One Interview Script Can't Do Justice to Every Expert

Every operations lead who runs expert interviews at scale eventually hits the same wall. The script that gets a former CTO talking about architecture decisions is the wrong script for a sales leader, a procurement head, or a practising clinician. Write one interview prompt broad enough to cover all of them and it asks shallow questions of everyone. Write a separate one for each role and you've signed up to maintain a small library of projects — and you still don't know which expert you'll actually get until the call starts.

This is not a tooling problem you can buy your way out of with a better questionnaire. It's structural. Below is why it happens, what it costs, and the model that actually resolves it.

The dilution problem

A single interview script is a compromise, and every role you add to its remit makes the compromise worse.

To do justice to a CTO it needs real infrastructure depth — migration triggers, build-versus-buy logic, what broke at scale. To do justice to a VP of Sales it needs the commercial machinery — deal cycles, who signs, what stalls a renewal, how quota pressure changes behaviour. To do justice to a procurement lead it needs vendor-selection mechanics — the shortlist, the switching costs, the contractual hooks that lock a buyer in.

Merge all of that into one prompt and each line of questioning gets thinner. The interview agent has to hedge: it can't go three questions deep on infrastructure because the same script has to work for someone who has never seen a data centre. The expert feels it. The questions read as generic, the follow-ups don't land, and the transcript comes back shallower than the expert actually was. You paid for an hour with a genuine insider and got a surface-level conversation because the script was built for an average that doesn't exist.

What "one script for everyone" actually costs

The dilution shows up in three places, and none of them are on the invoice.

Maintenance. The workaround most teams reach for is a separate interview project per role archetype. That feels organised until you're maintaining nine of them — nine prompts to keep current as the study evolves, nine places a change has to be mirrored, nine chances for drift between what you think you're asking and what's actually being asked.

Scheduling and routing overhead. Per-role projects only work if you know the role before the call. Often you don't — recruitment is fuzzy, an "ex-VP" turns out to have been an individual contributor, a "customer" turns out to have also worked for the target. Now someone has to triage and route each expert to the right project, and mis-routes produce exactly the diluted interview you were trying to avoid.

Depth left on the table. Every shallow transcript is a sourced, scheduled, paid-for expert who gave you less than they had. At scale that's the single most expensive line item nobody measures.

A pipeline that adapts at the start of the call

The model that resolves this is one pipeline that profiles the expert at the beginning of the conversation and then continues with a specialist line of questioning matched to who they actually are.

The handoff is silent. Same voice, no re-introduction, no "let me transfer you," no pause the expert can perceive. From their side it is one continuous conversation that simply happens to ask sharp, role-appropriate questions. From your side it is one pipeline that contains the role-specific depth you'd otherwise scatter across nine projects.

The reason this works — and the reason it isn't just "routing with extra steps" — is where it happens. It happens at the vetting stage, before the expert is a known quantity. Establishing who the expert is and what they can speak to is the entire point of that part of the call. Adapting the rest of the conversation on what you learn there is the natural use of that information, not a redundant extra step. If the expert's profile were already pinned down before the call, you'd simply assign the right line up front and no in-call adaptation would be needed. The adaptation earns its place precisely because the profile is genuinely open at that moment.

A worked example

A research team is running a channel study on a mid-market B2B software vendor. The recruited expert is described as "former engineering leadership." The interview opens with profiling questions and the expert explains, in passing, that their actual remit was running the procurement function that bought and later replaced the vendor — they sat on the buyer side, not the build side.

A one-size script would have spent the hour asking architecture questions the expert could only answer second-hand. Instead the conversation continues on the buyer-side line: how the shortlist was built, what the incumbent did to defend the account, the real switching costs, the moment the renewal conversation turned. Same call, same voice, no restart — but now the transcript is a procurement insider's account instead of a hedged generalist one.

That is the whole value: the expert was always going to be worth more than the script. The pipeline just stopped the script from being the ceiling.

What changes operationally

• Fewer scripts to maintain. One pipeline with role-specific lines instead of a project per role. Changes are made once.
• One call, not two. When an expert clears your vetting questions, the same call continues straight into the full interview — no second session to schedule, no expert lost in the gap between two bookings.
• Global panels, in-language. Confirm the expert's language at the top of the call and the interview continues tuned to that language and locale, rather than an English script translated live.

What this isn't

It is not magic, and it is not a reason to stop thinking about your instrument. You still design the role-specific lines — the depth comes from your question design, not from the routing. The pipeline's job is to make sure the right line is the one that runs.

It is also not warranted for every study. If a project only ever talks to one tightly defined role, a single well-built script is the right tool and anything more is overhead. The adaptive model earns its keep when you genuinely don't know who you'll get, or when "who you'll get" spans roles that need materially different conversations.

The result, in the cases that fit, isn't a flashier interview. It's a deeper transcript from the same expert, with less operational machinery behind it.

See how an adaptive interview pipeline is put together →