Sourced directly from the EHS database, every entry, every cycle, exact scores. No reconstruction, no estimates.
EHS isn't built to be read comment by comment or used to close the loop on individuals; it's a speedometer for the business, giving a steady, honest read on how culture is trending so a shift can be caught early. Seen that way, the data is reassuring: a fully named cycle (Cycle 1) and the first anonymous one (Cycle 4) landed at almost the same score, exactly what a trustworthy barometer should do. What actually moved the reading in between was who kept checking in, not a real shift in sentiment; the least happy voices from Cycle 1 didn't come back, so Cycles 2 and 3 read a touch rosier than the fuller picture.
None of that needs reversing anonymity to fix. What would genuinely move the reading, in EHS's current form, is more of the workforce answering, more regularly. One detail worth a direct conversation rather than a dashboard debate: Hopton On Sea and Five Lakes moved in opposite directions in Cycle 4, one to its best score yet, the other to its worst. Covered further down, it's exactly the kind of prompt a culture barometer is meant to surface.
Cycle 4's score (75.6) sits below Cycles 2 and 3, but it's essentially level with Cycle 1, which was 100% named. If anonymity itself were softening or souring the responses, Cycle 4 should look different from the named cycles. It doesn't; it looks like Cycle 1. The score moves cycle to cycle regardless of who can see the respondent's name.
| Cycle | Entries | Unique people | % revealed identity | EHS Score |
|---|---|---|---|---|
| Cycle 1 | 108 | 106 | 100% | 75.9 |
| Cycle 2 | 63 | 63 | 100% | 83.7 |
| Cycle 3 | 70 | 69 | 99% | 81.4 |
| Cycle 4 (anonymous) | 80 | 80 | 39% | 75.6 |
A fully-named cycle (Cycle 1) produced the same low score as the anonymous one (Cycle 4). That rules out "people got less honest once they were named" as the explanation for the Cycle 4 dip. The real driver, covered next, has nothing to do with anonymity at all.
Looking at Hopton On Sea and Five Lakes separately tells a very different story from the combined number above.
Hopton On Sea actually recorded its highest score of any cycle in Cycle 4 (84.2). Five Lakes dropped to its lowest by a wide margin (62.1), over 15 points below its own Cycle 1 baseline. The combined figure of 75.6 was averaging two opposite stories into one number.
The advocate effect explains part of this: Five Lakes' Cycle 4 respondents, whether returning or brand-new, both scored lower than Hopton's equivalent groups, not just the new arrivals. Most of the low scores at Five Lakes came in without any written comment, so the data doesn't tell us why. This looks worth a direct conversation with Five Lakes site leadership rather than something to conclude from the numbers alone.
Tracking individual respondents across cycles (not just headline counts) shows a clear pattern: how someone rated their happiness in Cycle 1 strongly predicts whether they ever check in again.
A small group of 10 people has answered every single cycle so far. Their average score across every entry they've ever submitted is 4.85 out of 5, essentially as positive as the scale allows. That group alone anchors a meaningful share of the "top box" result each cycle. Meanwhile, of everyone who scored themselves 1 or 2 in Cycle 1, only 27% ever responded again, compared to 61% of those who scored 4 or 5.
Put simply: it's not that sentiment improved between Cycle 1 and Cycle 3. It's that the pool of people still answering narrowed toward the people who were already happiest, while the least happy respondents from Cycle 1 largely didn't come back to be counted again.
This is the headline number, independent of anonymity or the advocate effect above. EHS has never reached more than about one in eight employees in a single cycle.
Raw entry counts make Hopton look like the more engaged site every cycle, but that's simply because it's roughly 500 people against Five Lakes' 300. Once adjusted for headcount, the two sites track almost identically: both hover in the same 7-14% band cycle to cycle, with the lead switching sides depending on the cycle. Site isn't a meaningful driver of participation here; the shortfall is company-wide, not localised to one resort.
A few further reads on the same dataset. None are conclusive on their own, but each is worth factoring into how EHS gets discussed and used going forward.
40% of Cycle 4's respondents had never appeared in Cycles 1–3 at all. Their average score (3.78) was noticeably lower than the returning "regulars" in that same cycle (4.19). That's consistent with anonymity doing exactly what it's meant to: giving previously silent or more critical people room to speak. It just hasn't happened at large enough scale to shift the overall number yet.
57 QR check-ins came from just 21 people, and two individuals alone account for 30 of those 57 entries. QR was intended to capture ad-hoc, in-the-moment sentiment from anyone passing a code. In practice it's mostly the same few people checking in repeatedly, not broad casual reach.
It had the highest participation (12.8%) of any cycle to date, and its score (75.9) sits notably below the two campaigns that followed. If reach is what determines how representative a score is, Cycle 1, not Cycle 2 or 3, may be the closest thing to a true baseline for the wider workforce's sentiment.
Once adjusted for headcount (Hopton ~500 staff, Five Lakes ~300), both sites track in almost the same 7-14% band every cycle, with the lead switching sides cycle to cycle. Retention of Cycle 1 respondents is also close (60% Five Lakes vs 52% Hopton), a modest gap on this sample size, not a meaningful site effect. Whatever is limiting reach, it isn't specific to one resort.
Overlaying EHS scores against active vs. inactive staff numbers hints at a correlation between low sentiment and people leaving shortly after. That's a plausible read given what the data shows here, but it's a hypothesis, not a finding: it needs checking against actual leaver/starter records before it's treated as anything more than a lead worth following up.
Across all 378 entries, a small number of themes recur consistently, and they don't all carry the same feeling. Each bar below shows the mix of positive, neutral, and negative scores behind that theme, not just how often it comes up.
The signal is stable and honest. The two things actually moving are how many people respond, and who: a shrinking, increasingly self-selected group rather than a broad cross-section.
Email and SMS are active invites, but the QR code and check-in link work passively too: on a staff noticeboard, ZenZap messages, a footer in internal emails. Reprint QR materials with the current cycle's question.
The people least likely to return are the ones who were least happy last time. Winning them back, not just the enthusiastic core, is what would make the score genuinely representative again.
Staffing, appraisal clarity, whatever tops the list: fix it and say so back to staff. That's what gives a dissatisfied respondent a reason to check in again next time.
Managers haven't been told plainly what EHS is now for, or why anonymity was introduced. A short, direct message explaining the shift would do more for genuine participation than another reminder to staff.