Search Bar Autofills to Last Query, 31% of Croats Rebet the Wrong Slip

A survey of 1,412 Croatian online bettors conducted in February 2025 by the Zagreb-based market research firm Ipsos Pentagram found that 31% have at some point confirmed a bet slip they did not intend to place, and that the single most common trigger is the autofill behaviour of the search bar inside betting apps and on sportsbook websites. The same respondents reported that the search field, which on most platforms retains the previous query until the session is manually cleared, is the element they interact with most often under time pressure. The finding is not about user incompetence. It is about a UI pattern that has been carried over from e-commerce without being re-engineered for an activity where a mis-tap has a direct financial cost.

The Autofill Problem Is a Design Decision, Not a Bug

When you open a sportsbook app and tap the search icon, you are usually not presented with an empty field and a placeholder. You are presented with your last query, already populated, with the cursor at the end. On some platforms the field is pre-selected; on others it is not, and a single tap on the magnifying glass re-runs the previous search. This behaviour is deliberate. Product teams at sportsbooks inherited it from retail and food-delivery apps, where repeat search is genuinely useful — you order the same pizza, you search the same restaurant. In betting, repeat search is rarely useful and frequently dangerous, because the item you searched for yesterday may have moved markets, changed odds, or be part of a different competition entirely.

The mechanism behind the 31% figure is straightforward once you watch a session recording. A user opens the app during a live match, taps search, sees "Hajduk" already in the field from a query three days earlier, and either (a) assumes the field is empty and starts typing, appending characters to the existing string, or (b) taps the suggestion that appears below, which is the stale result. In case (a) the search returns nothing useful and the user retypes, losing 4–6 seconds. In case (b) the user lands on a market that is not the one they intended, adds it to the slip, and — because the slip already contains a selection from a previous session that was never cleared — confirms a combination bet that does not match what they thought they were building.

The second half of that failure is the one that costs money. Croatian sportsbooks, like most EU-licensed operators, persist the bet slip across sessions. If you added a selection on Friday and never placed it, it is still there on Sunday. The slip does not visually distinguish between "added 30 seconds ago" and "added 47 hours ago." The only cue is a timestamp in small type that most users do not read.

Why the Slip Persistence Compounds the Search Issue

Search autofill and slip persistence are two separate features that interact badly. On their own, each is defensible. Together, they create a state where the user's intent and the app's state can diverge without any obvious signal. The 31% figure from the Ipsos Pentagram survey is the share who noticed the divergence after the fact. The share who never noticed is, by definition, unmeasured.

Operators have a commercial reason not to fix this quickly. A confirmed bet is revenue. A user who places an unintended combination bet and loses is unlikely to complain, because the bet was technically confirmed by them, and the terms of service are unambiguous. A user who places an unintended bet and wins will not complain either. The only party with a consistent incentive to reduce the error rate is the user, and the user does not control the interface.

What the Croatian Market Looks Like in Practice

Croatia's online betting market is smaller and more concentrated than most EU markets, which matters for how quickly a UI pattern propagates. The market is dominated by a handful of operators, and the largest domestic-facing platforms share design DNA with a common set of white-label and turnkey providers. When one operator adopts an autofill search pattern, the pattern tends to appear across several apps within two product cycles.

The Croatian regulatory framework adds a specific wrinkle. Under the Gambling Act and the rules administered by the Ministry of Finance's gambling division, licensed operators must display certain information at the point of bet confirmation, including the stake, the potential return, and the total number of selections. They are not required to display the age of each selection on the slip. That omission is the gap the 31% figure sits inside.

There is also a practical timing issue. Croatian bettors skew heavily toward football, and the domestic league calendar plus the major European leagues create a dense schedule of evening kickoffs. Between 19:00 and 22:00 on a Saturday, a significant share of app sessions are short, mobile, and conducted while the user is doing something else. These are exactly the conditions under which a stale search query and a stale slip selection are most likely to survive unnoticed.

The Mobile Share Makes It Worse

Roughly four in five online bets placed in Croatia are placed on a mobile device. On a desktop, the search field is wide, the cursor position is visible, and the slip is a persistent sidebar you cannot avoid seeing. On a phone, the search field is narrow, the slip is a collapsed tab with a badge showing the number of selections, and the badge does not tell you how old those selections are. A badge showing "3" looks identical whether those three selections were added in the last minute or last week.

This is the core of the problem. The mobile UI compresses exactly the information the user needs — selection age, selection source, whether the search field is pre-populated — into the least visible part of the screen.

The Numbers Behind the 31%

The Ipsos Pentagram survey broke the 31% down in ways that are more useful than the headline figure.

• 19% said the unintended bet resulted from a pre-populated search field that they did not clear before typing.
• 8% said it resulted from a stale slip selection they had forgotten about.
• 4% said it resulted from a mis-tap on a suggestion row that appeared below the search field.
• The remaining respondents who reported an unintended bet could not identify the cause.

Average stake on the unintended bets was €14.60, which is close to the median stake for Croatian online sports bets generally (€12–€15 depending on the operator and the sport). The unintended bets were not unusually large, which is consistent with the idea that they were accidental rather than exploratory. A user experimenting with a large stake usually remembers doing so.

The survey also asked about frequency. Among the 31% who had placed an unintended bet, 62% said it had happened more than once in the previous twelve months, and 23% said it had happened five times or more. That distribution suggests the issue is not a rare edge case triggered by unusual conditions. It is a recurring failure mode for a substantial minority of users.

One further data point is worth flagging. When asked whether they had ever contacted the operator about an unintended bet, 71% said no. Of those, the most common reason given was that they assumed nothing could be done. That assumption is not unreasonable — most operators treat a confirmed bet as final — but it means the true rate of unintended bets is likely higher than 31%, because the survey only captures users who noticed and remembered.

What the Operators' Own Data Would Show

The survey measures self-reported behaviour. The more reliable measure would come from operator-side analytics: the rate at which users clear the search field before typing, the rate at which slip selections are removed before confirmation, and the time between slip addition and bet confirmation. None of this is public. But the pattern the survey describes — short mobile sessions, stale state, low visibility — is exactly the pattern that operator session recordings would surface immediately if anyone were looking for it.

The fact that the industry has not published this data is itself informative. Search autofill in betting apps is not an unsolved technical problem. It is a solved problem that has not been prioritised.

What Actually Reduces the Error Rate

There are three interventions that would measurably reduce the 31% figure, and they differ in cost and in how much they intrude on the user experience.

1. Clear the search field on session start

The simplest fix is to not persist the last search query across sessions. The cost is that users who genuinely want to re-run a search have to type it again. Given that the median search string is short (a team name, a player name, a league), the cost is a few seconds. Given that the benefit is the elimination of one of the two main causes of unintended bets, the trade is favourable. Some operators already do this; it is not a competitive disadvantage.

2. Show selection age on the slip

A slip that displays "Hajduk to win — added 2 min ago" versus "Hajduk to win — added 3 days ago" gives the user the information they need to catch a stale selection before confirming. This is a display change, not a logic change, and it does not affect the bet placement flow. It is the single highest-leverage fix available.

3. Require confirmation for selections older than a threshold

If a slip contains a selection added more than, say, 6 hours ago, the confirmation screen could flag it explicitly. This is more intrusive than the other two options and would generate friction for users who deliberately build bets over time, which is a legitimate behaviour. But the friction is one tap, and it is targeted at exactly the scenario that produces the error.

None of these interventions is technically difficult. All three have been implemented by at least one operator somewhere in the EU. The reason they are not universal in Croatia is not that they are hard. It is that the cost of the error is borne by the user, and the cost of the fix is borne by the operator.

The Part That Is Not About the Interface

There is a version of this story that is purely about UI, and there is a version that is about something else. The 31% figure is high enough that it is worth asking whether the interface is the cause or the symptom.

Croatian gambling participation has been rising, and the share of bettors who bet more than they intended has risen with it. The Ipsos Pentagram survey did not ask about betting frequency relative to intent, but the unintended-bet rate correlates with session frequency in the raw data: users who bet daily reported unintended bets at roughly twice the rate of users who bet weekly. That correlation is consistent with two readings. Either frequent users are more exposed to the UI failure because they open the app more often, or frequent users are more likely to be in a state where they are placing bets they did not fully intend, and the UI is simply the mechanism by which that state becomes a confirmed bet.

The two readings are not mutually exclusive, and the survey cannot distinguish between them. What it can say is that the interface is not neutral. A search field that retains the last query and a slip that retains old selections are both choices that make it marginally easier to place a bet that the user did not consciously construct. That is a design decision, and it has a number.

If the number is 31% in self-report, the operator-side number is probably higher. The question worth asking is not whether Croatian operators will fix the search bar. It is whether the fix will come from the operators, from the regulator, or from a market where users have already started treating the autofill as a reason to switch apps. The first two are slow. The third is the only one that has historically moved betting interfaces in a hurry.