Andrew Wiseman considers the implications for survey design via a flat World Cup performance from the 3 Lions.
I watched the first 20 minutes of England v Ghana last night, then switched it off.
Call me a shitfan if you like. I’ll admit it. I’m always club first, country second when it comes to football. Maybe that makes me less emotionally invested. Maybe it makes me more honest. Either way, after 20 minutes I had a pretty strong sense of where the game was heading.
England had plenty of the ball, but something wasn’t quite clicking. Ghana, with no disrespect intended, looked organised, disciplined and very comfortable doing the difficult, unglamorous stuff well. England looked like a team trying to force something that wasn’t really there.
At that point, a 0-0 draw felt like it could be a likely outcome.
Of course, I didn’t know that for certain. Football is capable of making you look stupid. A deflection, a red card, a moment of individual quality, a goalkeeper doing something weird with his feet. Any of those things can change a game.
But you can still spot the shape of something early.
And because I have the sort of brain that can apparently turn a flat England performance into a market research blog, it got me thinking about quantitative surveys. Specifically, the importance of watching what happens early.
The importance of spotting friction early
We’ve all seen it. You craft the questionnaire. You agonise over the wording. You script it, test it, check the routing, stare at the survey link until the questions stop looking like actual words. Then you put it out into the big wild world and wait for the completes to roll in.
But sometimes… they don’t.
A couple of days in, response rates are poor. Dropout is higher than expected. The incidence rate looks odd. The average interview length is suddenly much longer than it was in testing. Or there is one question where people are giving up in suspiciously high numbers.
Sometimes the issue is obvious. A question is poorly worded. A grid is too long. An answer list doesn’t include the option people actually need. A concept is too complicated to process without more context. Maybe the routing has gone rogue and is sending perfectly good respondents into a dead end.
Sometimes the issue is less obvious, but the data gives you the warning signs.
That is why the first phase of fieldwork really matters. It is the survey equivalent of watching the opening 20 minutes. You are not trying to declare the final score with total certainty. You are trying to understand whether the game is unfolding in the way you expected.
The good news is that, in research, just like in football, you can use the hydration break (GRR!) to make changes.
You can simplify the wording. You can break a difficult question into two easier ones. You can add a short instruction. You can amend the routing. You can tighten a screener. You can make the experience better for the people giving up their time to take part.
It doesn’t mean your design was weak in the first place. Just that something didn’t land as intended, and paying early attention allows you to change direction.
There is sometimes a temptation to treat the questionnaire as sacred once fieldwork has started. As though any change is an admission of failure. I don’t buy that. Clearly, you need discipline. You can’t keep tinkering with core measures in a way that destroys comparability. You can’t rewrite the whole survey halfway through because you’ve had a wobble. You need to know what can safely be adjusted and what needs to remain fixed.
But there is a big difference between protecting the integrity of the data and ignoring clear evidence that something isn’t working.
The same applies to early analysis.
I remember a project where, after the first 20 interviews, there was already a pretty clear signal that a proposition we were testing wasn’t landing. Before the stats police kick the door in, I know 20 interviews is not a statistically robust base. I know you wouldn’t make a major decision purely on that number. I know confidence intervals exist.
This wasn’t the equivalent of laying a bet on that 0-0 result. But the early data suggested something awry. Something that needed to be looked at.
The proposition was confusing people. The language wasn’t connecting. The intended benefit wasn’t obvious. You could see participants circling around the idea, trying to make sense of it, but not quite getting there.
That early read gave us the chance to speak to the client before another 50, 100 or 200 interviews had been completed with the same problem baked in. We could explore whether the proposition needed to be reframed, whether the stimulus was doing enough work, and whether the analysis plan needed to account for what we were seeing.
That is where good quantitative research becomes more than data collection. It becomes active, responsive and intelligent.
The best researchers don’t just set the survey live and come back at the end with a PowerPoint. They watch the fieldwork. They check the patterns. They look for friction. They interrogate the early data without overclaiming from it. They know when to hold their nerve and when to intervene.
The early stages won’t tell you everything. They won’t predict the end result, whether that’s in a football or research sense. It will let you know whether something is clicking, whether people are struggling, and whether you need to change the approach before the opportunity has passed.
England couldn’t get the first 20 minutes back, but they could have adapted. They could have gone more direct, changed the tempo, made Ghana defend in a different way. Surveys are no different. Once the early responses are in, they are in. You cannot pretend they didn’t happen. But you can learn from them, adapt what comes next, and avoid letting the same problem play out until the final whistle.
Early monitoring of fieldwork is critical. The first responses reveal whether questions are clear, routing works, and the survey is behaving as expected. Just like the first 20 minutes of a match can signal the flow of the game, early survey data can prevent wasted time, reduce participant frustration, and protect the integrity of your research.
At Mustard, we help clients design, monitor, and adjust quantitative surveys intelligently – so you capture insight efficiently and avoid costly missteps. If you’re planning surveys in 2026 and want them to be active, responsive, and reliable, get in touch.