Feed discovery interviews with prior evidence
Discovery interviews often re-ask questions customers already answered in past calls, tickets, and surveys. NEXT reads that prior feedback on the interview topic and compiles what is already known. The researcher gets a short pre-read: what is settled, what is still open, and which questions are worth the live time.
Most research programs lose this context at the handoff. Whoever runs tomorrow's interview rarely has the last six months of related signal in front of them, so the session starts cold.
What the pre-read looks like
Here is a representative pre-read for a session on self-serve onboarding. Example output based on grouped onboarding feedback, support tickets, and prior interview notes.
Interview topic
Self-serve onboarding — first workspace setup
What we already know
Customers consistently stall at the data-source connection step, before field mapping. Several reached setup completion only after contacting support.
"We got to the data-source step and just stopped. Nobody on our side knew which credentials it wanted."
"We finished setup by booking a call with your team. It worked, but I wouldn't call it self-serve."
Accounts that raised this
19 accounts in the last two quarters, mostly mid-market, including three currently in onboarding.
Commercial context
About $380K ARR sits with accounts that flagged the connection step.
Demand summary
The friction is concentrated and repeating at one step. Prior signal points to credential ambiguity, not the mapping screen.
Still open
Do customers know which credential type to use before they start, or discover it mid-flow?
Is the blocker documentation, permissions, or product wording?
Does this differ for self-serve vs. sales-assisted accounts?
Signal strength
Strong and consistent on where customers stall; thin on the underlying cause — which is what the interview should resolve.
The brief is ready before the session, not reconstructed from scattered notes the night before.
How NEXT does this
NEXT reads where customers already speak — support tickets, call transcripts, surveys, reviews, and past interview notes. It keeps a continuously updated record of what customers have said, organized by topic. When a research session is scheduled, NEXT pulls the prior signal on that topic, separates what is well-supported from what is still unresolved, and writes a pre-read for the researcher with the open questions called out. The brief lands where the researcher prepares.
The researcher decides what to ask and how to run the session. NEXT only assembles the prior context so the interview starts from accumulated memory instead of a blank page.
Why these briefs take so long today
Preparing a discovery interview properly means reading old call notes, searching tickets, and remembering which survey covered the topic. Most researchers don't have the time, so they skip it and re-ask questions the memory already answers. The interview then confirms what was known instead of pushing past it.
The tools meant to help don't close the gap. The dashboard may be faster, but the brief still arrives too late — someone has to remember to open it and assemble the story by hand. An AI assistant returns what you ask, not what the session needs; it surfaces the loudest thread, not the unresolved one. Both wait on a person to do the work.
NEXT pushes the prior context to whoever is running the session, with no interface to adopt. It delivers an assembled pre-read, not a search box — grounded in how the research program actually accumulates evidence.
How this compares to the tools you already know
Approach | Where the evidence lives | What product ops does before the session |
|---|---|---|
Shared research repository | In docs and folders, by project | Searches across projects and hopes the relevant note is tagged |
AI assistant | Wherever you prompt it | Asks the right question and judges what comes back |
NEXT | In a continuously updated record by topic | Opens a pre-read that is already assembled, with open questions marked |
What changes for product operations
Today you prepare each interview from scratch, or you don't prepare and the researcher improvises. Either way, sessions vary in quality depending on who ran them and how much time they had.
With the pre-read in place, every interview starts from the same accumulated memory. You schedule a session on onboarding friction; before it starts, the researcher already has the 19 affected accounts, the two recurring quotes, and the three questions still unanswered. The session opens on the open questions, not on ground already covered.
The shift is small but compounding. The interview that used to spend its first ten minutes re-establishing context now spends that time probing the unknown. The researcher stops re-asking "where do you get stuck?" and starts asking "why didn't the documentation help?" Findings get more consistent across interviewers because everyone works from the same prior record. NEXT already supports product and GTM teams at companies like Deel and Visma in connecting customer evidence from calls, tickets, and reviews to product decisions.
NEXT supplies the prior context; the questions, the judgment, and the session stay yours.
Downstream effects
Interview time goes to open questions. Less of each session is spent re-confirming known facts, so more usable findings come out per interview.
Findings compound. Each session is written back into the record, so the next pre-read on the topic starts richer than the last.
Research gets more consistent. Different interviewers work from the same accumulated evidence, which is the operational consistency product ops usually tries to enforce by hand.
Where the human stays in control
You set what counts as enough prior signal to include, and how a topic maps to sources. You can require a human to review the pre-read before it reaches the researcher, or let well-supported topics flow through. This is configuration work, not approval work — you tune what gets compiled and when, then the researcher decides what to do with it. NEXT never runs the interview or decides the findings.
What the brief depends on
The pre-read is only as good as the prior signal on the topic. A few things to get right:
Source coverage. Calls, tickets, surveys, and past interview notes need to be connected, or the brief reflects only part of the picture.
Topic scoping. The topic has to map cleanly to how customers talk, or the brief pulls adjacent-but-wrong evidence.
Delivery timing. The pre-read has to arrive before the researcher prepares — early enough to shape the question set, not after it's locked.
Thin-signal honesty. On new topics with little prior evidence, the brief should say the signal is thin rather than pad it.
Where this breaks down
A brand-new topic has no prior evidence.
If the program has never touched the area, there is little to compile. The pre-read should mark the signal as thin, and the interview runs closer to cold — which is the correct outcome, not a failure.
The topic is scoped too broadly.
"Onboarding" pulls everything; "first data-source connection" pulls the right thread. A vague topic gets a vague brief with weak relevance.
Old evidence gets treated as current.
Signal from a year ago may describe a flow that has since changed. The pre-read should date its evidence so the researcher doesn't probe a problem that is already fixed.
The pre-read becomes a substitute for asking.
The brief is the starting point, not the answer. A researcher who treats the compiled context as settled truth stops probing — and the interview loses its point.
FAQ
Does NEXT replace the researcher or the interview?
No. NEXT assembles the prior context so the session doesn't start cold. The researcher still designs the questions, runs the interview, and decides what the findings mean. The pre-read changes what you walk in knowing, not who does the research.
How is this different from searching our research repository?
A repository holds notes by project and waits for you to search it well. NEXT keeps a record organized by topic and assembles the relevant prior signal into a pre-read before the session, with unresolved questions already separated from settled ones. You read it instead of hunting for it.
What if there's no prior evidence on the topic?
Then the brief says so. NEXT marks the signal as thin rather than padding it with loosely related material, and the interview runs closer to a cold start. That's the honest result — a sparse pre-read is more useful than a confident-looking but irrelevant one.
Won't a pre-read bias the interview toward what we already think?
It can, if treated as settled truth. That's why the brief separates well-supported findings from open questions and dates its evidence, rather than presenting everything as equally certain. Used well, it points the session at what's unresolved rather than confirming prior assumptions — the researcher still decides how much weight the prior signal deserves.
Can different researchers rely on the same brief?
Yes, and that's much of the point. When every session starts from the same accumulated record, findings get more consistent across interviewers instead of depending on who prepared and how much time they had. A researcher picking up a topic for the first time starts from the same context as someone who has run five sessions on it.
Does NEXT decide what we ask?
No. NEXT highlights which questions are still open based on prior evidence, and which findings are already well-supported. Which of those open questions you pursue, how you word them, and how you run the session stays with the researcher — NEXT supplies the map, not the script.