Election nights often begin with dramatic reversals. A candidate builds a ten-point lead in the first hours, then watches it evaporate by midnight. Or the map appears to have settled — until counting resumes the following morning and the outcome flips entirely. Understanding why this happens is not a matter of suspicion about the process; it is a matter of understanding how votes travel from the ballot box to a published figure, and how long each step actually takes.
The counting pipeline
When polls close, counting does not begin at a central location. It begins at each individual polling station, where the local election officials open the ballot boxes and count the votes by hand — typically in front of witnesses appointed by the competing parties. The officials then fill in a results form, sign it, and transmit those figures (in many countries, both physically and digitally) to the district or county commission.
The district commission aggregates the results from every station in its area and forwards the totals upward to the national electoral commission, which publishes them on its official portal. In large countries, this chain can involve three or four levels of administration, each of which adds a small but real delay. In countries where ballot boxes must be physically transported from remote areas to counting centers, the delay can be significant.
This is why results never arrive all at once. They trickle in, station by station, district by district, over a period of hours.
What "percent reporting" actually means
Alongside vote totals, broadcasters and results sites typically display a "percent reporting" figure — "34% reporting," "78% reporting," and so on. This number is widely misunderstood, and that misunderstanding causes a great deal of unnecessary alarm on election night.
Percent reporting refers to the share of polling stations (or precincts, in American terminology) that have submitted their results — not the share of votes that have been counted. A small rural polling station with 200 registered voters submits its results and moves the figure the same amount as a large urban station with 3,000 voters. The two are not equivalent, but the percent-reporting counter treats them as though they are.
This matters enormously in practice, because small stations tend to finish counting faster. They have fewer ballots, shorter queues, and simpler logistics. So when a result page shows "20% reporting," it may already reflect a disproportionate number of rural or small-town stations — and none of the large urban ones.
Why early results are often misleading
The order in which results arrive is not random, and it is rarely representative of the final vote. In most countries, rural areas report before cities. Rural polling stations are smaller and have fewer ballots to process; city stations may handle 10 times as many voters, take longer to clear the queues, and take longer to count. A candidate who runs strongly in rural areas will therefore tend to lead on election night, sometimes by a large margin, even if the overall election is close or trending the other way.
The United States provides the most prominent example of this pattern. In the 2020 presidential election, Donald Trump led in several battleground states late on election night, because in-person votes — which leaned toward Trump — were counted first. The millions of mail-in ballots, which leaned heavily toward Biden and were counted after in-person votes in several states, shifted the results over the following days. This was not fraud or manipulation; it was the predictable consequence of different ballot types being processed in a fixed legal order.
Many European countries do not have this complication, because they permit minimal or no postal voting and count everything on election night. Results in Germany, France, and the Netherlands are typically complete or near-complete by midnight. But countries with large absentee, mail-in, or early-voting programmes should be expected to take longer.
Projections versus official results
Alongside the official partial results, you will see something quite different: projections or "calls" from broadcasters and news agencies. These are not official figures. They are statistical estimates, produced by private organizations using a combination of exit poll data, partial results, and historical voting patterns. They are often accurate — but they are forecasts, not facts.
The terminology can blur:
- Exit poll: a survey of voters as they leave the polling station, conducted by a private research organization. Published the moment polls close. Often accurate, occasionally significantly wrong.
- Projection / call: a broadcaster's or agency's judgment that, based on available evidence, a candidate will win. Not binding. Can be retracted.
- Preliminary results: the official, incomplete count as published by the electoral commission. Accurate for the stations that have reported, but not the full picture.
- Certified results: the final, legally binding count, published by the electoral commission after all ballots have been processed and verified. This is what actually determines who takes office.
When results become final
Certified results rarely arrive on election night. The timeline depends on the country and the type of election. In parliamentary systems with relatively simple ballot structures, certification can happen within a day or two. In elections with multiple ballot types — in-person, early, mail-in, overseas, and provisional — the process can take a week or more.
In some countries, a recount is automatically triggered if the margin falls below a legal threshold. In others, candidates or parties may petition for a recount, which adds further time. Courts can also review challenges, occasionally altering the certified result. All of this is normal, functioning democratic process — not a sign that something has gone wrong.
The best way to follow an election is to keep these layers clearly separate. Preliminary results tell you the direction of the count, not its final destination. Projections tell you what experienced analysts expect to happen. Certified results tell you what actually happened. Each serves a different purpose — and only the last one determines who governs.