What Are the Benefits of Using a Fully Managed Survey Provider vs a DIY Tool?

DIY survey tools have made it easy to ask employees a question. But they haven’t made it easy to get an answer you can act on, and the gap between those two things is where engagement programmes commonly come unstuck.

This isn’t an argument that DIY tools are bad. They’re inexpensive, quick to set up and perfectly suited to some jobs. It’s an argument about what changes when the thing you are measuring is how people feel about working for you, and what they’ll tell you depends on whether they trust the process.

Where DIY genuinely works

It’s worth being clear about this first, because the honest comparison is more useful than the flattering one.

A form builder is the right choice when you need a quick, low-stakes answer: which date suits people for the away day, what they thought of last week’s training session, whether the new coffee machine was a good idea. The questions are simple, nothing hinges on precise measurement, and nobody is worried about being identified.

Where it starts to struggle is when the stakes rise. When the results will go to a board. When they’ll inform decisions about restructuring, reward or management capability. When you need to compare this year with last year, or your organisation with others. When people need to believe their answers cannot be traced back to them before they will answer honestly at all.

It’s worth naming what the real comparison is, because it’s not really a comparison between two pieces of software. Modern survey platforms are perfectly capable of putting a questionnaire in front of people and counting the answers. The genuine choice is between running the whole exercise in-house and having it run for you, and much of what separates a useful survey from a misleading one sits outside the software entirely.

Response rates: the number that decides everything else

Who answers your survey decides how much the rest of it is worth.

A survey with a low response rate does not give you a smaller version of the truth. It gives you a different one, because the people who do not respond are unlikely to be a random cross-section of the workforce. This is the part that is easy to underestimate: you cannot tell from the data which way the gap runs. Non-participation might be concentrated among people who have given up on the process, or among people who do not trust it, or among those with least time and highest workload. Each of those produces a different distortion, and a report that says “72% of respondents” tells you nothing about which it is. A thirty per cent response rate can look perfectly tidy on the page and still be systematically wrong about the organisation.

What a managed provider brings to this isn’t a better survey tool. It’s the set of things that sit around the tool:

• Independence with something behind it. Where a third party administers the survey, the organisation isn’t in a position to see individual responses. That’s a different proposition from HR running the survey from an internal account and assuring people it’s confidential, and it’s a claim that can be explained to employees rather than simply asserted.
• Communications designed to be believed. Who’s asking, why, what happens to the data, what will change as a result, and visible sponsorship from the top. This is where in-house surveys most often lose people, before a single question is read.
• Access for everyone. Email, SMS, QR codes, paper where it’s needed. In organisations with frontline, shift-based or non-desk workers, the people excluded by an email-only survey are often the ones whose experience differs most from head office.
• Questionnaire design that respects people’s time. Questions placed in an order that gets people to the end, demographics at the close rather than the opening, and a length that reflects what you will actually use.
• Live monitoring during fieldwork. Seeing response rates by group as they come in, so reminders go to the departments that need them rather than to everyone, which is how you avoid both low participation and reminder fatigue.

None of these are features of a platform. They’re decisions made by people who have run this before and know where it goes wrong.

Data quality: the questions you didn’t know were bad

Survey software gives you a blank question box, which is the most dangerous thing in employee research. Writing survey questions looks easy and it isn’t.

Double-barrelled questions (“My manager supports me and gives me useful feedback”) cannot be acted on, because you do not know which half people were answering. Leading questions invite agreement. Unbalanced scales skew results. Inconsistent scale direction creates response patterns that look like findings and aren’t. Questions written in the language of the HR team rather than the language of the workforce get answered inattentively or not at all.

A managed provider works from a tested question bank, which matters for two reasons. The questions have been used before, so their behaviour is known. And because they’ve been asked elsewhere, the answers can be compared to something.

There is also accessibility, which in-house surveys routinely overlook and which isn’t merely good practice. All employers have duties under the Equality Act 2010 to make reasonable adjustments for disabled employees, and public sector organisations are additionally subject to accessibility regulations covering their websites and digital services. If a survey isn’t navigable by screen reader or keyboard, does not allow font and contrast adjustment, and isn’t readable at a sensible reading age, a portion of the workforce has effectively been excluded from being asked.

A survey intended to reach everyone needs to be built against the Web Content Accessibility Guidelines, tested for reading age using a measure such as the Flesch readability test, and offered in paper or easy-read format where people need it. Independent verification of plain language, such as the Plain English Campaign’s Crystal Mark, is a reasonable thing to look for and a reasonable thing to ask about.

Anonymity: a design question, not a promise

This is the difference that tends to matter most to the people being asked, and it is worth understanding properly.

The weakest version is a survey run on a general-purpose form builder, where whoever holds the account can usually see individual submissions, including any identifying details collected alongside them. Employees work this out quickly, and the rational response is caution.

A more useful question than which tool is being used is a simpler one: who decides who sees the raw answers? Where an organisation runs its own survey, that decision sits with the organisation, however carefully it is made and whatever safeguards are put around it.

Bringing in an independent provider moves that decision outside the organisation altogether. Clients don’t have limited, logged or role-restricted access to raw response data. They have none. Results reach the organisation in aggregate, with a minimum reporting threshold applied throughout so that small teams cannot be inferred from the numbers. Free-text comments are reviewed and redacted to remove names, job titles and identifying detail before anyone at the organisation sees them. Where a comment raises a serious allegation, there is a protocol agreed in advance for handling it.

That sits alongside obligations an internal process cannot easily replicate. We’re a Market Research Society Company Partner, bound by the MRS Code of Conduct, registered with the Information Commissioner’s Office, and certified to ISO 27001 for information security management, independently audited rather than self-declared.

The point is not a promise to refrain from looking at who said what. It is that the organisation is never in a position to, and that is one of the reasons people tell an independent provider things they will not type into an internal form.

Reporting depth: beyond the average

Here’s where the difference becomes hardest to argue with, because it’s mathematical rather than a matter of judgement.

A questionnaire tool gives you averages and a bar chart. That tells you the score. It does not tell you what is driving it, and those are very different questions.

Key driver analysis. Using multiple regression, you can identify which factors are most strongly influencing engagement in your organisation specifically. This routinely surprises people. The lowest-scoring question is often not the one with the greatest statistical influence on engagement, which means acting on the obvious result can mean spending a year’s effort on something that was never going to move the needle. This isn’t analysis a questionnaire tool performs on your behalf, and running it in-house means having someone who can both produce a regression and interpret it correctly.

Segmentation that holds up. Results cut by division, location, length of service, role and protected characteristics, with a reporting threshold that protects small groups. And where the data supports it, intersectional analysis, looking at overlapping characteristics rather than each one in isolation. Single-variable analysis frequently misses the group whose experience is genuinely different.

Trend data that means something. Comparison year on year only works if the question wording, scale and population are consistent and documented. Where a survey is rebuilt each cycle by whoever has capacity, that discipline is easily lost: a question gets reworded, a scale changes, and the comparison quietly stops meaning anything. Managed providers hold that line, and can usually bring historical data from a previous provider forward so continuity is not lost at the point of switching.

Reporting that reaches the people who can act. Role-based access means a manager sees their own team, a director sees their teams, and project leads see everything, without anyone manually cutting and distributing spreadsheets. Done by hand, this is usually one person’s evening, repeated for every manager, with the attendant risk of sending the wrong team’s results to the wrong person.

Benchmarking. Your score of seventy per cent isn’t interpretable on its own. Seventy per cent on one question is strong and on another is a serious problem, and that depends entirely on how everyone else answers it. A benchmark built entirely from surveys the provider has conducted itself can be fully accounted for: where it came from, how it was collected, what it represents, and which subsets results can be compared against rather than one undifferentiated average. A survey built from scratch has nothing to compare against at all. And wherever a benchmark does come from, it is worth asking what it is built from, who is in it, when it was last updated and which subset your results are being measured against. A comparison figure is only as useful as the answers to those questions.

Action planning: the part everyone underestimates

In our experience, the most common way an engagement survey fails isn’t a low response rate or a badly worded question. It’s that the results arrive, everybody reads them, and nothing visible happens.

This is also the most damaging failure, because it’s worse than not asking. It confirms to employees that feedback is a formality, it makes the next survey harder to run, and it makes the answers you do get less honest. The CIPD’s guidance on employee engagement puts employee voice among its core enablers, and is clear that employers should give people effective channels for voice and listen carefully to what they say – which is only half done if nothing follows.

Software stops at the data. The survey closes, you download a spreadsheet, and everything that happens next depends on someone’s spare capacity.

What managed support adds here:

• Findings presented, not just delivered. A session with senior leadership that explains what the results mean, with clear recommendations and a position on what to prioritise. Not a 60-page deck left in a shared drive.
• Action planning built into the reporting. Managers create plans linked directly to their own team’s results, with owners, dates, automated reminders and RAG ratings. Project leads see every plan across the organisation on one dashboard, so the question “is anything actually happening?” has an answer.
• Guidance for managers who have never done this. A departmental workbook, support in holding the conversation with their team, and a steer towards no more than three priorities. Most managers do not need a strategy, they need to know what to do on Monday.
• Pulse surveys between main cycles, so progress can be checked without waiting a full year to find out whether anything worked.
• Consultancy support throughout, including facilitated action planning workshops where more hands-on help is useful.

The cost that isn’t on the invoice

Licence costs vary enormously, from a few hundred pounds a year for a basic form builder to per-employee pricing on a dedicated platform that runs well into five figures. Either way, the licence was never the real cost of running your own survey.

The real cost is the time: questionnaire drafting and redrafting, building and testing the survey, writing comms, chasing response rates, cleaning data, cutting reports for every manager, reading and redacting thousands of free-text comments, and building the deck for the board. That work is usually absorbed by one or two people in HR who already had full-time jobs, and it lands at the busiest possible moment.

And then there is the cost of getting it wrong. A survey that produces unreliable data on which a real decision is made is more expensive than no survey. A survey that damages trust because people did not believe it was confidential sets you back further than where you started.

What this looks like in practice

The surveys we run achieve an average response rate of 84%.

That figure is worth sitting with, because it’s not a marginal improvement on a survey run in-house. It’s a different order of participation, and it comes from the things described above rather than from anything clever in the software: genuine independence, communications people believe, access routes that reach everyone, and targeted reminders during fieldwork.

It also changes what the data can be used for. At that level of participation you are no longer reasoning about your workforce from a self-selecting slice of it. Team-level results become reportable rather than suppressed for being too thin. Differences between groups can be taken seriously instead of treated as noise. And when somebody at board level asks whether the findings really represent the organisation, there is an answer.

When running it yourself is the right answer

To be fair to the alternative: if you need a quick temperature check on something low-stakes, if nobody’s job or budget depends on the result, if your workforce is small enough that you already know what people think, and if you have genuinely got the internal capacity to design, run and act on it properly, then do it yourself and spend the money elsewhere.

The case for a managed provider isn’t that you can’t collect responses on your own. It’s that collecting responses is the easiest part.

The real question

The question isn’t really managed or DIY? It’s: what are we going to do with the answers, and what would have to be true for us to trust them?

If the answer is “we’ll have a look and see”, a form and a spreadsheet will do. If the answer involves a board paper, a budget decision, a benchmark, a trend line or a conversation about management capability, then the survey needs to be built to carry that weight, and that is a different exercise.

If you are weighing up the two and would like to talk it through, we would be happy to hear from you.

References

CIPD, Employee engagement and motivation factsheet (updated 23 July 2025) – https://www.cipd.org/uk/knowledge/factsheets/engagement-factsheet/

W3C, Web Content Accessibility Guidelines (WCAG) – https://www.w3.org/WAI/standards-guidelines/wcag/

Plain English Campaign, Crystal Mark – https://www.plainenglish.co.uk/services/crystal-mark.html

Market Research Society, Code of Conduct – https://www.mrs.org.uk/standards/code-of-conduct