20 ways to get more value from your research with Claude and Askable
•


Earlier this spring, I hosted a series of dinners for research, design and product leaders, as part of our Insights Exchange series. Many told me the same thing: AI was fast, but they couldn't always trust what it told them.
A few months later, the conversation had shifted. Many had plugged tools like Figma and their own data into Claude, and were asking all of it questions in plain language. My favourite discovery was teams running someone's vague explanation of a problem across their data until the real issue showed up. Say customers are "confused at checkout". Ask Claude what people actually said, and you might find it's fees that only appear on the final screen.
We built Askable's connector for Claude with a couple of uses in mind, like letting anyone on a team ask questions of the research without logging in to Askable. Since then, our customers and our own team have found many more. To celebrate Askable becoming an official Claude connector, here are twenty of our favorites, grouped by role.
These work whatever your role. If you only try three, start here.
New PMs and designers usually learn about customers slowly, through meetings and old decks. On day one, they ask Claude for the five things every new team member should know about your customers, each one backed by real quotes and a link to watch the moment it was said.
It's about as close as you'll get to sitting in on a year of interviews. It also shows them early that on this team, people back up opinions with evidence.
Try asking: "I just joined the payments team. What are the five biggest things our customers struggle with, and who said it?"
Research rarely lives in one place. Some sits in Askable, some in an older repository, some in NPS comments or a strategy document. Claude can hold several sources in one conversation, so you can ask questions across all of them at once.
Upload the strategy paper you're working from, and ask where the evidence agrees with it and where it doesn't. Ask Claude to keep the Askable citations visible, so anyone can see which claims rest on verified participants.
Try asking: "Here is our 2027 strategy. Which of its assumptions does our customer evidence support, and which does it contradict?"
The risk with putting research into an AI tool is that it turns into confident text with no trail behind it. Askable is built to prevent that. Over MCP, Claude receives findings and quotes with participant context, such as first name, age and location, and every answer links back to its source in Askable. And because every Askable participant is verified, you know each quote came from a real person who fits the brief.
Two habits keep it honest. First, when a finding is about to shape a decision, click through and watch the moment it came from. Second, resist turning interviews into fake percentages. "We heard this from people of different ages, in three countries" is truer and more persuasive than a percentage pulled from a dozen interviews, and it holds up when someone asks how many people you actually spoke to.
Try asking: "How consistent is this pattern? Who said it, and does it show up across different ages and countries?"
Use the evidence base to do less repeat work and spend more time on the questions that matter.
The answer to your next research question may already be in the evidence base. Questions get answered, the answer ends up in an old deck, and a new study gets briefed for something the team already knows.
Before you write the next brief, ask Claude what the evidence base already says. Sometimes the answer is there. More often you get part of an answer, which is still useful because it shows what's left to find out. And if nothing comes back, you have a clear signal the question needs fresh evidence.
It saves money as well as time. Every question you answer from the evidence you already have is a study you don't need to fund, which leaves budget for the questions that are new.
Try asking: "Have we ever heard customers talk about how they choose between plans? What do we know, and what's missing?"
Once you know what the evidence doesn't cover, you're halfway to the next study. Ask Claude to turn those open questions into a draft discussion guide: the open questions, grouped by theme, written to avoid leading the participant.
You still set up and run the study in Askable, and you still edit the guide. But you start from a list of questions you can't answer yet, instead of re-asking the ones you already can.
Try asking: "Based on what we don't know about onboarding, draft eight open-ended interview questions for our next study."
You ran a study before the redesign and another after. Or the same study with two different segments. Comparing them usually means two tabs, two sets of notes, and a lot of squinting.
Give Claude both studies and ask what changed: which themes got stronger, which disappeared and which are new. That saves half a day of side-by-side reading and gets you sooner to the more interesting question, which is why things changed.
Try asking: "Compare our onboarding study from March with the one from September. What do people complain about now that they didn't before?"
You remember a participant said something perfect, months ago, and now you need it for slide 4 in twenty minutes.
Describe what they said, roughly, and let Claude find it. You get the quote, the participant's first name, age and location, and a link to the moment in the session. It takes seconds, and you get their exact words.
Try asking: "Someone said cancelling should be as easy as signing up. Can you find it?"
A static deck gets skimmed once. Ask Claude to build an interactive report from the evidence instead: themes people can click into, the quotes behind each one, and a link back to the moment each was said.
Stakeholders explore at their own pace and go straight to the part that matters to them. And you get back the hours you'd have spent formatting slides.
Try asking: "Turn what travellers told us about hidden fees into an interactive report I can share, with themes, quotes and source links."
Check your plans against what customers said before the sprint starts.
Some ideas sound obvious in the meeting and are halfway onto the roadmap before anyone has asked the people they're for.
Paste the idea into Claude and ask the evidence to sort it into three piles: what supports it, what pushes against it, and what nobody has asked yet. The most useful part is the pushback. When we tested a rewards idea against travel interviews, the evidence pulled both ways. Some people said expiring credit pushed them to book a trip. Others said expiry dates and thresholds made them feel they had to work for a reward, so they took the cheaper option instead. Better to have that argument before you build it. It also helps the researcher in the room: pushing back on a leader's idea is easier when you're quoting customers rather than giving your own opinion.
Try asking: "We're thinking of adding streaks to the app. Based on what people have told us, should we build it, test it first, or drop it?"
Most PRDs rest on assumptions about behaviour that nobody has tested.
Paste the PRD into Claude and ask it to list each assumption, then search the evidence for each one. You get back a simple audit: supported, contradicted, or untested. The untested ones become your next research questions, and the contradicted ones are worth fixing before anyone writes code.
Try asking: "List every assumption in this PRD about user behaviour. For each one, tell me whether our customer evidence supports it, contradicts it, or doesn't cover it."
Your study keeps surfacing the same complaint. The next question is whether it's about your product, or whether everyone in your market says the same thing.
Ask Claude to run the question against your own studies, then against Askable's ongoing interviews across your industry, and put the answers side by side. If only your customers raise it, it's yours to fix. If the whole market does, it might be your chance to stand out.
Try asking: "Our customers keep mentioning fees. Search our studies, then the finance industry stream. Is this about us, or the whole market?"
Design from what customers said, in the words they used.
A brief that paraphrases the research loses the customer's own phrasing, which is often the most useful part.
Pull the finding into Claude, then use it to brief a builder like Figma Make or Lovable. The prompt carries the exact phrases participants used, so the first version is shaped by the evidence. Then put the prototype in front of participants on Askable, and the new evidence lands back in the same base.
Try asking: "Using what customers said about comparing flight prices, write a Figma Make prompt for a search results screen that fixes their biggest frustration."
Most journey maps are well researched in some stages, like onboarding, and lighter in others, like renewal.
Take each stage and ask Claude what customers have said about it. Where there is evidence, the map gets real quotes. Where there is none, you have found a blind spot, and you can say so plainly instead of papering over it.
Try asking: "Here are the six stages of our customer journey. For each one, what have customers told us, and which stages have no evidence at all?"
Digital twins and synthetic users are usually built from assumptions or general data. Build yours from what verified people in a segment actually told you, then talk to it while you design.
Use it two ways. In conversation, ask it questions and get answers in participants' own words, with the quote and source behind each one. As a survey, give it your draft questions and see how the segment would likely answer, before you field the real thing. Either way, you can see where the twin is speaking from evidence and where the evidence runs out. Where it runs out, you know what to research next.
Try asking: "Build a digital twin of frequent UK travellers from our interviews. Answering as them, what would make you book direct instead of through a comparison site? Cite the quote behind each answer."
Get the customer's view on a decision without waiting for a readout.
Leaders usually meet customer evidence second-hand, in a quarterly readout or on a slide someone else chose. By then, the decision it could have shaped has often been made.
With Askable connected, you can ask Claude directly and get a summary, the participant quotes behind it, and a link to watch each moment. It's the simplest way to stay close to customers without adding a meeting to anyone's calendar.
Try asking: "What are customers saying about our pricing this quarter? Show me the quotes."
Askable shares each participant's country alongside their quotes, and Claude can filter the evidence by country. That makes it easy to ask whether a pattern holds across markets or only shows up in one.
Ask how people in Australia, the UK and the US talk about the same problem, and you often find the same frustration in three different accents. Sometimes you find it in only one, which is just as useful when you're deciding where to launch first.
Try asking: "Do customers in the UK describe the booking problem differently from customers in the US? Show me examples from each."
Before a big decision, most leaders hear the team's view. It's worth hearing the customer's too.
Ask Claude for a one-page brief on the topic, built from the evidence, with every claim linked back to where it came from. You walk into the room knowing which positions have customers behind them and which are still hunches.
Try asking: "I have a decision on our onboarding redesign tomorrow. Give me a one-page brief on what customers have told us, with three sourced quotes."
Marketing teams rarely see the research, and there's a lot in it they could use.
The best headline for a landing page or ad is often a sentence a customer already said.
Ask Claude for the phrases customers use to describe the problem your product solves, then draft copy from them. You're starting from phrases a real customer used, not ones a copywriter guessed at. Just get the right permissions before quoting anyone directly in public.
Try asking: "How do customers describe the moment they decided to switch banks? Give me ten phrases I could build a campaign around."
Research often turns up things it wasn't looking for. We searched a weekly run of interviews about personal finance for anything on Black Friday. There was no study on it, but discount-driven buying came up anyway, in conversations about impulse spending and even scams. One person broke their own buying rules for a deal that turned out to be too good to be true.
Ask Claude about a seasonal moment, a news story or a trend, and see what the evidence already holds. It's a quick way to give a timely campaign or a point of view real human backing.
Try asking: "Is there anything in our research about how people feel about Black Friday sales, even if it wasn't the topic of the study?"
Before you stand up and make a claim about your customers, check it. We prepared our last webinar by asking Claude to find evidence for each talking point, and to flag the ones that don't hold up.
The result is a script where every claim is backed by a quote you can click on. And you find the weak points while there's still time to cut them.
Try asking: "Here are my five talking points for Thursday's webinar. Find the strongest customer evidence for each, and tell me which ones are weak."