How to verify the fidelity of synthetic users, ensuring their results are consistent, objective and useful in UX projects.
In recent months, synthetic users — sometimes called "clones" — have gained prominence in UX research and product design. Their promise is compelling: simulating conversations, reactions and decisions from real segments without relying on the constant availability of human participants.
At Interactius, we have incorporated this tool into our processes with a key premise: to be credible and useful, it must be built and validated with the highest methodological rigour.
The Interactius Approach: Internal Validation as a Guarantee
At Interactius we apply what is known as internal validation. This means verifying that the clone faithfully represents reality using the same data source from which it was created, avoiding the introduction of external variables that could distort the result.
In practice, if we build a clone from 20 real interviews, we validate it by comparing its responses and behaviours with those of those very same people in usability tests or concept testing. This allows us to measure, in a controlled way, the extent to which the model reflects the real pattern that gave rise to it.
Methodological Foundations
This principle aligns with recognised frameworks in model verification and validation (Naylor & Finger, 1967), which recommend:
- Face validity: the model must be recognisable as realistic by experts and users.
- Assumption validation: documenting and verifying the accuracy of the premises used.
- Output validation: verifying that the model's responses match real data in the same context.
In addition, we adopt criteria suggested in synthetic data validation: objectivity, consistency, realism and utility. By validating against the same original sample, we achieve consistency and can objectively demonstrate the fidelity of the clone.
Advantages of This Approach
- Methodological consistency: we measure the clone's ability to accurately reproduce patterns from the group that originated it.
- Bias reduction: without external samples, the risk of confusing real differences with model errors is minimised.
- Transparency: we can clearly explain the process to our clients and support the validity of the results.
As the Nielsen Norman Group notes, synthetic users should complement, not replace, research with real people. At Interactius we use this internal validation to accelerate hypotheses and early-stage testing, always anchored to real human data.
How We Do It at Interactius
-
Clone generation from real interviews and pattern analysis.
-
Parallel test design for both humans and the clone, with equivalent tasks and conditions.
-
Controlled execution and recording of responses in both cases.
-
Systematic comparison to measure alignment on key criteria (messages understood, barriers detected, motivations).
-
Documentation and communication with a clear report and direct evidence.
Benefits and Limitations
- Benefits: time savings during exploratory phases, early detection of issues, reuse of the clone in new projects within the same segment.
- Limitations: it does not replace direct empathy with real users, it may amplify biases present in the original sample, and its validity is specific to the context for which it was created.
Conclusion
Internal validation is a robust framework for synthetic users to gain genuine credibility and utility in UX. At Interactius, this method enables us to deliver clones that not only save time and resources, but are also built on solid human foundations and validated transparently.
In this way, we offer our clients a reliable tool to accelerate design and innovation, without losing the connection to reality that gives meaning to any user experience.
Request Information About Our Synthetic Users Service
If you would like to learn more about how synthetic users can benefit your research and design, fill in the form to get more information.
