Turn two hundred observations
into findings that hold up.

Break every session into atomic observations, one card each, with the source attached: the recording link, the transcript quote, or the screenshot. Spread the cards on a canvas and group the ones that describe the same underlying thing, ignoring which participant they came from. Name each cluster only after it forms, so the theme comes from the data rather than from your expectations. Then read the clusters critically: a theme spanning nine participants is a finding, a theme built from one vocal participant is an anecdote. Write the findings as their own cards and connect each one to its cluster. The result is a wall where every claim can be walked back to the evidence, which is what separates findings from opinions.

They do different jobs, and plenty of teams would use both. Dovetail is a research repository: it transcribes recordings automatically, supports structured tagging pipelines, and makes years of studies searchable, which matters enormously at scale. Storyflow does not transcribe and is not trying to be your archive. It is the synthesis wall: the spatial surface where observations get affinity-grouped into themes by dragging, where findings connect visibly to their evidence, and where personas and journey maps are built next to the raw research. If your pain is processing and retrieving sessions, choose Dovetail. If your pain is that collection is fine but the findings feel thin and untraceable, that is the synthesis gap, and it is the gap Storyflow is built for.

No, and that is deliberate. Storyflow's AI is structural: it lays out a research board from your study description so you are not starting from a blank wall, it can suggest how an overloaded cluster might split, and it can draft a frame for the findings section. Because it sees the whole canvas, its suggestions fit the study you actually ran. What it does not do is decide what your evidence supports. Synthesis is a judgment call about which observations belong together and which claims the data can carry, and a finding a machine produced is a finding a stakeholder can dismiss. The AI accelerates the arrangement; the conclusions stay yours, which is precisely why they hold up in the readout.

Not during early access. Storyflow is paid-only right now, with plans starting at $9.99 a month, and a Free plan for solo use launches before the end of 2026. Anyone a paid member invites to a board joins free today, so teammates and stakeholders can collaborate on your research wall without paying. If you need a free option this minute, Notion's free tier is genuinely capable as a research wiki if you are willing to build the structure yourself, and physical sticky notes on a wall remain a perfectly honest way to run a first affinity mapping session.

Attach the source at the observation level, not the findings level. Every observation card carries its own recording link, transcript quote, or screenshot from the moment you create it, so evidence travels with the card no matter how many times you regroup it. Findings then connect to clusters, and clusters are made of sourced cards, which gives you an unbroken chain: recommendation to finding to theme to observation to the exact minute of the session. This matters most six weeks later, when someone questions a decision and nobody remembers session four. On a board, you do not defend the finding from memory. You walk the connection back and let the participant make the argument.

Yes, and it is one of the main reasons to synthesize on a canvas rather than in slides. Build the persona on the same board as the interview observations it came from, and every trait can point to the cluster that produced it. Lay the journey map beside the usability test evidence, and every pain point traces to the sessions where participants actually hit it. Artifacts made this way are harder to dismiss and easier to update: when the next study lands, you can see immediately which parts of the persona the new evidence supports and which parts it breaks. An artifact that left its evidence behind in another tool tends to drift into fiction.

Send a view-only link. Stakeholders open the whole wall in a browser with no account: the findings, the themes behind them, and the quotes and screenshots behind those. That evidence trail is what makes research persuasive, because a stakeholder who can see nine participants hitting the same wall stops arguing with the recommendation. Comments let them ask questions on the board itself rather than in a meeting three days later. For readouts that have to live in a deck or a document, export the board as a high-quality image or PDF. The link usually works better than the deck, though, because it invites people to follow the reasoning instead of just receiving the summary.

Yes. Invite teammates to the canvas and they join free, so synthesis becomes a group activity rather than one researcher disappearing for a week and returning with conclusions. Group synthesis is genuinely better practice: when a designer, a PM, and a researcher cluster the same observations, the disagreements about where a card belongs surface assumptions that would otherwise stay buried until the roadmap argument. It also builds buy-in cheaply, because people rarely dismiss findings they helped construct. Stakeholders who only need to see the result get a view-only link instead, which keeps the wall itself in the hands of the people doing the work.

Five layers. Raw observations, one card per thing a participant said or did, kept atomic so they can regroup freely. Source evidence attached to each card: recording links, transcript quotes, screenshots. Themes, the named affinity clusters the observations form. Findings, written as explicit claims and connected to the clusters that support them. And the artifacts the research feeds: personas, empathy maps, journey maps, and recommendations, built beside their evidence. Frame each study into its own section so you can also synthesize across studies over time. What should not go on the board is anything you cannot source. An observation without evidence is an opinion wearing a lanyard, and the board is more useful when it refuses them.