Election polls and consumer confidence figures may offer a snapshot of public opinion, but they cannot reliably predict what voters, customers, employees or investors will do next, according to Kari Butcher, Burson’s chair for intelligence and insights for the Americas.
Her warning comes as the US midterm elections approach and the economy remains voters’ leading concern. Eighty-four per cent of registered voters told Pew it was “very important” to their vote, while CNN’s average of recent generic-ballot polls put Democrats ahead of Republicans by 52% to 42%.
Butcher, a former pollster, said such figures showed preferences at a particular moment rather than a guaranteed future outcome. Businesses that treat polling as a behavioural forecast risk cutting stock when demand remains strong, entering political debates customers did not seek or mistaking the difficulty of switching provider for genuine loyalty.
Consumer confidence and spending point in different directions
The contrast between sentiment and behaviour is evident in recent economic data. The Conference Board’s consumer confidence index fell 6.7 points in September to 81.9, its lowest level since 2014.
The following day, government figures showed that consumer spending had risen 0.9% in August, while personal income increased by 0.2%. Households bridged the gap by saving less, with the saving rate falling to 4.1%.
The figures suggest that consumers can feel deeply pessimistic while continuing to spend. A retailer relying only on the confidence measure could therefore reach the wrong conclusion about demand.
Polling can also change sharply depending on how likely voters are identified. About a week before Michigan’s Democratic Senate primary, one survey gave Abdul El-Sayed a 15-point lead, at 54% to 39% among likely primary voters. He ultimately won by less than a point.
In early October, YouGov’s chief scientist found that the same survey could produce a Democratic lead of between six and 16 points in the national House ballot, depending on the assumptions used to identify likely voters. He said the highest figure was almost certainly an overstatement, while more defensible estimates ranged from eight to 14 points.
The responses had not changed. The assumptions about how people would behave had.
Businesses urged to combine polling with behaviour
Butcher said polling itself was not necessarily failing, but that business leaders needed to establish what decision a particular poll provided enough evidence to make. A single figure could not explain whose views it captured, what lay behind them or whether respondents would act on them.
Evidence from Burson’s own research illustrates the difficulty. In a survey of 1,601 US adults carried out in late April, 61% said economic anxiety was causing them to save more and pay closer attention to companies’ words and actions.
Four months later, national data showed households saving less. The surveys covered different periods and were not a direct test of one another, but they demonstrated the risk of treating a stated intention as a forecast.
The Burson Pulse study also found that 48% of respondents said they had gradually lost confidence in a company without being able to identify a single defining moment. Another 52% said their confidence had been damaged by one event.
That suggests an overall reputation score may remain stable while confidence is quietly weakening. By the time the change appears in a headline measure, the underlying damage may already have taken place.
Businesses are therefore being advised to use polling alongside evidence of what people search for, share, buy and do, while continuing to ask them directly about their views. As Butcher put it: “What decision does this give us enough evidence to make?”
That question is particularly relevant to the midterm polling now dominating political discussion. The surveys may indicate who is ahead, but they do not, on their own, show what customers will do next.
