The election is now just one day away and while pollsters have been trying hard to fix the problems that failed to spot the winners in 2015, 2016 and 2017, the companies involved reckon they have it licked this time.
But have they? There’s one place where peoples’ true feelings about politics are often inadvertently revealed: their Google search history. Last time out, I spotted an interesting trend: judging by change in search patterns, people were more willing to give Labour and Jeremy Corbyn a hearing as the campaign continued, and at the expense of the Liberal Democrats.
Google has once again been keeping an eye on search trends for the election, so what can we learn? Let’s find out.
1) Labour are winning the search war
Over the course of the campaign, there have only been two days when the Labour Party hasn’t been the most searched for party, and then only by a nose:
Search volume doesn’t necessarily count as an endorsement of course (especially as demographically, younger people are more inclined to back Labour anyway), but Mr Corbyn will be grateful to get a hearing, and it looks like people are gradually responding, just as the polls suggest.
Between 6th November and 21st November, Labour got an average interest of 25, with the Tories on 13 and the Lib Dems on 12. Between 21st November and 6th December, the numbers have moved to 38 for Labour and 20 for the Tories with the Lib Dems stuck on 12.
2) Events matter
Those spikes you see correlate closely to events, by the way. The big Labour spike in graph one is for the party’s manifesto launch, and the smaller blue spike after is the day of the Conservative manifesto launch.
The same is true when you look at party leaders, albeit a little less clear cut. I’ve thrown Nigel Farage into the mix here, although he’s borrowed the Green Party’s colours due to a lack of customisation options in Trends. Send your complaint emails to Google – or to the Brexit Party, I guess.
That little green bump is the day the Brexit Party stood down in all Conservative-held seats, while that big red and blue spike is the day of the ITV election debate between Jeremy Corbyn and Boris Johnson. That rise at the very end is the BBC debate – as you can see, it’s quite a bit lower than the first one, but I suspect that’s because Google hasn’t quite caught up yet (this graph was made on 9th December, but Google’s data only goes up to the 6th.)
It responds to negative events too. Jo Swinson’s yellow bump is the day of the Question Time Leaders’ special where she took something of a clobbering, while the second red bounce is the day Jeremy Corbyn did himself no favours in the interview with Andrew Neil.
That said, people also started searching for Boris Johnson’s interview with Andrew Neil at the same time:
...which should hearten Labour a little bit.
3) But people generally don’t care...
But – and it’s an enormous but – people don’t really care. And here’s the proof:
This is the same graph as above, only instead of “Brexit Party”, the green bar represents “Premier League”. Only once have any of the parties cleared that line, and it was while domestic football was paused for an international break.
4) ...and it’s getting worse
Okay, but doesn’t that just mean that politics isn’t that popular? Well, yes, but it’s also getting worse. Here’s the graph of the three main parties again, only this time I’ve backdated it to 2017.
5) Local interest is based on how tight the race is, and on Leave/Remain lines
While those graphs are my own work, Google has been keeping up a Trends hub of its own for the election, and it has some interesting insights. If you’d assume people are more engaged if their vote is likely to make a difference, it seems you’d be right.
These are the places which were the most interested in the election last week:
Stroud MP David Drew is holding onto a majority of just 687, while Zac Goldsmith has one of just 45 for the Conservatives in Richmond. Beeston, meanwhile, is the home of Change UK leader Anna Soubry, where both her old party and Labour were just 863 votes apart at the last election.
The chart is fairly fluid, seemingly shuffling in newcomers every page refresh which is encouraging. Here’s how it sits going into the final week:
New entries for West Wellow and Portishead. The former is in Romsey and Southampton North, where Tory Caroline Nokes sits in a majority Remain area where the Lib Dems held the seat until 2010. The latter is in Remainish North Somerset, which is home to arch Brexiteer Liam Fox (majority 17,102).
6) Search interest is sometimes news led
There are curious mini spikes when you look for specific terms, like the Andrew Neil interviews. But not always where you’d expect.
Here, for example, are some popular Jo Swinson search terms:
Two of these search terms relate to a surprisingly weird fake news story that claimed the Liberal Democrat leader used to fire stones at squirrels with slingshots. “I’m a crack shot,” the fake article quotes her as saying. “I don’t go for the head because that’s too clean a death.” Chilling.
Here’s another more serious one. After the Conservatives were widely condemned for renaming their PR Twitter account “factcheckUK” during the ITV Leaders Debate, party chairman James Cleverly said it was all fine, because everybody knows what CCHQ stands for.
Judging by the search spike in the aftermath, they really don’t:
One more. This is the interest in “onanism” over time.
So what happened on 13th November? No, it wasn’t that a whole bunch of porn viewers deciding to use a simile for a change. Instead, it seems to be a direct response to a planned Boris Johnson speech where he compared Labour’s interest in referenda to masturbation. He didn’t give the speech in the end. What a tease.
7) ...but mostly it hasn’t moved since 2017
But these are short, fleeting moments with relatively little interest. And in that respect, Google reckons, the UK electorate’s main interest has barely shifted since the last election:
8) And we still don’t understand the single most important issue of the day
Lots of hard Brexit voters will tell you they know exactly what they were voting for in 2016. Maybe they did as individuals – but on aggregate, it really doesn’t seem likely if Google’s search data is a good gauge.
9) The questions people ask Google is sometimes quite telling
Last time out, I detected a real change in search terms over the election campaign. In April, the term “how many seats will Labour lose?” showed up in the top questions, but by June the phrase “will Labour win?” had replaced it.
Unfortunately, Google doesn’t seem as interested in its own Trends hub this time around, and it hasn’t been updated. The company has ignored my specific questions, so we’ll just have to read between the lines.
If I were a Tory strategist, I’d be hoping that “why vote Conservative” is only third because the message was already clear. I’d also hope that the phrase “how long have the Conservatives been in power” wasn’t accompanied by a sigh when typed into Google.
By contract, if I were a Labour strategist, I’d worry about the pessimistic results Google offers when “will Labour win the election” is typed in.
Still, at least both have better results than The Brexit Party:
Despite being called “The Brexit Party”, two of the questions suggest people don’t know what the party is all about. The third suggests some doubt that could lead Brexiteers to vote tactically, and the fourth will likely disappoint, given the party’s decision to stand down in 317 seats.
Still, it could be worse. Spare a thought for Change UK:
So what will happen?
It’s pretty clear that outside of hardcore enthusiasts and Twitter propagandists, there just isn’t that much enthusiasm for this election in the real world. Or at least the real world as reflected by Google search interest.
The polls suggest that Labour is catching up with the Conservative Party, but not by enough to make the difference required – and the search data doesn’t make me question this as it did a little in 2017.
But there are a couple of key caveats: firstly, there's still a day to go, and there’s still a lot of “don’t knows” that will (probably) put their vote somewhere. Secondly, we haven’t had a December election since 1923 – with the possibility of snow. That could do strange things to the result that even Google can’t predict, so we’ll just have to wait until Friday 13th December for the full post mortem.