We compared synthetic users with a real sample and achieved a 93% match rate. A new way to validate ideas and test hypotheses — and it works.
In the first part we explored how synthetic users created with AI allow you to anticipate behaviours, motivations and needs when you don't yet have access to real users. In this second instalment, we go one step further: can these profiles accurately replicate the results of a real study?
Yes. And not just "more or less". We're talking about an average match rate of 93%.
The comparison: 20 synthetic users vs. 150 real ones
We started with a structured survey, with more than 30 questions, applied in two contexts:
- To 150 real people in the process of adapting to a new environment (anonymous online survey).
- To 20 synthetic users, created from in-depth qualitative interviews with 20 real people.
Context is key.
These synthetic users were not generated in a generic way: they were specifically designed for a concrete digital product design context. Each one responds to a realistic scenario related to the challenges, expectations and experiences of people interacting for the first time with digital services in a new environment.
The profiles were modelled from real accounts, with a clear focus: exploring barriers, needs and decisions in an onboarding flow for new services. This meant that, when answering the survey, the synthetic users' decisions were consistent with the framework being designed.
Results: a 93% match
We compared responses by thematic blocks, measuring how closely the aggregated answers aligned between both groups:
What does this mean for design, product and innovation teams?
- You can use synthetic users to test surveys, product ideas, onboarding flows or communication strategies, before having access to real users.
- They are useful for validating hypotheses and refining criteria in the early stages of the research process.
- If they are built well, based on qualitative and structured data — as in this case, where each profile started from a real interviewed person — the results are highly realistic and representative.
- And most importantly: if you design them with the specific context of the product or service you are building in mind, synthetic users can behave with astonishing precision.
What if you had them available 24/7?
Imagine this:
- Profiles designed for your context.
- Actionable at any time.
- No downtime or loss of focus.
Synthetic User-as-a-Service
Synthetic User-as-a-Service is not science fiction.
It is a new way of designing with AI without losing the human touch. It is researching before researching.
Synthetic User-as-a-Service is a solution that enables you to work with synthetic users created from real data, ready to answer surveys, test flows or take part in simulations. They are not invented: each profile is built from in-depth qualitative interviews and is aligned to the context of the product you are building.
Because designing well does not always depend on more data, but on the right data, at the right moment. Click here to view the Case study: Montserrat and the Apple Watch.
At Interactius we are already helping teams explore this new frontier.
Are you ready to make the change?
