The 2019 General Election delivered a disappointing result for pro-Europeans. While 53 per cent of votes went to parties that supported Remain or backed a second referendum, the first-past-the-post electoral system delivered a majority government on a minority of votes.

Best for Britain worked to encourage voting pacts between opposition parties at constituency level, encouraged voter registration, as well as running a large tactical voting operation. This report is both an examination of the impact Best for Britain had as polling day neared, and a look at the valuable lessons that have been learned.

Read the full Impact Report below, or download a PDF here.

With the honourable exception of 60 constituencies in which Unite to Remain participants (Plaid Cymru, Greens and Lib Dems) fielded a single candidate, the failure of the Remain parties to ally, and replicate the Farage/Johnson pact, meant those wishing to stop a Conservative majority were left with tactical voting as the only weapon in their arsenal.

Our disappointment at the election outcome is tempered by a quiet pride that we had such a significant impact, in a situation where the odds were stacked heavily against us. As this report shows, if we had not done what we did, the result could have been even worse.

Key Figures

• Our digital channels achieved more than 200 million impressions and in the six weeks prior to election day we reached 25 million people in a 45.8 million electorate.
• More than 800,000 people voted tactically in line with Best for Britain’s advice.
• Our efforts helped Labour to victory in six seats they may well not have otherwise won.
• The party we recommended received a significant boost because of our tactical voting advice in 19 seats.
• More than 4.5 million unique visitors came to getvoting.org for advice
• We recommended the highest placed remain candidate in 97% of constituencies where we made a recommendation

Tactical Voting

Tactical voting – voting for a party which isn’t your first choice – was a big part of Best for Britain’s election strategy, and with good reason: It’s extremely significant. According to the election day Ashcroft poll and analysis by our own data experts:

• 10 per cent of the electorate voted Labour even though another party was their first choice. This group would largely have been Remain voters who felt Labour were the best chance for a Final Say vote.
• 5 per cent of the electorate voted Lib Dem even though another party was their first choice. This group would largely have voted Labour in 2017.
• 7 per cent of the electorate voted Conservative even though another party was their first choice. Some will have preferred the Brexit Party but there will also have been quite a few who were Lib Dem leaning but very anti-Corbyn (eg people who voted Conservative in 2017, Remain in 2016, Lib Dem in the Euros).

Was our data accurate?

Best for Britain is a data-driven organisation, so it is vital that we have trustworthy figures to use for recommendations. Looking at the election results and our recommendations, the accuracy of our data and resulting predictions were unparalleled. Best for Britain recommended the highest placed Remain candidate in 97 per cent of constituencies where we made a recommendation.

We used a technique called ‘Multilevel Regression with Poststratification Analysis’ (MRP). MRP starts with a large poll – in this instance, a poll of up to 46,000 respondents – then adds in other data sets, which could include census age and income information, to produce a more accurate reflection of voting intentions in every constituency.

MRP accurately predicted the surprise wins for Labour in Kensington and Canterbury in 2017, as well as the 2016 election of Donald Trump. Data science company Focaldata carries out the analysis for us, using the most up-to-date polling available. While some tactical voting websites chose to use the 2017 General Election as a baseline for their findings, we insisted on using up-to-the-minute MRP to capture changes in voting sentiment.

Read the full Impact Report below, or download a PDF here.