Is the product broken?
It is the obvious question. When a start rate is near zero, the standard checklist runs: bad UX, wrong audience, broken email flow, unclear value proposition. We have worked through each of these — and the system is early enough that some issues were real. Routing bugs. Email delivery gaps. Infrastructure problems we diagnosed and fixed over the pilot period.
The technical issues are being addressed. But there is a pattern underneath them that technical fixes alone will not resolve. We have 879 people who recognized themselves in the description of cognitive overload and still did not begin the assessment. 29 opened the app. The funnel is reaching people. People are not starting.
Some of that is product. Some of it may be something else worth looking at.
The gap worth naming
The Decision Load Index was designed to measure cognitive overload — specifically, the accumulated weight of unresolved decisions that sits in a person's working memory and degrades their capacity to function clearly.
The people who signed up to take it are, by definition, people who recognized something in the description. They saw themselves in the framing. They wanted to know their number.
Then they did not start.
Consider what it actually takes to begin a structured self-assessment when you are cognitively overloaded. You need a contiguous block of attention. You need to believe the output will be worth the cost of engagement. You need to not have seventeen other things competing for that same window. And you need to make a decision — the decision to start — at the exact moment your decision-making capacity is most depleted.
This pattern is consistent across the signup cohort. Whether product issues explain all of it or only part of it, the underlying dynamic is real: the people who most need a decision-load measurement face genuine friction at the moment of initiation.
The target user and the user who finds initiation easiest are not the same person.
What happens at the "Start Assessment" button
We do not have session recordings. But we can reason from what we know.
The moment a person arrives at the assessment entry point, they encounter a discrete choice: start now, or not. For someone with low cognitive load, this is a trivial decision. For someone carrying 40 unresolved items in working memory — the exact person our tool is designed for — it is one more demand stacked on top of all the others.
There is a body of research on what is sometimes called ego depletion, though the precise mechanisms remain debated. The general pattern is well-supported: repeated decision-making draws on a finite cognitive resource, and when that resource is taxed, people avoid new decisions. They defer. They close the tab.
We built an assessment for people in that state and asked them to make a voluntary, effortful choice to begin it. The completion rate reflects how that friction lands when the audience is exactly who we intended.
What this means for assessment design
The standard model for behavioral health and cognitive assessment tools assumes a motivated, available user. Complete the intake form. Work through the modules. Return next week.
That model may work fine for populations with moderate load. It appears to work poorly when the presenting condition is the very thing that makes initiation difficult.
A few design directions follow from this observation:
Friction-aware onboarding
If the cost of starting is high for the target user, the entry point needs to cost less. A shorter first interaction — even a single question — may function better than a complete assessment form, because it does not require the same upfront commitment. The user gets a partial result and a reason to return.
Progressive disclosure
The full DLI score does not need to be delivered in one session. Breaking the assessment into segments with meaningful partial outputs at each stage changes the decision calculus. The user no longer has to commit to the whole thing — just the next step.
Async completion
Email-based or SMS-based assessment delivery, where a user answers one question per touchpoint over several days, offloads the initiation problem entirely. There is no "start" moment. There is only a response to a prompt that arrived when it arrived.
These are not speculative improvements. They are structural responses to a structural problem. The initiation gap suggests that the standard assessment delivery model is mismatched to the population it is trying to serve.
Curious where you land?
The Decision Load Index measures cognitive friction from unprocessed decisions. Takes about 5 minutes — or you can start with a single question.
Check your DLI scoreWhat we are building toward
We are treating the initiation gap as a design constraint, not just a launch problem. The next version of assessment delivery is being scoped around progressive disclosure and lower-friction entry points.
We do not have clean completion data yet — the system is early and has had known technical issues during this pilot period. What we have is a directional signal: 879 people recognized the framing, signed up, and did not begin. That is worth designing around, regardless of the precise numbers.
We do not know yet whether friction-aware redesign will close the gap. We know the current design does not.
That is where we are. The measurement will continue.
879 people recognized they might be overloaded. Are you one of them?
The assessment takes 5 minutes. The result is a single number that describes your current decision load.
Start measuring