Research decks fill easily with memorable sentences. A customer says the process is “a nightmare.” Another wants “everything in one place.” Someone asks for automation. The team highlights the words, groups them into themes, and feels closer to the answer.
Without a product decision, those quotes never leave the slide. What does “a nightmare” actually describe? Which force sits behind “everything in one place”? What would the product do differently because of it? Getting there means reading past the words to the forces underneath.
Quotes preserve the customer’s voice but don’t interpret the forces behind it
“I want a dashboard” may mean “I’m tired of being surprised in leadership meetings.” “Automate this” may mean “I don’t want to be the person who remembers every follow-up.” “Everything in one place” may mean “I can’t defend a decision when the evidence is scattered.”
The stated solution is rarely the deepest usable finding. People may ask for a summary, an agent, or an automation when what they really need is confidence, control, relief, a second set of eyes, or a safer way to act.
A strong synthesis moves through seven steps
The chain starts with what the person actually said and ends with an observable reaction that would confirm the decision helped. Between the quote and the decision sit the situation that produced it, the force it reveals—Push, Pull, Habit, or Anxiety—and the progress the person was trying to make, functionally, emotionally, and socially. Push and Pull move people from the status quo—the current workflow—toward a brighter future; Habit and Anxiety hold them in the status quo.
Sequence map
Follow the chain
Carry a quote from what the person said to the reaction that would confirm the decision helped.
- Quote: What did the person actually say?
- Situation: What was happening when this became important?
- Force: Which force does the statement reveal: Push, Pull, Habit, or Anxiety?
- Job: What progress was the person trying to make—functionally, emotionally, and socially?
- Decision: What product choice changes because of that progress?
- Behavior: How should the system work, respond, or recover differently?
- Test: What observable reaction would confirm that the decision helped?
Skip the middle and teams jump from customer language to feature requests. The output looks customer-centered while the reasoning remains unexamined. “Users want bulk export” is easy to put on a roadmap. “Users are switching because the experience doesn’t produce confidence at the moment they need to present the output to leadership” takes more thinking, more design judgment, and more willingness to reconsider what the product is actually for.
One quote can change the product’s evidence model and confidence behavior
A customer-success lead needs to explain an AI recommendation in a weekly risk review.
Example
Follow the chain: a weekly risk review
- Quote: “I need the AI to explain why it picked this account.”
- Situation: a customer-success lead is preparing for a weekly risk review and expects leadership to challenge the prioritization.
- Force: Pull draws the user toward the recommendation; Anxiety holds the user in the status quo because accepting it creates social accountability.
- Job: decide where to intervene first and be able to defend that decision in the room.
- Decision: don’t show a bare risk score. Surface the signals, source dates, competing evidence, and the action the score supports.
- Behavior: when evidence is weak or contradictory, switch from “recommended action” to “review needed” rather than presenting false certainty.
- Test: in a realistic review scenario, can the user choose an account and explain the decision without reopening three other tools?
The words asked for an explanation. The job showed what the explanation was for: deciding where to intervene first and defending that decision in the room. The recommendation creates Pull, but Anxiety holds the lead in the status quo because accepting it creates social accountability. In high-stakes work, the user is often not only deciding for themselves. Good explainability helps the user act now and defend later. So the decision changed what the product shows, and the behavior changed what it does when it can’t back the claim.
Differences between interviews often reveal conditions
Synthesis shouldn’t sand every interview into one average user.
Experienced users may want direct control while occasional users need a prepared recommendation. High-stakes cases may require evidence that routine cases don’t. One team may treat a field as authoritative while another knows it’s stale.
Instead of deciding which participant is “right,” ask what changed the job. Role, stakes, frequency, reversibility, available context, or social exposure may explain the split.
A decision ledger shares the reasoning, not only the final screen
The chain produces a decision. The ledger writes down how the team reached it, including the alternatives it rejected.
Fill-in table
Decision ledger
Document each important research-backed choice.
| Field | What to write down |
|---|---|
| Situation and trigger | What was happening when this became important |
| Evidence and representative quotes | Observed or reported customer evidence, and the quotes that represent it—not the team’s theories |
| Forces at work | Which of Push, Pull, Habit, or Anxiety the evidence reveals, and why |
| Job interpretation | The progress people were trying to make—functionally, emotionally, and socially |
| Decision made | The product choice that changes because of that progress |
| Alternatives rejected and why | The options the team didn’t choose, and why |
| Behavior or component affected | How the system should work, respond, or recover differently, or which component changes |
| Test and success signal | The observable reaction that would confirm the decision helped |
| What would change this interpretation | The evidence that would send the team back to this decision |
This gives designers, product managers, engineers, and AI agents access to the reasoning—not only the final screen.
When conditions change, the team can revisit the decision without restarting the research from memory.
The handoff test
The handoff test checks whether the chain holds for quotes the team already has.
Pick five quotes from the last study. For each one:
0 of 5 done
If the chain breaks after the quote, the research created empathy but not direction. The goal isn’t to collect language users recognize. It’s to turn their reality into product judgment the team can act on.
