The Scale of the Defeat


The Scale of the Defeat

  • This was a terrible defeat for Labour – but one that was a long time coming
  • The losses were widespread but concentrated in the North, Midlands, North Wales and Scotland
  • Labour lost support among all groups of voters compared to 2017 – but the losses were greater among lower income, manual or routine workers
  • The generational and educational divide grew further at this election
  • Labour was less successful in turning out new voters than the Conservatives
  • Labour’s vote is increasingly based in cities, not towns
  • Many of the seats we lost had become contestable because of what happened in 2017
  • The seeds of the loss go much further back to the 2000s and before
  • The geographical landscape of this defeat is very different from comparative elections
  • There is now a mountain to climb to become the largest party or win a majority

This was a terrible defeat for Labour – but one that was a long time coming

Labour members, candidates, staff and supporters worked incredibly hard in this election. The numbers out on the doorsteps, in the particularly challenging circumstances of a December election, were unprecedented.

This makes the result both more painful and more important to understand.

Labour lost 59 seats, the second largest number lost by any Opposition for a century. There are now only 202 Labour MPs in the House of Commons – the lowest number since 1935.

The Conservatives now have one of their biggest majorities since the Second World War – exceeded only in 1959, 1983 and 1987.

The only other time an opposition party has lost so many seats to the incumbent was in 1983, when the Conservatives had been in power for four years. For a major party to fall this far behind after nine years in opposition – and four elections – is historically unprecedented.

On the face of it, this result seems a sharp reversal of the advances made by Labour in the 2017 election. But a deeper investigation of the patterns and drivers behind the 2019 vote points to underlying trends and dynamics that were already at work in 2017 and – as we explore in Chapter 2 – much earlier.

The losses were widespread but concentrated in the North, Midlands, North Wales and Scotland

In England, Labour lost seats in every region except the South East, where it is starting from a low base. In London, Labour gained one seat but lost another. The largest number of seats lost were in the North West, West Midlands, and Yorkshire and Humberside.

In Wales, Labour lost six of its 28 seats (five being in North Wales, where we only held one seat out of six), its vote share falling from 49 to 41 per cent.

In Scotland, Labour retained just 51 per cent of its 2017 vote, and collapsed to its worst ever vote share [1]. As well as losing six of its seven seats in Scotland, Labour is now out of the top two parties in more Scottish seats than ever before, as the Conservatives solidify their position as second placed party across much of the nation.[2]

This pattern of seat loss is an effect of where Labour held seats with majorities that were vulnerable to voters switching to other parties. 

Labour lost votes everywhere. The most significant swings were seen in the North East, West Midlands, Yorkshire and Humberside and East Midlands. 

Labour lost support among all groups of voters compared to 2017 – but the losses were greater among lower income, manual or routine workers

There is a debate about how current polling and statistical categories relate to the modern realities of economic “class”.[3] There is also nothing inherently new about support for the Conservative Party among some lower income or less educationally qualified workers, which was a recognised reality in the early 1960s. [4]

But it is clear that in this election Labour lost proportionately more support among workers in lower paid, more manual or “routine” occupations, with important electoral consequences.[5]

Using NRS Grades[6], likelihood of sticking with Labour was noticeably lower among 

  • C2 (“skilled manual workers”), and 
  • DE (“semi-skilled and unskilled manual workers”; “state pensioners” [that is, pension-age voters without any personal or occupational pension], “casual and lowest grade workers, unemployed with state benefits only”) respondents. 

Likelihood of sticking with Labour was higher among 

  • ABs (higher and intermediate “managerial, administrative and professional” workers) and 
  • C1s (“supervisory, clerical and junior managerial, administrative and professional” workers).

Ipsos MORI data also indicates that, compared to 2017, the deepest net falls in Labour’s vote share were among C2s. Figure 5 below shows in 2019 the Conservatives took the lead among DE voters, overtaking Labour. 

The generational and educational divide grew further at this election

This result displays an electorate starkly polarised by age and education level.

  • Labour won a large share of voters under 35 - according to Ipsos MORI, 62 per cent of 18-24 year olds and 51 per cent of 25-34 year olds. But Labour was well behind the Conservatives among voters over 45 - by 18 per cent among 45-54 year olds, 21 per cent among 55-64 year olds and 47 per cent among voters over 65.
  • According to YouGov, Labour was ahead of the Conservatives among graduates, but 17 per cent behind among voters with medium level qualifications, and 33 per cent behind among voters with no qualifications.

Of Labour’s 2017 supporters, younger voters were more likely to stick with the Party than older - 79 per cent of 18-24-year-olds voted Labour again, compared to 69 per cent of voters aged over 55.[7]

These divisions have been apparent over a number of recent elections, but their depth is unprecedented.

Since at least 1992, younger voters have been somewhat more likely to vote Labour, and older voters Conservative. However, this polarisation sharpened dramatically in 2015, even more in 2017 and further in this election. 

Labour was less successful in turning out new voters than the Conservatives

Over 3.2 million people registered to vote during the election campaign (although it is likely that 1 million may have already been registered).[8] While not all of these will have been first time voters, two thirds of them were under 35. This was more than the 2.3 million who registered during the 2017 campaign.

However, overall turnout was slightly down on the 2017 election, at 67.3 per cent. The biggest increases in turnout were mostly in cities; turnout was generally down in the seats that Labour lost, with the sharpest fall in Stoke-On-Trent North.[9]

Though non-voters have always been difficult to study (as panels tend to heavily over-represent the politically interested, and under-sample the politically disengaged), the large YouGov dataset which Datapraxis used for this Labour Together review includes a much larger and more balanced sample of non-voters than most previous political pollsters’ datasets. It therefore enables us to make the first serious estimates of the role of previous non-voters in the 2019 election - both in terms of those who did not vote in 2017 but turned out in 2019, and those who voted in 2017 but not in 2019. 

This analysis indicates that in 2017 Labour benefited much more than the Conservatives from 2015 and 2016 non-voters, particularly in the 18-44 age bracket.

In 2019 the data indicates that well over 4 million voters turned out in 2019 who had not voted in 2017. 

Labour’s main gains in this election came from 1.8 million people who hadn’t voted in the 2017 election, most of them young. 59 per cent were aged 18-34 and around a third were newly registered. However, these were offset by 1.8 million 2017 supporters who abstained in 2019.

Meanwhile, the Conservatives turned out around 2 million who didn’t vote in 2017. This mobilisation of previous non-voters accounted for almost two thirds of the Conservatives’ increased vote in this election.

Almost half of these previously abstaining Conservative voters had voted Leave in 2016; most of the rest were habitual non-voters. They had a balanced mix of ages and social grades, and were spread right across the nations and regions, with greater concentrations in the West Midlands, Yorkshire and Humberside and the South West, and fewer in Scotland, London and Wales. Only a small fraction had voted Labour at previous elections since 2010.[10]

Labour’s vote is increasingly based in cities, not towns

Behind the general swing against Labour, this election continued the concentration of our support in metropolitan areas. The one seat Labour gained in this election (Putney), as well as the 12 others in which our vote share increased, and those in which we suffered the lowest swing, tended to be:

  • in or around large cities, especially London but also Manchester, Leeds, Birmingham and other multi-constituency cities
  • characterised by both high measures of deprivation and significant numbers of professional workers, often with relatively high numbers of BAME voters or students

Although this election saw a swing against the Party that was manifested in almost every constituency, many of these seats had such large Labour margins that they could absorb these setbacks without changing hands. In a few cases such as Canterbury, Portsmouth and Putney, Labour consolidated support. Across 18 cities outside London that are large enough to contain three or more constituencies, Labour still holds 53 of 73 seats.[11]

Labour retained a strong advantage among Black and Minority Ethnic voters, winning 64 per cent of BME voters against the Conservatives’ 20 per cent.[12] According to research from Datapraxis, Labour retained a comparatively high proportion (over 80 per cent) of Labour 2017 voters identifying as Arab, white and black Caribbean, Bangladeshi and Pakistani, compared to 73 per cent of voters identifying as White British.

Retention rates among voters who identified as Indian were even lower, however, at 68 per cent – likely reflecting the growing support for the Conservatives among voters identifying as Hindu. Hindu voters who had supported Labour in 2017 were 42 per cent likely to withdraw their support in 2019 – compared to Muslim voters who were 80 per cent likely to stick with Labour. Data also suggests that Labour had difficulty retaining voters identifying as ethnically Chinese.[13] However, this commission notes that some care needs to be taken when interpreting current estimates of ethnic minority voting patterns as BAME voter groups are often underrepresented in polling samples and panels.

The majority of the seats Labour lost were centred on towns and small cities rather than within larger conurbations. Many of these seats were once centred in mill towns, potteries or collieries. Today they are characterised by:

  • below-average living standards – though not as low as Britain’s poorest constituencies
  • a comparatively stable and ethnically homogenous population
  • slightly higher-than-average rates of home-ownership but comparatively low property values.[14]

Many of the seats we lost had become contestable because of what happened in 2017

Some of the seats Labour lost might be seen as typical “bellwether” swing seats that have been won by the Conservatives in the recent past.[15] However, many others are seats that have never before returned a Conservative MP and would not long ago have been regarded as core Labour strongholds.

The loss of these longstanding Labour seats to the Conservatives may be shocking. However, the shifts in support that have led to this outcome have been slow, not sudden. Most of these seats were lost as a result of swings that continued a long-term trend of fragmenting Labour support and a rising Conservative vote apparent over previous elections. 

Many of these constituencies had already been left highly vulnerable by the larger advances made by the Conservatives in the 2017 election.

Analysis shared with us by Professor Rob Ford of the University of Manchester shows that in the 18 English and Welsh seats lost in 2019 for the first time since 1945, the Conservatives had increased vote share by an average 14 per cent in the 2017 election. This meant they only needed a smaller increase in 2019, combined with a fall back in Labour’s share, to take these seats in 2019. The Conservatives had already paved the way to victory in these seats back in 2017.

In line with this, analysis by the Resolution Foundation has shown that, across all constituencies, it was in 2017 that the Conservatives made the largest increases in vote share relative to the strength of the 2016 Leave vote. The further increase in Conservative vote share relative to Leave votes was smaller in 2019, but enough to tip many more constituencies over the edge.[16]

Data reveals that 2017 was also when the Conservatives made their biggest increase in support from voters in “working class” occupations (which under this NS-SEC-based classification make up the largest group in the electorate), though it was not yet enough to tip the balance in many seats. Analysis by the British Election Study indicates that the Conservatives achieved their big increase in support among “working class” voters in 2017, an increase which was consolidated in 2019. 

In Scotland, there was clear evidence in 2017 that Labour was losing support to the Conservatives among anti-independence voters, as nationally the Conservatives overtook Labour in terms of vote share. 

However, as shown by Figure 14 in 2017 the impact of this in terms of seat gains was partly offset by a fall in turnout among SNP voters that was reversed in 2019. This allowed Labour to win back some seats from the SNP in 2017, despite Labour only gaining less than 10,000 additional votes between 2015 and 2017.

The seeds of the loss go much further back to the 2000s and before

A longer view of voting trends in these seats indicates that the roots of these losses go further back than 2017. 

Over the last two decades, the Conservatives have steadily increased their votes and seats across all the nations and regions of Great Britain, except in London. They took the lead in the Midlands a decade ago. Meanwhile Labour declined dramatically in every region between 2001 and 2010, losing most seats in Southern England as well as many in the Midlands and North.[17]

In many English and Welsh constituencies, the Conservatives have been achieving significant increases in vote share since 2010. These Conservative advances often followed in the wake of falls in Labour support evident from the 2001 election onwards. 

This pattern of Labour support dropping and remaining at a lower level, followed after some delay by cumulative increases in Conservative vote share, can be seen in seats that Labour lost in this election (such as Rother Valley or North West Durham). This is also true in seats that had already been lost (such as Amber Valley or Cannock Chase). 

For example, the biggest swing to the Conservatives since 2005 was seen in Mansfield. This was lost narrowly by Labour in 2017, but data highlighted by Professor Rob Ford clearly shows that the decline of Labour support and rise in Conservative support was in evidence by 2010 and continued in 2019, so that it now has a relatively large majority.

Seats lost in 2019 such as Bolsover, Sedgefield and Walsall North are among 12 seats in England and Wales that have seen a cumulative swing from Labour to Conservative of more than 25 points since 2005.

A further 27 seats in England and Wales, most of them in the North and Midlands, have seen swings of more than 21 points from Labour to Conservative since 2005 – some of them lost before 2019 such as Tamworth or Amber Valley, and others such as Normanton, Pontefract & Castleford and Barnsley East that were retained.

The 2019 election result thus confirms and exacerbates profound shifts in political representation that, in some cases, have been underway for many years.

Election results do not give us definite data on how specific groups of voters shift their support from election to election. However, analysis of the available evidence by the British Election Study [18]and others [19]suggests that many traditional Labour voters started moving away from the Party around two decades ago.

For a time, these “lost” voters abstained in elections, lent support to the Liberal Democrats or Conservatives, or lodged protest votes through smaller parties such as UKIP, the Brexit Party or the BNP. However, in recent elections, and particularly since the 2016 European Referendum, the Conservatives have been increasingly successful in bringing them into their electoral coalition.[20]

In Wales, Labour’s vote declined from 2001 to 2010, recovered in 2017 but has now fallen back again. Meanwhile the Conservatives have been increasing their vote share steadily since 2001 and are now a closely placed second to Labour in vote share.

In Scotland, Labour’s vote held up between 2001 and 2010, but collapsed after the 2014 independence referendum. As Figure 13 shows, although 2017 produced a slight revival, the Party has now fallen even further.

The geographical landscape of this defeat is very different from comparative elections 

One way of seeing how Labour’s “base” has changed is to compare Labour’s result in 2019 with the seats it held onto in 1983, its most comparable result in terms of seats since the war. 

Greg Cook, has estimated that in England and Wales there are about 35 seats – most of them in the West Midlands (11), North West (6), East Midlands (5), North East (5), and Yorkshire and Humberside (7) – that were Labour even in 1983 that have now gone to other parties. Meanwhile, Labour now holds around 80 seats in England and Wales – most of them in Greater London and the North West – that it did not in 1983.

This shifting “core” in England and Wales is visualised below: seats coloured grey were held in 1983 but lost in 2019. Most of them are in the North and many are geographically large town-based constituencies. Seats coloured red were not part of Labour’s core in 1983 but were retained in 2019. Most are in densely populated urban areas in London and the North West. In Scotland the picture is particularly dramatic. In 1983 Labour held 33 seats that it doesn’t today, while the sole Scottish seat Labour holds today, Edinburgh South was actually won by the Conservatives in 1983. This comparison illustrates the significant shift in Labour’s core support over the past four decades. 

Another relevant comparison is the defeat of 1987, when the Conservatives won a comparable advantage in terms of both seats and vote share to 2019. However in 1987, according to YouGov estimates, social grades C2 and DE made up 62 per cent of the electorate, and 78 per cent of Labour’s support; today they make up 43 per cent of the electorate, and 40 per cent of Labour’s support.[21]

There is now a mountain to climb to become the largest party or win a majority

To be the largest party at the next election, Labour now needs to make a net 82 gains from the Conservatives, requiring a swing of 7.9 per cent. This is almost as much as the 8.8 per cent swing achieved in 1997.

To win a majority at the next election, Labour now needs to win 124 seats. This has few recent parallels.

In 1997 Labour won 145 seats, but the Party had already won 271 at the previous General Election. The 1997 landslide amounted to a 54.2 per cent increase in seats.

In 2010 the Conservatives won 96 seats from a much lower level of 210, which amounted to a 54.5 per cent increase. 

To win the 124 seats needed to have a majority in the next Parliament, Labour needs to increase its number of MPs by more than 60 per cent. This has never been achieved by any major party.


  • This was a terrible defeat for Labour. This result, with losses across the North and Midlands, Scotland, and North Wales, poses profound questions about the future prospects of our Party. Labour’s electoral coalition had been fracturing for a long time and was broken in 2019. We were rejected by many of the communities we were founded to represent.
  • We lost all types of voters everywhere compared with 2017, except in London.
  • Age, education and place are the new electoral divides even more than traditional conceptions of class.
  1. We have seen dramatic changes in relation to older voters, those with lower levels of education and qualifications.
  2. Labour lost votes across every region and country in the UK; Labour’s vote share declined most in small, medium, and large towns, but consolidated in cities.
  3. Labour lost support amongst all classes but amongst working class communities the most.
  • The swing away from Labour in our heartland seats in the 2017 election, masked by the much better than expected result, foreshadowed our 2019 defeat. The Conservatives made significant gains in 2017 in seats they would go on to win in 2019.  
  • Labour faces a substantial challenge to win the next election, with a historic swing of over 10 per-cent needed to gain a majority of one seat. No major party has ever increased their number of MPs by over 60 per-cent, which is what Labour would need to do to win in 2024.
  1. Datapraxis analysis of YouGov data for Labour Together
  2. Datapraxis analysis of YouGov data for Labour Together
  3. See for example: Luke Cooper and Christabel Cooper, The Devastating Defeat: Why Labour lost and how it can win again; Part 1: Britain’s new political divides in the Brexit election, Europe for the Many, 2020:; Paul Mason, After Corbynism: Where Next For Labour?, December 19:; Ash Sarkar, ‘It’s a myth that Labour has lost the working class’, The Guardian, 10 December 2019:
  4. Robert McKenzie and Allan Silver, Angels in Marble: Working Class Conservatives in Urban England, University of Chicago Press, 1968.
  5. British Election Study, Mar 2020:
  6. For background see ‘Social Grade’, National Readership Survey:
  7. Datapraxis analysis of YouGov data for Labour Together
  8. Benjamin Butterworth, ‘At least 1m of the 3.2m people who registered to vote in the general election aren’t new voters’, The I, 26 November 2019:
  9. Tory Landslide, Progressives Split: A Datapraxis Analysis of the UK General Election, Datapraxis:
  10. Tory Landslide, Progressives Split: A Datapraxis Analysis of the UK General Election, Datapraxis:
  11. Peter Kellner, ‘Five crucially important but frequently ignored facts about the 2019 election’, Prospect, 16 December 2019:
  12. ‘How Britain voted in the 2019 election’, Ipsos MORI, December 2019:
  13. Datapraxis analysis of YouGov data for Labour Together
  14. Charlie McCurdy et al, Painting the towns blue: Demography, economy and living standards in the political geographies emerging from the 2019 General Election, Resolution Foundation, February 2020:
  15. Lewis Baston, ‘The Myth of the Red Wall’, Lewis Batson, The Critic, 18 December 10:
  16. Torsten Bell, ‘Election dissection’, Resolution Foundation, 13 December 2019:; See also Matthew Sowemimo, Labour’s Last Stand?, Compass:
  17. Datapraxis analysis of YouGov data for Labour Together
  18. Geoffrey Evans and Jonathan Mellon, ‘The Re-shaping of class voting’, British Election Study, 6 March 2020:
  19. David Cutts et al, ‘Brexit: the 2019 General Election and the Realignment of British Politics’”, The Political Quarterly, Vol 91, Issue 1, February 2020:
  20. David Cutts et al, ‘Brexit: the 2019 General Election and the Realignment of British Politics’”, The Political Quarterly, Vol 91, Issue 1, February 2020:
  21. Roger Liddle submission to the Labour Together Review, 2020