🎤 Community Update
We've been quiet for a few weeks, so this one is worth the wait: two brand-new studies that put synthetic users side by side with real people, a practical way to measure how exposed your own week is to AI, and 15 research roles posted in the last fortnight.
Research Lunch Club's November matches go out on Saturday 31 October. Join before then and you'll be in the round with 3,700+ researchers across 17 cities: no cold DMs, no scheduling back-and-forth, just lunch with a small group of researchers who get what you do.
Join Research Lunch Club here: https://researchlunchclub.com/join
📰 Article Picks
🤖 Synthetic Users vs. Human Respondents: Same Conclusion, Less Variation by Rachel Banawa for Nielsen Norman Group | Banawa took one open-ended question from an earlier NN/g survey, in which 604 tech and design professionals named the field that best describes research-led product work, and put it to 120 synthetic respondents generated on the Synthetic Users platform. Both datasets reached the same headline conclusion: UX is still the dominant name for the work. Underneath, the humans used more than 50 distinct labels while the synthetic respondents collapsed onto about a dozen, and none of the unusual or specialised terms from the human data appeared, even though only 20% of the synthetic sample were UX professionals against 67% of the humans. Banawa calls this mode collapse, where synthetic data reproduces the most common answer and drops the long tail, and she also flags that the platform quietly reworded the question, a confound researchers can't control. Her practical line: if you only need the most common response, synthetic data may be enough; if you need unexpected language, emerging themes or ideas for further study, you need people.
📈 Digital Twins Systematically Overestimated Business Software SUS Scores by Lucas Plabst, Jeff Sauro and Jim Lewis | MeasuringU built a digital twin for each of 289 real users of six business products, from GitHub to Slack, using a persona generated from that person's own survey answers with their SUS ratings held out, then asked each twin to complete the SUS the way its human had. The twins scored every product higher than the humans did, by about 10 points on average and by around 16 points for people whose real scores sat between 50 and 70; GitHub moved from a D to a B and ZipRecruiter from a C to an A. They did get the rank order of products mostly right (correlations of .94 to .95) and tracked individuals reasonably well (around .8), and giving them more data made no difference. The warning is in the shape: twins gave a perfect 100 almost four times as often as humans, so the error is largest for exactly the mid-scoring products that have the most room to improve.
💼 AI doesn't care about your job title by Patrick Neeman | Neeman argues that AI exposure runs through the tasks in your week, not the title on your badge, and sets out a five-step audit with a ready-made Claude prompt for each step. The core is his spec-and-check test: a task is exposed if a finished version can be described in two sentences and checked faster than it took to make, so twenty variations of an empty state pass while deciding which of three problems to fund this quarter fails. He grounds it in research rather than vibes, including Stanford's finding that employment for 22- to 25-year-olds in exposed occupations sits 19% below trend while experienced workers show no comparable gap, and an Anthropic study of Claude Code sessions in which knowing the problem domain, not the tool, did most to double success rates. The action items are concrete: score your week, move one hour a week from your most exposed task towards decisions, stakeholders or time with users, and rebuild your strongest case study around the decisions you made.
🎙 Podcast Pick
💼 Job Picks
Are you a hiring manager looking to share research opportunities with a global network of over 7,500 researchers in more than 60 countries? We'd love to hear from you! Just drop us an email ✉️.
Europe & UK
North & South America
👀 Sponsor us?
Our weekly newsletter is a fantastic way to connect with a global audience! With over 7,500 researchers tuning in from more than +65 countries, it's a great chance for sponsors to boost their brand visibility and make a real impact. - 💬 let’s talk!
👋 About us
We’re People of Research & Research Lunch Club - a global hybrid community uniting researchers across the industries, from UX Research to Behavioural Science. We connect, collab and grow. Created by @faysel.


