Employee benefits data: how to audit yours in an afternoon

Average benefits utilisation industry-wide is just 13%. Roughly 87p in every pound of benefits spend goes unused.

Now set that against intent. 77% of UK employers link their benefits to at least one business objective, and the goals they're naming are real ones: retention, engagement, and productivity. Yet only 33% of the employers who review their benefits say those benefits fully meet their aims. The intent is there. The proof is thin.

When that gap shows up, most teams reach for a new benefit. Another provider, another login, another launch email, and another line on the budget. It feels like movement.

But there's a cheaper option, and it should come first. Understand the data you already own. Every vendor you pay makes a usage report, or holds one behind a login somebody set up in 2023. Pull twelve months of it into one sheet and you'll know what your people use, who they are, and what you're funding for nobody at all. 

Why has nobody pulled your utilisation report?

For most businesses, benefits data sits behind five or six separate vendor logins, each with its own export format and its own idea of an active user, so the job of pulling it together seems arduous.

Ask a People team when they last read a full year of usage and you'll get a pause. The renewal pack gets read. The invoice gets paid. The usage report lands in March and it's still sitting there in September.

This habit also shows up in the research. 22% of employers have no set goals for their benefits at all, and around 15% of the employers who do set goals don't review against them. A benefits budget that's never reviewed is a standing order.

In professional services, the spend per head runs high, and so does the cost of replacing anyone who walks. So unused spend hurts more there than almost anywhere, and it's the easiest thing in the world to miss.

What counts as good employee benefits utilisation?

Utilisation is the share of your headcount who used a benefit at least once in twelve months, counted benefit by benefit, and read against the number of people that benefit was ever built to reach.

Two numbers matter. Reach, which is unique users divided by headcount. And depth, which is how often those users came back.

There's no single pass mark, so set your own. Do it per benefit, before you pull anything, so the result doesn't get graded after the fact. The CIPD employee benefits data hub is a sensible place to sense-check what your peers offer.

You'll want one more column. Cost per user, which is the annual cost divided by unique users. That's the column that turns a usage report into a decision, and it takes one formula. If you'd like a rough cost per head before you start, the Heka employee benefits calculator does it in a minute.

Where does your benefits data already live?

Four sources cover almost everything you need: your vendor portals, your payroll and expenses file, your insurer or broker claims summary, and your HR system for headcount by team, location, and length of service.

Pull these five things:

• Unique users per benefit, for the last twelve months
• Total spend per benefit
• Users by team, by site, and by length of service
• Booking dates, so you can see the seasons
• Anything a vendor calls a registration, kept apart from real use

Registration is the number vendors like to send. A registration counts a person who made an account once. Use is the number you want, and it's the one that's harder to get.

If a vendor won't give you unique users, ask again in writing. Then ask for it as a standing report at renewal. Every serious provider should hold that number already. If they don't, you've learned something.

How do you run the audit in one afternoon?

Block three hours, put one person on the exports, and work benefit by benefit: list, export, count unique users, divide by headcount, then flag whatever sits at the top and the bottom of the range.

Two rules make the afternoon work. One person does every export, because switching logins is the slow part. And you write the three sentences on the day, while you've still got the surprise fresh.

We go wider on all of this, in our webinar, The September reset. Get the recording on demand.

What does low use actually tell you?

Low use is a fit signal: people can't find the benefit, can't book it in under five minutes, or sit in a life stage the benefit was never built for, and all three are fixable this quarter.

It’s important to note here that a utilisation audit shows you where the money goes unused. The audit never tells you why. The reason sits with the people who never logged in, and the only way to get it is to go and ask them. Five twenty-minute conversations will teach you more than the spreadsheet did.

For example, in a law firm or a consultancy, the answer is usually time. Mental health support that needs a form, a wait, and a weekday appointment loses to a billable hour every single time. People don't skip it because they don't value it. They skip it because they've got a deadline.

Why does personalised delivery change the number?

Personalised delivery lifts use because the right benefit reaches the right person at the moment they need it, which is what turns a wide list into a reach number worth reporting.

Run the audit and the pattern shows up fast. Reach is thin, depth is thinner, and cost per user is highest on the benefits nobody could find. Personalisation works on the middle column. The same spend, routed by who someone is and what they're already searching for, reaches fewer people on paper and gets used by more of them in practice.

On Heka, that shows up as depth. 98.5% of bookings are for preventative health, and employees who feel their benefits meet their needs are 3x more likely to stay. Those are repeat-use numbers, which is the column your audit is about to make you care about.

How Heka works covers the mechanism if you're rebuilding a benefits mix this autumn.

What can your benefits data predict?

Twelve months of anonymised usage is a forecast: the categories your people book now show where pressure is building, so you create a solution before the absence, the grievance, or the resignation arrives.

Heka's tracked 100,000+ bookings across 36 months, which is enough to watch the shapes repeat. Mental health clusters. Financial pressure spikes. Fertility journey signals.

Those patterns flow back into your benefits strategy, your workforce planning, and your DEIB (diversity, equity, inclusion, and belonging) reporting. Every level is anonymised, so no individual is ever identified.

Mental ill health causes 41% of long-term absence in the UK. So a report showing therapy bookings climbing across one team in August is a staffing warning for November. Read it in August and you've got three months to prepare. But read it in December and you've got a potential vacancy.

Where do you start this week?

Start with one afternoon and twelve months of your own usage data, because a benefits decision built on evidence beats one built on instinct every single time.

Come back to the 13% average benefits utilisation figure. That's the average share of a workforce using their benefits at all, industry-wide, and it's why roughly 87p in every pound of benefits spend goes unused. You've been paying for that number without ever seeing it.

The right support, in front of the right person, at the right moment, and the data to show you where it went.

Want the shortcut? Book a demo and we'll show you what your usage data looks like on Heka. Pricing is there if you'd rather see the numbers first.

Frequently asked questions about employee benefits data

What is a good benefits utilisation rate?

There's no single pass mark, because reach depends on what a benefit is for. Average benefits utilisation industry-wide sits at 13%, so anything above that is ahead of the field. Set a target per benefit before you pull the data. A flu jab and a fertility benefit shouldn't be judged against the same figure. 

How often should we audit benefits data?

Once a year, timed about three months before renewal, so the findings land while you've still got room to change something. Add a lighter check at the halfway point: unique users per benefit, nothing else. That second look takes twenty minutes and it catches a benefit that's quietly stopped working in month four.

What data can we ask a benefits provider for?

Ask for unique users, total transactions, spend, and a breakdown by month, all for the last twelve months. Ask for it as a standing quarterly report, written into the contract at renewal. Providers hold these numbers already. A provider who can't produce unique users is telling you something useful about how closely they watch your account.

Do we need a data analyst to run this?

No. A spreadsheet, one afternoon, and a person who can log in to each portal covers it. The only skill involved is division: unique users divided by headcount. Bring in a data analyst later, once you want the picture split by team, site, and length of service, because that's where the interesting differences show up.

What should we do with a benefit nobody uses?

Talk to five people before you cut anything. Low use often means the benefit is hard to find or slow to book, and neither of those costs much to fix. If the benefit still sits unused three months after you've fixed the route to it, move the money to a category your data shows people already search for.