The only poll that matters is at the coffee shop

In an election that defied pollsters, it was the old-fashioned tally room at Miss Maud that called the result correctly.

Almost 32,000 votes were cast across the 18 Miss Maud stores in Western Australia, with customers dropping a bean into a box for the party they would give their first preference.

Despite the obvious limitations of the ballot not being secret and potential double voting, the Miss Maud poll was highly accurate.

For example, 39.6 per cent of customers picked the Liberal Party as first preference, while 2.3 per cent picked the Nationals, for a combined total of 41.9 per cent.

Nationwide, the coalition primary vote was about 41.4 per cent, according to the Australian Electoral Commission, with about 76.3 per cent counted.

The coalition primary in WA was 45.3 per cent.

Miss Maud customers picked Labor 28.4 per cent of the time, much less than the 33.9 per cent who voted for the ALP nationally, but close to Labor’s WA primary vote of 30.6 per cent.

Minor party votes were also close.

The Greens scored 10.3 per cent of Miss Maud votes, while picking up 11.4 per cent of WA primary votes.

One Nation picked up a primary of 5.4 per cent in the bean counting poll, and 5.1 per cent in WA on Saturday.

Interestingly, One Nation performed better in the bean poll than on voting day, which potentially puts paid to suggestions that right leaning voters are shy about being honest in polling.

Miss Maud was also close to the United Australia Party's numbers, predicting 3.7 per cent compared to 3 per cent nationally on the day.

A YouGov Galaxy exit poll for the Nine Network on Saturday had predicted a 52 per cent to 48 per cent victory to Labor, with a coalition primary of 38 per cent.

The final Newspoll of the campaign had favoured Labor 51.5 per cent to 48.5 per cent after preferences.

Miss Maud claims the poll has accurately predicted eight federal elections and three state elections.

The company says the reason the poll has been accurate is simple.

“Customers only put in a bean if they feel like it and if they have something they want to express,” Miss Maud said.

Their second argument is that people are more relaxed over a coffee than a quickfire answer when called by a pollster.

Method in the madness

There are a couple of feasible explanations for the accuracy.

It’s actually a much bigger sample size than many official polls, with even the final Newspoll on the Friday before election day only interviewing 3008 voters nationally.

Notably, however, there’s a barrier to entry- people have to buy a coffee to participate.

That means it’s less likely to be skewed by repetitive voting manipulation in the way some online polls are.

The second reason is that pollsters usually use complex methodology to interpret results, including, in some cases, calculating preference flows based on previous elections rather than on the stated intent of the voters surveyed.

For the Miss Maud poll, simplicity might be the virtue.

Ken Wyatt, a Liberal frontbencher who was returned in the seat of Hasluck on the weekend despite polls predicting otherwise, offered a pithy analysis.

"The Miss Maud Poll has never let me down and has been far more reliable than so many other polls, which seem to have failed miserably at this election," Mr Wyatt said.

"Maybe the likes of Newspoll, YouGov, Galaxy and Ipsos should spend more time sipping coffee and counting beans! It’s Miss Maud’s one, SportsBet zero."