As Theresa May fired the starting gun for the 2017 UK General Election, a vox-pop response from a voter in Bristol to the news of yet another election went viral. It had been just two years since the last one, not to mention the EU Referendum in between. “Not another one!”, Brenda exclaimed. Little did she know that there would be yet another election at the end of 2019, marking four times in five years that the UK electorate had been asked to go to the polls.
Democratisation of polling
With elections comes polling – and therefore, with more elections comes an avalanche of polling. While Britain had to wait five years from 2019 to 2024 for another General Election, snap policy decisions throughout the COVID-19 pandemic and two changes of Prime Minister meant politics never really went away, and polling continued apace.
It has led, in Britain, to a democratisation of polling and data. In theory, the electorate is more informed than ever. But does more information, more research and more data create greater clarity or more confusion? In short – does the proliferation of polling in the UK actually benefit the electorate, or does it simply create more noise?
The growth of polling has been meteoric. 97 polls were conducted between the dissolution of Parliament and polling day in 2015, but by 2024, there were 152, an increase of more than 50%. The types of poll, or model, that pollsters have released have also increased in sophistication. In 2015, there were no multi-level regression and post-stratification (MRP) models released by pollsters, but by 2024 there were 18.
These models not only give a headline voting intention, but also seat-by-seat results in all 632 Great British constituencies. They provide a snapshot of which sitting MPs may lose, where in the country parties are strong or weak, and a sense of the scale of potential victory or defeat for the incumbent government and main opposition. At constituency level, these models carry a wider margin for error – meaning that having lots of them during a campaign can create far more noise than simply having more ‘traditional’ polling.
In a sense, this is also the democratisation of polling, but in another form. The data is more accessible to the public, but the barriers to entering the industry have never been lower either. Cheap online panels, accessible survey software, social media as a distribution channel; the tools are there for anyone who wants them. And media outlets want polls – they easily fill column inches, contextualise existing stories, and give commentators something to talk about.
The industry has responded – and then some.
I genuinely believe the democratisation of polling is a positive development. The idea that only a handful of large firms should have the ability and tools to measure public opinion feels outdated. New entrants bring new thinking and new perspectives, and that is welcome. Most importantly, they often bring new technology and innovation; indeed, some of the newer entrants to the UK polling market performed very well at the 2024 General Election, using new and untested methods.
They, along with an increasingly volatile electorate, are forcing the more established names to innovate, too. In this sense, the arms race among pollsters feels more progressive than regressive, and therefore, a positive development.
Standards
But democratisation without standards can create problems. With an increasing number of polls and pollsters, it becomes difficult to distinguish between the rigorous and the unreliable. It goes without saying that being new to the industry does not, of course, equal unreliability. But it can create a greater sense of confusion among those that aren’t polling obsessives (i.e., 99.99% of the population) about which polls can be trusted and which cannot. Similarly, with an increasing appetite among journalists, campaign operatives and voters for data, there is a risk that less rigour in reporting may be applied.
We regularly see how polls without a clear provenance can enter the political ecosystem and take on a life of their own. They get picked up by campaigns. They get reported by outlets that may not have the time or resources to interrogate methodology. They can shift prediction markets and shape narratives. And by the time anyone pauses to ask where the data actually came from, the damage has already been done.
In the UK, membership of the British Polling Council (BPC) is often perceived as a badge of honour – a quality mark of a certain standard. But the BPC rightly refrains from positioning itself that way; it is a self-regulatory membership body with transparency at its core, and ruling or commenting on methodology is not within its remit.
The transparency standards the BPC holds its members to allow journalists, political operatives and the public to make their own minds up over the validity of political research. But whether this is practical, or the best solution, given the sheer number of polls and pollsters, could be debated. An expanded remit for the BPC feels unlikely, given it is a self-regulatory body, but some polling agencies may well wish to ascribe to an organisation with emphasis on quality, rather than just transparency.
The future
I’ve worked in political polling for 12 years and have watched the industry change. The proliferation of polls does not just affect public trust in polling; it impacts trust in research more broadly. Savanta is a consumer research agency, with published political polling a tiny part of our global operation. But it is one of the highest profile parts, and ultimately, as pollsters, we are the only subset of the wider market research industry that has our homework marked against an effective census every time there is an election. When a dubious poll makes headlines, the public understandably does not distinguish between that outfit and the firms that have spent years investing in methodology, transparency, and rigour. The reputational consequences tend to be shared.
The answer is certainly not to wish away a crowded polling environment. That genie is not going back in the bottle. But could the industry benefit from stronger self-regulation? I can certainly see more frequent calls for it, not just from outside the industry, but inside it too. Bad, or rogue, polling is a stain on all our houses.
Media organisations could do more to apply due diligence before publishing data from unfamiliar sources – but equally, it is impractical for all journalists to have as firm a grasp of polling, and the industry, as those within it. Those of us who do this professionally could be better at explaining what good polling looks like and why it matters, but that has sometimes led to personal attacks and antagonism that further hurt, rather than help, the industry more generally.
And in an increasingly divided political environment, pollsters’ methodologies are increasingly coming under fire from parties and politicians who seek to undermine those measuring public opinion because the results do not align with their narrative or worldview. Savanta has been caught in the firing line of this before, and I was particularly grateful to the industry peers who voiced support for our position. But these were individuals, not companies, and certainly not an industry body whose remit it was to defend us. Maybe such an organisation could exist, and provide greater scrutiny of its membership at the same time.
The polling industry is not broken – the best firms are producing genuinely excellent work. But that work risks being lost in the noise. It is in all of our interests – but especially pollsters, media and the public – to find ways to turn the volume down.