fee that stops recurring. The attribute only becomes value once the alternative is named.

That is the list test, and it is unforgiving in a useful way. If three answers land in internal labour, the segment is not mid market insurers. It is companies that already employ the analysts who would absorb the work. You can see that from the outside, so you can list it. If two answers land on the auditor, the segment is companies whose compliance work already sits with an external firm. The sale there is a budget transfer, not a new line.

Both of those are lists you can build. If what you have cannot become a list of names, you have described a mood.

The segment makes the next decision possible, which is one category. A category is not a label you pick for how it sounds. It is the shelf your buyer puts you on, and the shelf carries the comparisons, the vocabulary and the number they have in mind before the first call. The bucket with the most best customers in it tells you which shelf your own evidence already belongs to.

Then it gets captured, once, so that the homepage, the accelerator or investor portfolio page, the LinkedIn tagline and the sales conversation all repeat the same sentence. That step looks administrative and is not. What the committee hears is whatever version reached them last, and the materials each function needs are built out of that settled sentence rather than in place of it, which is the argument I made in materials for each stakeholder in a buying committee.

Past Ellis's heuristic there are no figures I can hand you on how many interviews are enough, or what share of answers should fall into one bucket. I went looking. What the work does have in its favour is cost: five phone calls, a blank document and an afternoon, which puts it at the front of the queue when there is no budget. I set out what else belongs in that queue in how to build a marketing strategy with limited resources.

The most common answer in B2B is that nothing would happen: they would live with it. That is not a failed interview, it is the finding, and it means the thing beating you is the status quo, which never sends a proposal and never appears in a bake off.

What that bucket asks of your research is a different output. Not the benefit of buying: the people in it have already shrugged at that. What they need named is the cost of the month that follows when nothing is decided, in units they already report: days, headcount, late filings, write offs, the March scramble. The customers in that bucket can supply those units. They lived there before you arrived, and they remember the texture of it. That is what the second question in the published exercise is for, and it is why the answers have to be written down verbatim and left ugly.

The method I use for the rest of it is called The Loom: week one is a diagnosis drawn from your own calls, wins and words, week four puts the positioning document in your hands, and weeks five to ten turn that document into a go to market your team runs without me. It ends rather than renews, which is the whole difference from a monthly seat: one person sets the direction and builds it, and if that week four document has not shown you the system, you cut the thread, the second payment never happens and you keep everything delivered.

This is where research stops describing the market and starts testing your own go to market. If your deals pass through risk, IT, compliance or procurement, two of the twenty statements in the Committee Readiness Audit sit on exactly this seam, in the audit's own words: “When a deal is lost, it is almost never because the buyer misunderstood what we actually do.” And: “We can articulate the cost of the buyer not acting, not just the benefit of acting.”

What the audit gives you back is the loosest of four threads, named, and read against the replacement answers you have just collected. A low score on the cost of the buyer not acting tells you the “nothing, we would live with it” bucket is the one to work on first. Twenty statements, five on each thread, rated 1 (not true of us at all) to 5 (completely true), a score out of 100 in one of three bands, a few minutes. One email, given before you start, opens both the interactive version and a printable PDF of the same twenty statements.

Score each statement against your last real deal rather than against what you meant to build. The next step after those five calls is to find out which thread will give way first: take the Committee Readiness Audit.

Sources: Sean Ellis's product market fit survey. Obviously Awesome, 2019.