1. Heywood was not a Labour stronghold

The media has made much reference to the seat being “rock solid” or “very safe” for Labour. It wasn’t. Jim Dobbin’s majority was only 5,791: very close to what would technically be called a marginal. Up to the 2010 general election Heywood was indeed a stronghold: in 2005 Dobbin’s majority was 11,083. But since 2010 the seat has been what could be classed “provisionally safe”: the kind Labour could and should be able to defend fairly comfortably, so long as nothing is taken for granted.

There are dozens of seats like this across the country. The incredibly close result of Thursday’s poll ought to encourage the party to pay those seats rather more attention.

2. The opinion polls in Heywood were wrong

We only had two of them, but fieldwork for the most recent was carried out between 30 September and 4 October: just one week before polling day. It suggested Labour were on 47% (six more than they actually won) and Ukip were on 28% (11 less). There looks to have been a hefty shift in opinion towards Ukip in the 72 hours ahead of the vote. This is not new; it happened in Eastleigh in 2013. I wonder if this opinion poll bred a degree of complacency within Labour and, to refer to my point above, a sense of taking things for granted.

3. Turnout stuck to the trends

I thought the added spice of a Ukip insurgency might boost turnout in Clacton above levels seen in recent by-elections. It didn’t. Turnout in the two most recent contests where Ukip was a strong challenger – Newark in June 2014 and Eastleigh in February 2013 – was exactly the same: 52.8%. In Clacton on Thursday it was 51.2%. In 2010 turnout in Clacton was 64.2%.

Meanwhile Heywood & Middleton fitted the pattern for much lower turnouts when the fight is between Labour and Ukip. It was 39.3% in South Shields in May 2013, and 28.2% in Wythenshawe & Sale East in February 2014. In Heywood on Thursday it was 36.0%. In 2010 turnout in Heywood was 57.5%.

4. Douglas Carswell won fewer votes than in 2010

I know turnout in Clacton was lower than the general election, but the Carswell “brand” has shed a few votes along the way. He won the support of 22,867 people in 2010; on Thursday this figure fell to 21,113. Conversely his share of the overall vote rose from 53% to 60%.

5. The Liberal Democrats suffered the worst electoral result for any party in government since the war

The party’s 1.3% share in Clacton wasn’t quite the worst it has ever achieved. According to Professor John Curtice, that particular honour remains the 1948 by-election in Glasgow Camlachie, when the Liberal party won 1.2% of the vote.

At least the Lib Dems didn’t end the night £1,000 out of pocket. Although they lost their £500 deposit in Clacton, they kept it in Heywood by a whisker, securing 5.1% of the vote.

6. Ukip won’t hold the balance of power at the election

Any party hoping to play kingmaker in May 2015 will need a lot more than two or three MPs to their name. Not even nine will be enough for Ukip. With whom would they form a coalition? Nobody but the Conservatives would countenance working with them. And the Conservatives will need far more than nine extra MPs to form a coalition government if they continue to bump along in the low 30s in opinion polls. They’d phone up the Lib Dems first, and keep phoning until they got a deal.

7. The Tories will want to prolong the Rochester & Strood campaign as long as possible

We still haven’t got a date for this by-election, though I can’t imagine it will take place this month. A date of 6 November feels plausible. This would mean the result would be announced early in the morning of 7 November: precisely six months before the general election.

The Tories have already announced they intend to choose a candidate through a postal primary, allowing them to contrast such a process with the behaviour of their erstwhile colleague Mark Reckless. I suspect they’ll want to use a long campaign to grind down the opposition. It could be counter-productive, making more people more determined to vote Ukip. At least it would give us the chance for a couple of fresh opinion polls – and preferably as near to election day as possible.

8. Both Labour and the Tories will be desperate to avoid any more by-elections before polling day

And as a fan of such contests, I’d rather we did see them. It would be fascinating to see a by-election in a top Labour target or a Tory marginal. It’d also be intriguing to see how Ukip fared in a contest held in one of those metropolitan centres so despised by Nigel Farage – Birmingham, Cardiff, or an inner suburb London. Then we’d really get the measure of the “People’s Army”.