For weeks Coalition strategists have been predicting a tightening of the polls. At times, it seemed like waiting for a drought to break.

Then, about half way through the campaign, a few fat drops of rain fell. Napthine's people started sounding vaguely optimistic.

There was talk of snaring Labor seats such as Yan Yean, to Melbourne's north, Eltham and Ivanhoe to the north-east and the Ballarat seats of Wendouree and Buninyong. Even the south-eastern seat of Narre Warren North was seen as a possibility.

Then the heavens closed again. In the second last week of the campaign the optimism on the Coalition side seemed to evaporate and a sense of gloom once again settled.

Finally, right at the end of the campaign, the latest Fairfax Ipsos poll shows the long-awaited tightening has now started to occur after a relentlessly negative advertising assault and spending spree in marginal seats.

Labor's primary vote is down, the Coalition's is up and, depending on how you measure it, the two-party preferred result is on a knife edge.

The question is whether it will be too little, too late for the Coalition.

First, more than 1.1 million people – representing about 29 per cent of total enrolments – have already voted.

Never before have voters flocked in such numbers to early polling booths.

In retrospect, however, Labor was clever to have launched its campaign well before the start of the caretaker period. In general, Labor has also conducted a slicker, smarter campaign.

Second, if the Coalition's primary vote really does end at up at 42 per cent as the poll suggests, it is probably still too low to win.

At the 2010 election the Coalition won with a primary vote of 45 per cent, and then it only just scraped over the line with the narrowest of majorities.

On the other hand, according to the poll, Labor's primary vote is now just 35 per cent, a far from emphatic endorsement from voters. To win with a primary vote like that, Labor will probably need to mop up something like 85 per cent of Greens' preferences, slightly above the level of the 2013 federal election.

Third, the final result will depend on how things pan out in the key marginals, particularly in the "sandbelt" seats to the south-east. Labor seems relatively confident it will win back one or two of the four seats it lost in this area in 2010, although there are no guarantees. But it is likely to lose the Ballarat seat of Ripon.

In net terms, Labor needs to win just two seats. As things stand now, it will probably achieve this, although things are less certain than a week ago. Chances are it will be close.