Customers say a feature is essential, yet the usage report shows that few open it. The tempting response is to decide which source is more trustworthy. A more useful investigation asks whether the two sources are measuring the same thing: perhaps one administrator uses the feature on behalf of an entire team, or perhaps “essential” means available when an occasional emergency occurs.

Triangulation examines a research question through multiple methods, sources, investigators or perspectives. It can strengthen an explanation, reveal different aspects of an experience or expose a disagreement that deserves further work. Agreement is useful, but it is not the only worthwhile outcome.

Choose evidence with a clear role

In the fictional feature example, event data could establish recorded use, interviews could explore the circumstances in which the feature matters, and observation could show how work is shared between colleagues. Each contributes something specific rather than serving as another vote for a conclusion chosen in advance.

Nielsen Norman Group’s guide to triangulation discusses combining research approaches to address their limitations. Multiple sources are more informative when their coverage and weaknesses are understood. Three dashboards built from the same tracking error do not provide three independent confirmations.

Triangulation can also involve different populations, time periods or analysts. The appropriate combination depends on the uncertainty and the stakes; a straightforward wording problem does not automatically justify three studies.

Compare the claims, not just the headlines

Before interpreting agreement or disagreement, check the audience, time period and question behind each finding. A satisfaction survey among paying administrators and observations of new end users may describe genuinely different experiences. Treating them as contradictory without that context creates a problem the evidence never posed.

Analyse each strand competently and preserve its limits. A survey percentage requires attention to recruitment and missing responses; an interview interpretation needs support in the accounts. Combining them does not cancel weaknesses in either.

When the strands address different aspects of a question, explain how they fit together. When they address the same claim and disagree, examine plausible reasons and identify what remains unresolved. The report should not hide inconvenient evidence in an appendix while presenting agreement as the main story.

Let the comparison change the conclusion

Suppose the investigation shows that a rarely used feature enables an administrator to complete a critical monthly task for many colleagues. The appropriate conclusion may concern concentrated value, rather than either widespread adoption or an unnecessary feature. That changes the product decision as well as the description of usage.

Mixed-methods research often uses this kind of integration, although triangulation can occur within qualitative or quantitative work too. Neither guarantees certainty or establishes causation by the number of sources involved.

The useful output is an explanation of what each source supports, how the evidence relates and where confidence remains limited. Sometimes that is enough to act; sometimes the disagreement defines a narrower and more valuable next study.