Your user research is a focus group of people who agreed to be in a focus group.
You recruited them, paid them, and waited three weeks for them to lie to you politely. Then you wrote a 40-page report nobody read, and shipped the thing anyway.
I have done this. You have probably done this. Most product teams I talk to are still doing this, and they are proud of it. They call it being user-centric.
I want to make an uncomfortable case: most of what passes for user research today measures the wrong thing, arrives too late to matter, and survives mainly because the alternative used to be impossible. The alternative is no longer impossible. So let’s talk about the funeral.
What your research actually measures
Picture the last study you ran. You recruited maybe eight people. You screened them so they fit a profile. You scheduled them, paid them, and put them in front of your product while someone watched and took notes.
Now ask what you actually observed.
You observed how eight people behave when they know they are being watched, compensated, and asked to produce opinions on demand. That is not a neutral window into your user. It is a performance of a user, shaped by every incentive in the room.
This is not a knock on the people. It is the structure. The moment you recruit, pay, and observe, you introduce three problems at once:
Recruiting bias. The people who say yes to a research session are not a random slice of your market. They are the people with time, willingness, and a certain comfort being studied. Your hardest, busiest, most valuable users are the ones who never show up to your study.
The observer effect. People behave differently when watched. They narrate, they rationalize, they try to be helpful. They tell you what they think you want to hear because a human is sitting right there and politeness is a stronger instinct than honesty.
Tiny samples. Eight people. Sometimes five. We make roadmap decisions worth months of engineering on a sample size that would embarrass a high school statistics class.
None of this is secret. Every good researcher knows it. The defense is usually that qualitative research is not meant to be statistically representative, it is meant to surface insight. Fair. But insight that takes three weeks to surface, on a biased sample, in an artificial setting, is a very expensive way to be directionally unsure.
The lag is the real killer
Even if you accept the bias, there is a problem you cannot accept away: timing.
Real user research arrives after you have built something. You design the flow, build the flow, recruit, schedule, run sessions, synthesize, and present. By the time the report lands, the decision that mattered is already made, the engineering is already spent, and the team has already moved on emotionally. The report becomes a document you cite to justify what you were going to do anyway.
The whole point of research is to change a decision before it is expensive. Traditional research consistently shows up after the expense. That is the part that should bother you most.
I am not anti-research. I am anti this version of it.
Here is where I want to be precise, because the funeral framing is deliberately provocative and I do not want to hide behind it.
I am not saying understanding users is dead. Understanding users is the entire job. I am saying the specific apparatus we built to do it, recruit, pay, observe a handful of people, write it up, is slow, biased, unscalable, and badly matched to how fast teams now need to move.
What I am against is guessing dressed up as rigor. A study of eight recruited strangers is not rigor. It feels like rigor because it is effortful and it produces a document. But effort is not evidence, and a document is not an outcome.
What replaces it
If the problem with research is that it is slow, biased, and tiny, then the fix has to be fast, grounded, and large. That is what simulated user studies are.
At Marketrix we run user studies with simulated users instead of recruited ones. The mechanics, in plain terms:
Ground personas in the actual product. We point the platform at your product and generate personas from what is really there, not from a workshop full of sticky notes. These are not generic archetypes. They are shaped by the flows, the language, and the decisions your product actually asks people to make.
Run real tasks, at the UI layer. Each persona attempts real journeys the way a person would, clicking through the real interface. No SDK, no instrumentation, no code change required.
Run hundreds in parallel, before launch. This is the part that breaks the old constraints. You are not waiting three weeks for eight people. You are running a large, diverse set of personas through the flow in minutes, and you can do it before the feature ships, or before a single real user has seen the change.
Get back what a study gives you. Notable quotes, emotional reactions, friction scores, where each persona gets confused, where each one quits. The same kind of qualitative texture, without the recruiting, the incentives, or the wait.
Two proof points so this does not read as theory. We ran a full UX study on HubSpot’s buying flow with zero human participants, and it surfaced a pricing-journey insight that most teams would only catch after losing the deal. And Cut+Dry, our first paid contract, ran more than 900 test cases through this approach. Real product, real flows, no recruited panel.
The obvious objection
A simulated user is not a real user, so how can this be trusted?
It is the right question, and the honest answer is that simulation does not eliminate real users, it changes when you need them and what you need them for. You simulate first to catch the obvious friction, test variants, and kill bad ideas cheaply, before you spend the scarce, expensive, high-signal time of actual humans. Real feedback still matters. It just stops being the thing you wait three weeks for to learn something you could have known on day one.
This is the same shift cloud forced on infrastructure. We stopped provisioning servers months ahead on a guess and started spinning up capacity on demand. Simulation does that for product understanding. You stop guessing months ahead and start testing the journey the moment you can describe it.
Time to stop guessing
The next era of product teams will not run a study, wait, and hope the report shows up before the decision is locked. They will simulate the journey first, across hundreds of personas, in the time it takes to read the brief. The teams that do this will ship with more conviction, waste less engineering on flows that were always going to fail, and spend their real-user time where it actually counts.
The rest will keep recruiting eight people, paying them, and writing the report nobody reads.
User research as we built it is dead. The understanding it was supposed to deliver is more alive than ever. It just runs on a different engine now.
A million users before your first user.
If you are building a product and want to see a simulated user study run on something real, reach out at irosha at marketrix dot ai, or find me on LinkedIn. And if this resonated, subscribe below for more on simulation, product, and building software that actually works for people.