Consulting & AdvisoryFree
Qualitative Interview Data Synthesis Report
How many raised it, how many were asked and did not, and how many were never asked, on every theme you report.
River reads every transcript in full and returns a coded extract sheet plus a theme register. Each theme carries three counts: how many people raised it, how many were asked and did not, and how many were never asked. It also carries where the first mention came from in each transcript, and a re-identification check against the number of people who hold that role, because a theme held by four of five supervisors names all four.
Search this and page one is a doctorate. Six phases of reflexive thematic analysis, familiarise then code then construct then review then define then write, a thematic map on sticky notes, NVivo. It is a real method and it was built for research with no client in the room. A consultant has three weeks, nineteen named employees of one company, and executives who will read the report and try to work out who said what.
Built for the management consultant, the operations reviewer and the fractional executive running discovery inside a client. Run it once the interviews are done, against the plan the interview programme set, so the coverage targets and the actual counts sit side by side. Where the recordings are still recordings, a quote bank carrying timecodes comes first. What comes out is the evidence layer under findings that name their criteria, and themes nobody can attribute safely change channel instead of vanishing.
The published test for who can be identified
There is a published test for whether a reader can identify your source, and it is not whether you used a name. 45 CFR 164.514 sets the standard as whether the risk is very small that the information could be used, alone or in combination with other reasonably available information, by an anticipated recipient to identify an individual. Both qualifiers are the point. The anticipated recipient is the executive reading your report, and the other reasonably available information is the headcount of the team.
The same section adds a second condition that translates directly. A body relying on the standard must not have actual knowledge that the information could be used, alone or in combination, to identify the person. You do have that knowledge. You know there are five shift supervisors because you counted them to build the schedule. So the check runs against the role population in the client organisation, not against the number of people you happened to interview.
The denominator is the other half. Federal survey standards plan a nonresponse bias analysis whenever the response rate for any item used in a report falls below 70 percent, and the logic carries straight into an interview programme. The people who were never asked a question are rarely a random subset: they are the ones you interviewed first, before the theme existed. So a theme reports three numbers, and five of nineteen never gets written as widespread.
How it works
Send the transcripts
Every recording or transcript, however it was captured, plus the ones with no speaker tags.
Give the headcounts
How many people hold each role you interviewed, because that is the identification denominator.
Read the register
Each theme with its three counts, its first mention, and whether it can be attributed.
Set the channels
Written, verbal, or held back, decided per theme against what the reader can work out.
What you get
- Every theme with three counts: raised it, asked and silent, and never asked at all
- The provenance of each first mention, so a theme you named is flagged as yours
- A re-identification check per theme against the role population, not the interview count
- A reporting grain and a delivery channel set per theme rather than per report
- The sponsor's own position on every theme, recorded even where it contradicts the finding
- One row per coded extract, with the role, the level, and the exact transcript location
- Themes held inside a single source cluster flagged as one voice, however many people repeated it
Common questions
What does it need from me?
The transcripts in whatever form you have them, including the ones with no speaker tags. Then the headcount for each role you interviewed, because four of five supervisors and four of forty are the same count and completely different findings. What you promised people about confidentiality shapes the channel column, so say that too.
An executive says nobody said that to them. What now?
You give the interview number and the timecode, and the conversation ends. Every extract in the register carries where in which transcript it came from, so the answer is a reference rather than an argument about your recollection. The same discipline over the client's paperwork is the cited evidence log.
Why does it matter who mentioned a theme first?
Because agreement with a proposition you supplied is not evidence for it. Where a theme appears in a transcript only after you named it in a question, the register says so and the theme is held rather than reported. It goes back out as a neutral question in the next round, and often comes back smaller.
I promised confidentiality. Can I report a theme only two people raised?
Yes, at a grain that does not identify them. The check compares the theme's holders against how many people hold that role in the client, and where the arithmetic points at individuals the theme is reported at a coarser level, delivered verbally, or held. The theme is never quietly dropped.
What happens when the sponsor disagrees with a theme?
Their position gets recorded against the theme and the theme stays. Deleting the sponsor's view makes the finding fragile, and saving it for the board makes it a confrontation. So it goes to them in writing first, with both counts visible and their objection quoted at the same length as the theme.
Is this the same as thematic analysis?
It uses the same coding. Then it adds the three things a client engagement needs and a dissertation does not: a denominator per theme, the provenance of the first mention, and a check on whether a named executive can work out who spoke. For participants who are strangers rather than employees, user research synthesis is the version to use.
Where does this fit in the engagement?
Straight after the interviews, before anything gets drafted. The register is what a readout is built from, and the counts are what stop a theme two people raised becoming a slide that says the organisation believes. Where the same evidence has to be scored rather than narrated, a rubric with observable anchors is the neighbouring job.
Qualitative Interview Data Synthesis Report
Fill in the form and your workspace opens with the work already underway.