There is a particular kind of confidence that turns up on the Friday morning after May's local elections. By eleven, someone on a panel show is explaining what the council results mean for the next general election. By teatime, "crushing blow" has been used at least four times. Much of it will be nonsense, and not because the pundits are fools. It is because local elections are not a national opinion poll with fewer cameras.

If you want to read the results properly, you need to know what you are looking at. Here is a working guide.

Turnout is the first number to look at, and the one nobody mentions

Turnout in English council elections sits a long way below what you see at a general election, and it varies wildly from ward to ward. Some wards comfortably pass half of registered voters. Others struggle to reach a quarter. Weather, school holidays, a lively local row and whether another contest shares the ballot paper all move it.

That matters for the story. A party that turns out its base in a low-turnout election can win seats on a modest number of votes, while a party whose support is softer can see a decent poll rating fail to appear at the ballot box. Rising turnout in one part of a council area and falling turnout in another often explains a swing better than any national trend does.

So before you accept anyone's claim about a shift in the country's mood, find out whether the same number of people actually voted. Compare like with like: the last time those seats were contested, not the last general election. And remember the register itself moves. In wards with high turnover, the electorate on the day may look very different from the one that voted four years ago.

Seats exaggerate almost everything

Vote share and seat totals are different animals. First past the post in single-member wards, and something close to it in multi-member wards, can turn a small, uneven change in votes into a dramatic change in seats. A party can drop a couple of points across a council and lose eight councillors, because the same few hundred voters in the same few wards tipped the same few contests.

Multi-member wards add their own layer. Where three councillors are elected at once, voters can use all three votes, split them between parties, or plump for a single candidate. That produces results that look strange to anyone used to Westminster: a ward returning two councillors from one party and one from another, or an independent finishing second on the strength of a well-run local campaign.

The useful exercise is to ask where the seats changed, not how many. Ten gains spread across a county tell a different story from ten gains concentrated in two towns.

The calendar and the boundaries do half the work

Different councils run on different cycles. Some elect every councillor at once, every four years. Some elect a third of the chamber at a time, and some elect half. A council elected by thirds changes slowly, by design, and one bad night rarely hands anyone control. A council holding an all-out election can be transformed in a single evening.

Then there is the map. Boundary reviews redraw ward lines and change how many councillors each area returns. When that happens, the honest answer to "how is this party doing compared with last time?" is: we don't know exactly. Analysts produce notional results — estimates of how the old vote would have translated onto the new boundaries — and they are exactly that. Estimates. Useful, and worth treating with a pinch of salt.

Ward-level quirks worth knowing about

Some results have almost nothing to do with the national picture, and it pays to spot them early:

• Long-standing independents and residents' associations. Some wards have returned the same independent for decades. A win there is not a verdict on any party leader.
• Student and highly mobile wards. The register turns over so quickly that comparisons with four years ago can be meaningless.
• Small rural wards. A few dozen votes can decide the outcome, and one well-attended public meeting can move them.
• Planning rows. Nothing shifts a suburban ward faster than a contentious development — often a local issue wearing national colours.
• Split voting in multi-member wards. Look at how many votes each party's candidates received, not just who came top.
• Uncontested parish and town council seats. Where there are fewer candidates than seats, nobody votes at all. Local democracy has quieter corners than results night suggests.

What a national projection can and cannot tell you

Broadcasters and academics often publish a projected national share of the vote, estimating how the country might have voted if every seat had been up. It is a genuinely useful comparator, and better than adding up raw votes across incomparable councils. But it rests on assumptions. It assumes a uniform swing. It is built from the wards that happened to be voting that day, which are not a random sample of the country. It cannot see the personal vote for a well-liked councillor, the campaign to save a local hospital, or the voter who quite likes their council but wants to give the government a bloody nose.

Treat it as a rough reading of temperature, not a forecast. Anyone who tells you the next general election is settled by a set of district council counts in May is guessing — and so is the person agreeing with them.

How to read next May's results without getting it wrong

Keep it simple, and do the same checks every time:

1. Establish the baseline. Which seats were up, who held them, and were the boundaries the same last time?
2. Compare turnout with the last comparable contest, not with the general election.
3. Count the seats actually contested before talking about gains and losses. A party defending a third of the chamber cannot lose control of it.
4. Look at where change happened — the specific wards, towns and kinds of neighbourhood — rather than the council-wide percentage.
5. Wait a week for the projections and the proper analysis, then read them with the local detail in mind.

Local elections are the most detailed picture of public opinion this country produces, and also the most misread. They reward patience. Read the turnout, count the wards, notice the quirks, and you will end up with a better account of what happened than most of the people talking about it on television.

Photo: Edmond Dantès / Pexels