research methods
Triangulation
What is triangulation in research?
Triangulation is checking a finding against more than one source, method or perspective, and trusting it most when they converge. Every method has its own blind spots and biases - analytics can't tell you why, interviews capture what people say not what they do - so a conclusion that holds up across several of them is far more trustworthy than one resting on a single, fallible view.
Also known as: triangulation, mixed methods, converging evidence, method triangulation
Prefer to watch?
Watch the recap 0:52
The demo
A finding to test: “users abandon at the payment step.” Switch on your sources of evidence one at a time, and watch how much you should trust it change.
What this demo shows (text version)
A candidate finding - "users abandon at the payment step" - can be checked against three sources: analytics (a measured drop-off), interviews (people saying so), and a usability test (watching people quit there). As you switch sources on, a confidence meter rises. With only one source, confidence is low and the demo flags the method's blind spot (analytics can't say why; an interview could be one vocal person; five users is a small sample). With all three converging, confidence is high.
That's triangulation: because each method fails in different ways, a finding that holds across several of them is far more trustworthy than one resting on a single view. Combine quant and qual, multiple sources and times; trust convergence; and when sources conflict, treat the disagreement as the most interesting question in the study rather than noise to dismiss.
No single method is trustworthy on its own - each has characteristic blind spots, so cross-check findings across sources (analytics + interviews + usability testing, quant + qual). When they converge, believe it; when they conflict, you've found something important to dig into, not an annoyance to ignore. Triangulation is how you tell a real insight from an artefact of one method's particular bias.
One source said "users abandon at payment" - but was that a real problem or just a quirk of how that method sees the world? Add a second that agrees, then a third, and the confidence climbs from "maybe" to "act on it". That convergence is triangulation: not more data for its own sake, but different vantage points agreeing on the same truth.
The logic is that methods fail in different ways, so their errors don't line up. Analytics tells you what and how many but never why, and only counts people who got far enough to be measured; interviews and surveys capture stated attitudes, skewed by memory and social desirability; usability tests show real behaviour but on small, artificial samples. Each is partial - which is exactly why agreement across them means something.
So combine across methods (quantitative and qualitative), across sources (different participants, teams, datasets), and across time. When they converge on the same finding, your confidence is well-founded. When a single method screams something the others don't, treat it as a hypothesis to verify, not a fact - it may be that method's particular blind spot talking.
Conflicts are a feature, not a failure. When analytics says people use a feature heavily but interviews say they hate it, you haven't got bad data - you've got a question worth answering (maybe they use it because they're forced to). Triangulation turns "the numbers say one thing, users say another" from a contradiction to resolve into the most interesting finding in the study.