B2B research is facing a quality crisis – one that’s quietly eroding confidence, inflating risk, and threatening the credibility of the decisions insight teams support. As Anthony Shephard-Williams explains, if we don’t talk about it openly, we can’t fix it.
Anyone working in B2B research today knows the landscape has shifted. Reaching niche audiences has always been hard work, but quality, true, defensible, verifiable quality, is becoming harder to guarantee than ever.
And yet, strangely, we don’t talk about it enough.
When 85% of your sample has to be removed
At the Market Research Society B2B networking event in September 2025, a story stopped the room: a project where 85% of the sample had to be removed because it was simply unusable. Eighty-five percent.
Whilst this may be an extreme case, it wasn’t an anomaly. It was a reflection of something many of us are quietly wrestling with behind the scenes.
What we’re seeing first-hand at Mustard
We regularly go out to three or four sample partners on every B2B brief. Not just for price, but to test feasibility, quality controls, and the overall thinking behind the quote.
And the contrast in supplier responses can be staggering.
The partners who “get it”
The suppliers who cost more aren’t doing so arbitrarily; they’re often the ones:
- putting real thought into feasibility;
- flagging challenges we’re likely to face;
- adding caveats that demonstrate due diligence;
- suggesting mixed method approaches to reach difficult audiences;
- showing their working because they’ve actually done it.
This aligns with our values at Mustard; being real, savvy, and confident. We need to be able to stand behind the data we deliver, and that means partnering with people who take sample quality seriously (give yourselves a big pat on the back, as you know who you are!).
And then there are the others…
On the other end of the spectrum are the suppliers who come back within 30 minutes with:
- no questions;
- no caveats;
- impossibly cheap pricing;
- and implausibly high incidence rates.
It’s the “Yep, no problem, we can get 200 completes in a week” response. Every researcher knows that sinking feeling of something’s not right here.
We’ve had (potential) suppliers claim they could deliver more respondents from an incredibly niche education audience than there were actual institutions in the UK. A panel apparently larger than the population itself!
Another recent brief required an ultra-specific audience that even the most experienced partners approached with caution. Given how niche it was and how much of a challenge I thought it would be to deliver, I upped the number of ‘RFQ’s’ from my usual 3 or 4 to 8. Now a few came back and refused outright, whilst some brought thoughtful, pragmatic alternatives. A few also came back quite quickly with bargain-basement prices and promises that didn’t add up.
When we pushed one of them for verification details (just a simple “How are respondents validated?”), they never replied. Months on, we’ve heard nothing. And we won’t be contacting them again.
Why this matters
Price pressure is real right now. Budgets are tight. Procurement is tough. And it can be tempting, understandable even, to choose the cheapest quote.
But in B2B research, the cheapest option is rarely the safest. Or the most accurate. Or the most ethical.
Poor sample doesn’t just skew data; it skews decisions. And when clients are using our recommendations to shape strategy, investment, product development, and planning, we owe them more than numbers on a page, we owe them confidence.
Industry benchmarks confirm the scale of the problem
The inaugural Global Data Quality (GDQ) benchmarking report which was recently released, shows an unfiltered look at what removal rates are like across the industry between January and June:
Overall, there’s twice as many removals vs B2C respondents and the post survey removal rate is over three times higher. Prior to this study, we’d been very aware that we needed to pay extra close attention to the questions that we put into our B2B surveys with a view to “tripping people up” and bringing nonsensical answers to the front so it’s good to have some stats to back this up.
Where AI helps and where human verification still matters
Alongside these challenges, there’s also a new wave of panel providers emerging (or existing ones evolving), who are using AI to help identify, verify, and manage respondents.
In many ways, this is a positive shift. These tools can make screening smarter, spot inconsistencies faster, and reduce the volume of low-quality or fraudulent participants entering the ecosystem. But even with these advancements, technology alone isn’t enough. Verification still needs human care behind it. The partners we trust most are the ones combining AI with rigorous manual checks: double-verifying respondents, conducting phone validations, reviewing IDs, and putting in the legwork to ensure that every participant is who they say they are. And because of that commitment, there are signs of improvement and real reasons to feel more confident about where B2B sample quality could be heading.
We work with a supplier on regular project with vets and part of the conditions for signing up to the panel is that you must send an email with documentation to prove that you are a veterinary practitioner. This must be signed off (manually) by someone at the panel before you can take part in surveys.
As a result of this method, we have had a 100% success rate on quality for this project – surely something as simple as this translates well into the wider B2B market?
Time for the industry to break the silence
This issue isn’t going away. If anything, it’s worsening as demand for niche B2B audiences increases and some suppliers stretch their panels (and claims) far beyond reality.
At Mustard, we believe it’s time to bring more transparency to the conversation. And we’re committing to doing exactly that.
What we’re doing going forward
When we present a sample cost that sits above the cheapest quote, we’re going to:
- explain why;
- outline the checks we’ve done;
- show the reasoning behind feasibility;
- and be open about the risks associated with too-good-to-be-true pricing.
Because our job isn’t simply to field survey completes. It’s to deliver high-quality B2B evidence that clients can rely on with absolute confidence.
If you’d like to learn more about how we approach sample quality or the standards we uphold across every project, get in touch with theteam@mustard-research.com.