Research

How to write an analytic memo from qualitative interviews

Dylan de Heer, Co-Founder & CPO

Dylan de Heer

How to write an analytic memo from qualitative interviews

An analytic memo is a working note in which you develop an interpretation of your qualitative material. It records what you think an account or pattern might mean, the evidence behind that thinking, the uncertainty that remains and what you should examine next.

To write one, choose a focused question, identify the relevant material, explain a possible relationship, challenge it with contrasting evidence and leave a concrete next check. You do not need to wait until every interview has been coded or the final report is ready.

This guide uses a fictional interview example to show the difference between describing the data and thinking with it. The steps are a practical drafting method, not a required format for every research tradition or a claim that software can carry out your analysis for you.

What an analytic memo adds to your notes

An interview summary gives an account of what was discussed. An analytic memo uses selected material to explore a question: why two people describe the same process differently, how a possible concept is taking shape, or what might explain a surprising case.

The Tobacco Control Evaluation Center at UC Davis describes analytic memos as short pieces of analysis written during and after data collection. Its guidance allows their length and focus to follow the developing material. Read its guide to writing analytic memos.

For example, “three people mentioned unclear ownership” is a descriptive starting point. A memo might explore whether the uncertainty concerns the person doing the work, the person approving it, or the point at which responsibility changes. Those explanations would lead to different follow-up questions.

A memo can stay provisional. “Perhaps this depends on the handover boundary” is useful if you show what prompted the idea and what would challenge it. Smoother prose should not make a weak inference look settled.

Before you start

Have the relevant interview notes or transcripts available, with enough source information to find each passage again. Use the recording where permitted and useful. Keep any visual material or documents needed to understand the account alongside it.

You also need a question small enough to explore. “What are our insights?” is difficult to test. “Why do people disagree about who can approve a changed deadline?” gives you a starting point without assuming the answer.

Distinguish your source material from your working interpretation. Keep original accounts intact and create a separate memo, rather than rewriting a transcript until it agrees with your explanation. If you later revise the memo, retain enough history to understand what changed. Our qualitative research audit-trail guide covers that broader traceability task.

Use the terminology and methodological expectations of your study. In an assessed or formal research project, your supervisor, team or research design may specify a different approach. This guide does not replace those requirements.

Step 1: name the question and bound the material

Give the memo a working title that says what you are exploring. “Approval uncertainty after a deadline change” is easier to return to than “Interview insights”. Add a date or version and identify the material you have actually reviewed.

Be specific about scope. “Interviews A–C, passages about revised deadlines” describes a selection. “All project interviews” would be misleading if you only searched for one word and read three results.

Start with a short statement such as: “I am exploring whether people use ‘owner’ to mean responsibility for delivery or authority to approve a change.” The first person here belongs to the hypothetical researcher writing their own memo, not an account of the author’s experience.

If you do not yet have a possible explanation, begin with the puzzle: “A and B appear to describe the same approval route differently.” A good memo can start with a contradiction or an unanswered question.

Step 2: describe the observations before interpreting them

Record a few concise accounts of the relevant material. Keep source references beside them. Use an accurate quotation where the wording matters, and distinguish it from a paraphrase.

For the fictional example in this guide:

  • Interview A: a delivery lead says the project owner agrees deadline changes with the client.

  • Interview B: a coordinator says they update the schedule after the delivery team confirms a new date.

  • Interview C: a sponsor describes approving a revised commercial commitment before the client is told.

These are invented accounts, not extracts from real interviews. They deliberately leave several things unresolved: whether “owner” names the same role in every account, whether all three people mean the same kind of change, and whether they are describing a rule or a particular occasion.

Do not resolve those differences by selecting the most senior participant’s answer or the most polished sentence. First check what each account concerns.

Step 3: propose a relationship worth exploring

Now explain what the observations might mean together. Move beyond repeating them, but keep the inference visible.

A first possibility is that the team lacks a shared approval process. Another is that different participants are describing different decisions: proposing a feasible date, updating an internal plan and approving a commitment to the client.

Write out the connection. “These accounts may differ because the same word, ‘approval’, refers to several points in a sequence” is more useful than the isolated label “misalignment”. It suggests what you need to clarify.

Avoid writing a conclusion that the source cannot carry. The fictional accounts do not establish that there is no process, that communication is poor across the whole organisation, or that any participant has acted outside their authority.

A memo develops a possible explanation. It should not quietly promote that explanation into an organisational finding.

Step 4: bring in a contrasting case or alternative account

Ask what might make the proposed explanation incomplete. Look for a case that behaves differently, a passage that qualifies your interpretation, or a situation where the apparent pattern does not occur.

Suppose a fourth fictional interview describes a small internal rescheduling decision that required no client approval. This could weaken the idea that every difference reflects confusion. It suggests that the type of commitment may matter.


An orange cloth ribbon takes one route through three separate cream rings while a second opening remains unused, representing an alternative explanation that still needs examination.

Do not search only for phrases that confirm your current idea. Read surrounding context and try different terms. Someone may describe authority without using “approve”, or express uncertainty without saying “unclear”.

If no contrasting evidence has been found, say how you looked and what remains unexamined. Absence from a limited search is not evidence that no contrasting case exists.

Step 5: record what you still do not know

Make uncertainty specific enough to act on. “More research needed” does not explain what would change the memo.

For the fictional example, useful uncertainties include:

  • whether participants were describing the same project and period;

  • whether a changed deadline also changed cost or contractual commitments;

  • whether “project owner” refers to a named role or an informal responsibility;

  • whether an account describes usual practice, an exception or a proposed future process;

  • whether an approval rule is available in an existing document.

Also consider your own contribution to the account. Did your question introduce the term “owner”? Did you ask one participant about a formal rule and another about what happens in practice? Could your expectations have made one explanation appear more obvious?

You do not need to answer every uncertainty in the memo. You do need to keep consequential gaps visible, so they do not disappear when someone copies a paragraph into a report.

Step 6: leave a next evidence check

End with an action that could improve, qualify or challenge the explanation. Choose the check because of the uncertainty, rather than collecting more material without a purpose.

For the example, a next check could be to ask participants to walk through one recent deadline change from proposal to client communication. Another could be to compare their account with the relevant approval document, where access is appropriate.

Write down what the check might distinguish. If everyone describes the same sequence for the same kind of change, the terminology explanation becomes more plausible. If accounts still conflict, you have a different problem to investigate.

That is a direction for further analysis, not an experiment result. Do not write as though the check has already happened.

A fictional worked analytic memo

The following memo is invented for this article. It demonstrates a provisional interpretation and its limits, not a recommended finding about a real organisation.

Working title: Different meanings of approval in deadline changes.

Material reviewed: fictional Interviews A–D, limited to their accounts of deadline changes. The researcher would attach exact source locations in a real study.

Question: Are these accounts inconsistent, or are participants describing different decisions in a sequence?

Starting observations: A describes a project owner agreeing changes with the client. B describes updating an internal schedule after delivery confirmation. C describes sponsor approval before a commercial commitment changes. D describes a small internal rescheduling decision with no client approval.

Provisional interpretation: The accounts may concern different boundaries. Delivery feasibility, internal planning and a client-facing commitment could require different people to act. The broad phrase “deadline approval” may be hiding those distinctions. This remains an interpretation, because the accounts have not yet been matched to equivalent changes.

Alternative explanation: There may be inconsistent practice even for the same kind of change. Differences in project, timing or formal authority could also explain the accounts.

Researcher check: The interviews did not use identical follow-ups. One question asked who updates the plan; another asked who can make a promise to a client. This difference may partly explain the answers.

Next evidence check: Revisit a specific recent change with the participants and identify the proposal, feasibility check, schedule update, approval and communication. Compare with the relevant document if available. Preserve any differences instead of rewriting them into agreement.

Current limit: This memo does not establish that the organisation lacks a process or that any role has failed. It identifies a distinction worth investigating.

Notice that the memo explains a relationship, offers another account and leaves a concrete next step. It is longer than a topic label but still much narrower than a final report.

Review the memo against the source material

Before treating a draft as usable, return to the passages behind it. Check whether each source supports the sentence attached to it, whether the wording is a quotation or paraphrase, and whether surrounding context changes the interpretation.

Weeve can help you work with saved research interviews and revisit relevant material through Chat with recordings. Define the recordings you mean, ask a focused question and follow the available source references. Review the underlying accounts yourself.


A genuine Weeve recording-scope menu with synthetic Product planning and Weekly check-in examples, showing that the reviewed source set can be selected.

Genuine Weeve interface with synthetic demonstration content. The menu illustrates source scope, not an automated analytic-memo method or the fictional study above.

A suggested request is: “Within the selected recordings, locate passages about who can approve a deadline change. Keep each account separate. Include surrounding qualifications, source references and any contrasting account. Do not infer that the accounts concern the same type of change.”

The output can help you locate material; it does not validate your interpretation or prove that every relevant passage was found. Search results and generated summaries need checking. Optional external ChatGPT or Claude connections can send relevant text to the selected service, so choose a route appropriate for the material and your information-handling rules.

When to revise, split or finish a memo

Revisit the memo when a new account changes the explanation or when a source check exposes a mistake. Add the new evidence and explain the revision. You do not need to preserve an idea as correct merely because it appeared early in the project.

Split a memo when it contains two different questions with different supporting material. For example, the meaning of approval and the practical delay in receiving approval may belong in separate notes. Link them where the relationship matters.

There is no universal length or number of memos that proves the analysis is complete. Use a working note while it helps you think. When preparing a report, make a separate judgement about which interpretation the full analysis can support and what limitations readers need.

Mistakes that make a memo less useful

Retelling every interview. Return to the memo’s question and select material that helps examine it. Keep the fuller interview summaries separately.

Naming a theme without explaining it. Replace “ownership issues” with the relationship you suspect and the accounts that prompted it.

Mixing source and interpretation. Label your inference and keep a route to the original passage. Our guide to paraphrasing and summarising covers preserving meaning when rewriting source material.

Adding an impressive quotation without context. Check which situation the person meant and whether the passage supports the claim you want to make.

Treating an AI draft as completed analysis. Check the selection, the interpretation and the missing evidence. A source link makes inspection possible; it does not make the attached sentence correct.

A useful analytic memo leaves you with a clearer explanation of your thinking and a better next question. Before closing it, make sure another authorised reader could find the evidence, recognise what remains provisional and understand what you intend to check next.