Why Official Election Results Sometimes Take Weeks

The count you watch on election night is not the result. In most countries it is not even an official document — it is a running tally assembled by the electoral commission from polling-station protocols as they arrive, published for the convenience of the public and the press. The result, the one that seats a parliament or swears in a president, is a legal act that comes later. Sometimes hours later. Sometimes a month.

This gap is the single most common source of confusion I see in the questions readers send me, and it is worth understanding properly, because the two numbers are genuinely different — not just in precision, but in what they mean.

The four stages almost every country goes through

The names differ, but the sequence is remarkably consistent across democracies:

1. The preliminary count. Polling stations close, votes are counted on the spot in front of observers, and each station's protocol is transmitted upward. This is the count that appears on live results sites, including this one. It is fast, it is usually accurate, and it is not binding.
2. Consolidation. The commission assembles the protocols, resolves arithmetic errors, and adds what the election-night count could not include: postal ballots, diaspora votes, stations whose protocols arrived late or were contested. Colombia calls this stage the escrutinio; São Tomé and Príncipe calls it the apuramento geral. It is where the numbers move.
3. Adjudication. Candidates and parties file challenges — a station's tally disputed, a ballot type ruled on, a recount requested. A court or the commission itself rules on each one.
4. Proclamation. A named body declares the definitive result. In Peru it is the National Jury of Elections (JNE); in São Tomé and Príncipe it is the Constitutional Court. Only at this point does the result exist in law.

A worked example: São Tomé and Príncipe, 2026

The presidential election of 12 July 2026 is an unusually clean illustration, because the two published figures differ in a way that looks paradoxical until you see the arithmetic.

The National Electoral Commission's provisional count, released on 20 July, put the incumbent Carlos Vila Nova on 55.94% — 31,596 votes out of 56,485 valid ballots. Eight days later the Constitutional Court proclaimed the definitive result: Vila Nova re-elected in the first round with 55.88%, on 31,613 votes out of 56,574 valid.

Vila Nova gained 17 votes and lost 0.06 points of vote share. Both statements are true at once, because the apuramento geral added the late-arriving and diaspora tables, and those added 89 valid ballots to the denominator while adding only 17 to his numerator. Everyone's raw count went up; the shares rearranged slightly. The order of the candidates, and the fact that he cleared half the valid vote and so avoided a run-off, never changed.

That is what a normal, well-run consolidation looks like: the shape of the result is settled on election night, the third decimal place is not. It is also why we publish a result twice — a provisional pass as soon as the outcome is beyond doubt, and a second pass with the certified figures — rather than waiting a month to publish anything at all.

How much movement is normal

Across the elections on this site, the provisional-to-final drift is usually a few hundredths of a point to a few tenths. Iceland's EU membership referendum of 29 August 2026 closed provisionally with No on 52.54% and certified at 52.84% on a turnout of 82.52% — three tenths of a point, from the final constituencies and the reconciliation of blank and invalid ballots.

Peru's 2026 run-off shows why that margin sometimes matters enormously. The JNE proclaimed Keiko Fujimori the winner on 3 July 2026 with 50.135% — a lead of roughly a quarter of a point. In a race that close, the difference between a preliminary count and a completed escrutinio is not a rounding detail; it is the presidency. This is exactly the situation in which the slow, boring, legally exacting stages earn their keep.

Why some countries are fast and others are slow

The length of the gap has less to do with competence than with four structural facts:

• Diaspora voting. Ballots cast abroad physically have to arrive. São Tomé's diaspora tables are a large part of why its consolidation took eight days — and its diaspora turnout was so low that nearly four voters in five abroad did not vote at all.
• Postal voting. Countries that accept ballots postmarked by election day but delivered afterwards have a built-in waiting period, by design.
• Who certifies. A commission that certifies its own count moves faster than a system where a constitutional court must rule first.
• Mandatory challenge windows. Many systems keep the result open for a fixed number of days whether or not anyone actually files a challenge.

None of these is a defect. A count that can be legally contested and corrected is more trustworthy than one that cannot, and the delay is the price of that property.

What to do with this as a reader

Three habits make election-night numbers much easier to read honestly:

• Check what stage a figure is from. On this site, provisional results say so explicitly, and the figure is replaced once the certifying body publishes.
• Treat margins under about one point as unsettled until proclamation. Above a few points, consolidation has essentially never changed an outcome.
• Expect turnout to be the last number to stabilise, not the first. It depends on the final registered-voter roll and the final count of blank and invalid ballots, which is exactly what the consolidation stage is for.

The short version: election night tells you who won. The proclamation tells you by how much, and makes it real.