How this works

1How the 10 issues were chosen

The issues are the 10 most pressing questions in Israeli society ahead of this election, chosen by their importance in current public debate rather than limited to any one topic area.

For each issue we wrote a short neutral description, a question, and the most common real-world positions on it (3 to 6, depending on the issue). In the survey, you pick one of these positions for each issue. Each position describes a policy approach and its stated reasons, worded so that its own supporters would recognise it as fair. The Hebrew and English versions say the same thing, each written to read naturally and neutrally in its language.

Before the issues were finalised, a separate review checked them for loaded language and for major positions that were missing.

2Which parties were researched, and how

Not every list on the ballot was researched in depth. Which lists to research was decided when the research was done. The other lists are shown with their official details only.

The parties that were researched in depth are included in the matching, and each has its own research page. Parties running together on a joint list were researched separately, each with its own page.

For the other parties, the site shows only official details: the party’s name, its list, its ballot letters, its leader and the top candidates on the list. Where the party has a platform of its own, there is also a link to it. We have not checked that platform. These parties are not included in the matching, and the site says nothing about their positions.

Each research page follows the same structure: overview, leadership, the party’s position on each of the 10 issues, other notable positions, coalition stance, track record, legal matters involving its leaders, and sources.

The research focuses on facts rather than campaign messaging. Evidence is ranked by reliability: actions (votes, actions in government, legislation) come first, then formal commitments (the party platform), then statements. Where a party’s actions contradict its statements, the research says so. Legal matters are reported only from court or official sources.

The research was compiled with AI research tools, working from public sources. Each document was then checked against its sources by a separate AI step, and all the documents were reviewed together for consistent depth and tone. Every substantive claim cites its source, and each page shows the date it was researched.

No one’s personal view decides what a party’s position is. The site’s operator did not review or edit the content about the parties. Each check was done by a separate AI step that had not written the content it checked.

3How matching works

Matching follows fixed rules and uses content prepared in advance from the research. No AI runs when you use the site, and the same answers always give the same result. Everything in a result comes from the published research pages, so you can check it on this site.

For each issue, the research was used to record which of the survey’s positions each party holds. Some parties are recorded with a main position and a second one, and a position that rests on weaker evidence, or that the party holds only in part, is recorded as such and counts for less. The site compares the position you picked with what is recorded for the party. If you picked the party’s main position, that is full agreement. If the party’s position is close to yours, it counts partly. If it is far from yours, it counts for little or nothing. How close any two positions are was set once for each issue, and is the same for every party.

Issues you rank higher count for more. Issues you mark as “doesn’t matter to me” count for nothing.

Where the research found no documented position for a party on an issue, that issue adds only a little to the party’s match, and the result says that the party’s position is unclear. The result also says “unclear” where the research has only weak evidence of the party’s position and you picked a different position. Where the research shows that a party rejected a position, for example by voting against it, the result says the party disagrees with it, even if the party’s own position is close to it.

A party’s actions count for more than its statements. This is applied when the party’s positions are recorded from the research. The recorded positions and the explanations shown in a result were each checked against the research by a separate AI step.

Polls, party size and the electoral threshold play no part in the match. They appear only as factual notes on the result.

All of this is worked out on your own device. Your answers are never sent anywhere.

The result shows the five closest parties in full, and you can open the rest of the researched parties below them, in order. Parties next to each other in that order are often almost equally near to your answers, so a small difference in position means little. For each party you can see, issue by issue, whether it agrees, partly agrees or disagrees with the position you picked, or that its position is unclear, with a short explanation that links to the relevant part of its research page. Each party also comes with a few “also worth knowing” points about notable positions outside the 10 issues. These points don’t affect the match.

If no party matches your positions closely, the result says so instead of presenting a weak match as a strong one.

4Polls and the electoral threshold

A researched party’s page shows its list’s polling average from a fixed snapshot, with its source and date. Pages of parties that were not researched show no poll information. If the list of a party in your result is polling below the electoral threshold (3.25%), the result notes that, with the date. Lists that don’t appear in published polls are noted as such. These notes are factual only; they are not advice about how to vote.

Israeli election law restricts publishing poll results in the days before election day. From the Friday before election day, all poll information is removed from the site, and it isn’t shown again afterwards.

5Limitations

No tool can capture everything that matters to you. 10 issues and a few fixed positions on each simplify a complicated reality, and your own view may not be exactly any of the positions offered.

Not every list on the ballot is included in the matching. A party that was not researched in depth can’t appear in your result, whatever its positions are.

The research is a snapshot as of the date shown on each page. Parties may change their positions after that.

Some of the researched parties, especially new or small ones, have little public record. Their pages say so, and so does the result if one of them is among your matches.

Research and matching can contain mistakes, even after checking. Read the research pages, and make your own judgment.

The result is a decision aid, not a recommendation of who to vote for.

6Neutrality

This tool has no party affiliation and no ads. No party funds it, influences it, or reviews or approves its content. Every party that was researched was researched with the same structure and the same evidence standards. All the parties, researched or not, are listed together in alphabetical order, never by size or polls.

We never publish how many people matched with which party.

7Corrections

Research pages are corrected only to fix errors. The date on each page shows when it was last researched.