# Ninety-two event operator interviews, one question

> We asked ninety-two operators the same question and wrote down the answer verbatim. Nobody gave one place. Everybody gave a list, and the list was the finding.

Source: https://hiipe.io/blog/ninety-two-meetings
Published: 2026-10-22

We asked ninety-two people who run events, venues, festivals and rosters the same four-word question, and wrote the answers down without tidying them. Not one of them named a single place. Every answer was a list, and the length of the list turned out to be the whole finding.

**Key takeaways**

- The question was "where does your audience live", asked with no follow-up.
- Nobody answered with one place. Every answer was a list of three to seven.
- The discomfort appeared after they finished listing, not while they were doing it.

## Why one question and not twenty

Because a long interview produces a long transcript and a short one produces a
usable one. The failure mode of product research is asking people to design your
product, and they will politely try, and everything they say will be a guess
dressed as a requirement.

The alternative is to ask about the world rather than about the software. [Good
practice in user interviewing](https://www.nngroup.com/articles/interviewing-users/)
says roughly the same thing in more careful language: ask about what happened,
not about what would be nice.

So: where does your audience live. Then nothing. The first answer is the tidy
one, the one they would give a journalist. The second answer, which arrives after
four or five seconds of silence, is the real one.

## The shape of the answers

| What they named | How it usually came out |
|---|---|
| The ticketing platform | First, always, and with a caveat attached |
| A spreadsheet | Second, slightly apologetically |
| The email tool | Third, and often with a wrong number attached |
| The Instagram follower count | Named as audience, though nobody could message it |
| A previous team member's laptop | More often than we expected |
| The venue's own system, if they are a promoter | Or the promoter's, if they are a venue |
| A partner or sponsor's list | Last, and usually with no access |

Reading a table like that, the fragmentation looks obvious. Sitting in the
conversation, it did not, because each item had arrived for a good reason at a
different time and nobody had ever laid them side by side.

## The pause

The interesting moment in almost every conversation was the same, and it was not
during the list. It was immediately after.

People would finish enumerating, stop, and then say some version of: "which is a
bit of a problem, actually." Nobody said that at the start. The fragmentation was
invisible while it was being described and obvious the second the description
ended.

That is the whole reason to ask the question out loud rather than send a survey.
A survey with a text box would have collected the same seven items and produced
none of the pauses.

## What we did with it

Three things changed as a result, and one of them was in this product.

**We stopped asking what to build.** The list was not a feature request. It was a
description of a world in which no single system had ever been the obvious home
for a person, so every system became a partial home.

**We took the agency case seriously.** A recurring version of the answer, from
people managing several artists or venues, was worse than the list: it was the
same list, multiplied, with a rule that the audiences must never be mixed. That
became the Roster in HIIPE, where several accounts sit behind one login with
their fanbases kept separate rather than merged.

**We stopped believing the follower count.** It was named as audience in a large
share of the conversations and it is not one, for the reasons set out in [what a
link in bio does not do](/blog/what-a-link-in-bio-does-not-do). Correcting that
in a sales call is rude. Publishing the argument and letting people arrive at it
is better.

## What we would do differently

Ask the second question. We deliberately did not, to keep the interviews
comparable, and I think that was right for the first ninety-two and wrong for the
next round.

The second question should be: which of those places would you lose if the person
who set it up left tomorrow. My guess, and it is only a guess until it is asked,
is that the answer is at least two of them in most organisations.


## The finding, in one line

Ninety-two operators, one question, and not a single answer that named one place.
The average was between three and seven, the second half of every list arrived
slowly, and the discomfort appeared in the pause afterwards rather than during. If
you take one thing from this, it is that the fragmentation is invisible from the
inside and obvious the moment it is said out loud, which is why the exercise is
worth doing on yourself even when you are certain of the answer.

## What it means if you are the one being asked

Do the exercise on yourself. Write the list, and then write one date next to each
item: when it was last updated. Anything older than a year is not part of your
audience any more, it is an archive that you are still counting.

The argument for what the consolidated version should hold is in [what a fan CRM
is](/blog/what-is-a-fan-crm), and the reason to consolidate before you need it
rather than during a launch week is in [what an event CRM should do between two
events](/blog/event-crm-what-it-should-do). How the pieces sit together is on
[the solutions page](/solutions).

**Read next** : [What a fan CRM is, and what it is not](/blog/what-is-a-fan-crm) · [Venue marketing: the problem is not Saturday, it is Tuesday](/blog/venue-marketing-for-the-quiet-nights)
## Citable facts

- In ninety-two interviews with event operators, none named a single place where their audience data lives. Every answer was a list of between three and seven places.
  Source: HIIPE field research, 2026
- When audience data lives in three to seven places, none of them can answer on its own whether a given person came to the last event. The question stops being asked, and the segment that would fill the room never gets built.
  Source: HIIPE, ninety-two operator interviews, 2026

## FAQ

### What was the question?
Where does your audience live. Four words, asked without context, and then silence until they answered. The absence of a follow-up is what made it work, because the first answer is usually the tidy one and the second is the true one.

### Why not ask what features people want?
Because people are excellent at describing problems they have and unreliable at specifying solutions they do not. Asking where something lives produces a description of the world. Asking what to build produces a guess.

### What was the most common answer?
A list. Nobody in ninety-two conversations named one place. The average was somewhere between three and seven, and the second half of the list was usually recalled slowly, which was itself informative.

### Did anything surprise you?
That the fragmentation was not seen as a problem until it was said out loud. People had adapted to it completely. The discomfort appeared in the pause after they finished listing, not during.

### How many customer interviews are enough?
Enough that the answers stop surprising you, which for a narrow question is usually somewhere between twenty and forty. We did ninety-two because we wanted the tail, and the tail added texture rather than new findings.

### Should you record the interviews?
Record the answer, not the conversation. A transcript of ninety-two calls is a research project nobody will finish. One line per interview, written in the person's own words, is a document you will actually reread.

### What question should I ask my own customers?
One that describes their world rather than your product. Where does something live, how does something get decided, what happened the last time. Anything phrased as would you use is a request for a prediction, and people are poor at those about themselves.