They got the points, a win is a win, but they're chastened by the experience.

Except it wasn't really like that, it just seemed that way through the unfolding count. In actuality the voters' decision – a Coalition win with 76 or 77 seats in the lower house against Labor's 68 or 69 and a five cross-benchers – was locked in at 6pm on Saturday, July 2 (add a few hours for the west).

It just took a long time for the result to become clear.

The Coalition appears to have received about 50.6 per cent of the national two-party-preferred vote. No federal government has lost with more than 49 per cent support. Sitting governments routinely enjoy an advantage in the vote-seat equation, and usually 50.6 per cent would translate to a comfortable seat victory. But not in 2016.

It's a close seat outcome by Australian standards, but it's a working majority.

Yet when heads hit pillows on Saturday evening the score was 67 seats each, with five others and 11 undecided – a likely hung parliament – and it provoked a conservative backlash including outright demands from parts of the wider Liberal movement that Prime Minister Malcolm Turnbull fall on his sword.

We now know those early numbers were very wrong, but those attacks can't be unspoken, nor the damage inflicted on the leader's authority undone.

Why did the later counting turn around so much in the government's favour?

The answer has several elements: the steadily growing proportion of early votes; electoral law changes put in place six years ago to facilitate the inclusion of more pre-poll votes in the election night count; and unexpected skews in party support in various classes of vote.

For the first few hours on election night, the only available results came from what are called ordinary votes, those popped in ballot boxes on the day. These showed a roughly 3 per cent swing to the ALP from their 2013 result of 46.5 per cent. That looked grim for the government, but help, they hoped, was on the way from another class of votes, whose results would enter the tally from about 9pm. These, pre-poll ordinary votes, are taken before polling day, in the elector's home electorate.

Pre-poll ordinary votes have only been disaggregated from other pre-polls, and counted on election night, since 2010, and at that election and in 2013 they favoured the Coalition, compared with earlier ordinary votes, by about 4 per cent after preferences. With a record proportion of them this year, more than one in six votes, the later evening count was expected to claw back those early Labor gains.

But astonishingly this time pre-poll ordinary votes swung to the ALP by about 9 per cent and favoured Labor by 3 per cent more than ordinary votes. As the evening dragged on the scoreboard got worse for the government. And that was where it largely sat for Sunday and Monday.

Starting early the following week, several other classes of votes were counted, of which easily the largest, and most important, were postal votes. In recent elections postal votes favoured the Coalition by about 5 per cent more than other votes.

In 2016 postal votes, on counting so far, favour the Coalition by about 7 per cent. And their numbers are at a record high, accounting for almost one in 10 of turnout.

Postal votes were the main driver of the Coalition's clawback, but there were a couple of others.

Absent votes are those performed on election day, but in another electorate (in the same state or territory). They usually favour the ALP by several per cent, and they did this time, but they shrank as a proportion of the total. It seems many people who would previously have voted absent made a pre-poll ordinary instead – and so had their vote counted on the Saturday evening.

And yet another category, pre-poll declaration votes, also favoured the Coalition more than usual.

So all in all, the votes counted on election night were particularly unrepresentative, skewing to Labor much more than anyone realised at the time.

Why did this happen? Much more analysis is required, but we should note that the Liberal Party put in a particularly energetic campaign this year. It is unclear to what extent their efforts merely convinced supporters to deliver their vote via that method rather than on the Saturday. If that was a meaningful contributor (we'll never know for sure), then the Liberal postal vote campaign was partly responsible for the discomfort inflicted on its leader, and the campaign team, over the following days.

As for the future, an ever-increasing proportion of postal votes, unable to be counted until days and weeks after the election, is not sustainable. Something will have to be done to include most of them in the election night totals.

That will surely be a popular topic at the Joint Standing Committee on Electoral Matters Inquiry into the 2016 federal election.

Peter Brent is an Adjunct Fellow at Swinburne University.