All articles ←

What happens after you say no

5 September 2026·8 min read·By The Zugot Project team

A stated preference is a hypothesis. Behaviour is evidence. When the two contradict each other there is only one reasonable decision — and most dating systems make the opposite one.

The question asked at the end

At the close of a virtual date each side rates the experience and separately marks whether they want to continue. Anyone marking no is asked why, and given three options: a religious gap, a lifestyle gap, or simply no chemistry.

In most systems that answer is collected into a report. Here it does something immediate: it changes the profile of the person who gave it, before their next date. Each of the three reasons has its own mechanism, and there is a fourth that runs after every date — including one that went well.

One rule that precedes all the others

Before any mechanism, a constraint applies to all of them: feedback changes only the profile of the person who gave it. The other side's profile is never touched. Under no circumstances.

This isn't a technical limitation but a decision. A system that downgrades someone every time another person says no is a system where one bad experience creates a slope: less exposure, fewer dates, less chance. And worse — it turns feedback into an instrument of judgment about another person, which immediately distorts what people write in it.

Mechanism one: a religious gap

Someone declining for religious reasons is effectively claiming that outlook decided it — even though the system judged the gap tolerable. The reason for the discrepancy is almost always the same: the importance rating the person gave themselves was too low.

The mechanism examines the three core items — halachic observance, Torah study, attitude to the modern world — and looks for those where the gap between the two was two steps or more. On every such item, the rater's own importance rating is pinned to the maximum.

The result is that from then on, that item functions for them almost as a hard filter. They said it was flexible, discovered it wasn't, and the system updated what they said according to what they did.

Mechanism two: a lifestyle gap

Here the logic leans on the leisure tier model. The system looks at the other side's interest tags sitting in the top two tiers — core spiritual and core active — that the rater did not share.

These are the likeliest sources of the gap: not a series they hadn't watched, but a regular practice or a joint activity forming part of the structure of the week. The groups those tags belong to receive half their weight for the rater from then on.

The crux is that the discount is specific rather than blanket. The system doesn't conclude that interests matter less to this person — it concludes that one particular group of interests misled them, and lowers exactly that one.

Mechanism three: no chemistry

This answer is the most interesting of them, because it reports that the calculation was fine. No values gap, no lifestyle gap, the numbers matched — and still nothing happened.

This is precisely the situation the research on virtual dates describes: people are experience goods, and the experience doesn't follow from the data. There is no filter that can be tightened to fix it, and so the system doesn't try to filter differently.

Instead it changes the date itself. Next time, that user receives universal experiential conversation cards in the room — ones that generate conversation between any two people, rather than ones leaning on shared interests. If the conversation never took off, the problem may lie in its starting point rather than in the match.

Mechanism four: initiative level, after every date

The last mechanism differs from the three in that it isn't triggered only by a rejection. It runs after every date, including one that ended in two yeses.

What it learns is which level of initiative from the other side actually suits this person. Each report moves the target one step in the direction reported. And since a successful date answers that question just as clearly as a failed one, there is no reason to ignore it.

There's a small detail here that is easy to miss and worth noticing: when someone reports that the level of initiative was exactly balanced, nothing happens. The target doesn't move, and it also doesn't decay back toward the middle. The reason is that such a report is evidence the current setting is right — and a system that pulled the target back to centre every time things worked would be systematically erasing precisely the learning that succeeded.

The shared principle

The four mechanisms say the same thing in four ways: what a person declares on a form is a starting point, and what actually happened to them overrides it.

This isn't contempt for their answers. It's recognition that nobody knows themselves precisely on these questions in advance, and that the only way to find out is to go on a date and see. A questionnaire that hasn't moved after ten dates is a questionnaire that learned nothing from ten sources more reliable than itself.

And for anyone in the process, the practical conclusion is simple: it's worth marking the real reason. Not the polite one, not the one that's easy to explain. The reason you mark is the one that will change what you're offered next time.