Common questions.
The data-protection detail is on its own page, and every agreement is in the due-diligence pack.
Your data
How long does the questionnaire take, and do we have to finish it?
About an hour. Most questions are one click, “not applicable” counts, and you can stop and come back. Your first answer produces a comparison immediately. Scenarios opens at halfway; peer comparison in full at 80%.
Do we need to integrate anything?
No. It’s a guided questionnaire, and we benchmark policies rather than employee records, so there is no HRIS integration and far less for IT to review.
Can other members see our data?
No. Members see anonymised aggregates only. When fewer than three employers answered a question, the whole card is hidden, and your raw submissions are never exposed.
Where is it hosted, and does the AI see our data?
On Amazon Web Services in London, with multi-factor authentication for every user; lumi is Cyber Essentials certified. Ask lumi and the written reads use Anthropic as a sub-processor under the UK addendum, on aggregated and derived figures only, and nothing is used to train models. You can turn the AI features off for your organisation. Every agreement is in the due-diligence pack.
The benchmark
Where does the data come from?
The reward data of UK organisations that shared it with us directly, quality-assured against our anchor register of graded published UK sources. As members join, their answers join the same pool. Nothing is scraped and nothing is bought.
Who are “our peers”?
Your sector and size band by default. You can switch to all peers, to the twelve employers most like you, or draw your own group. Every figure names the group and the number of employers behind it, and you are never counted among your own peers.
Why is it a community and not a subscription?
Because the data doesn’t exist unless members put it in. Every member contributes, and the pool is the product.
The package and Scenarios
Are the package scores real money?
No. Cost, value and efficiency are modelled from your answers on lumi’s ratings of each benefit, then placed as percentiles among peers’ packages. They are estimates, not £ amounts and not an employee survey, and the page says so wherever they appear. The benchmark figures beside them are not estimates: they are what members answered.
What does a difficulty rating mean?
Whether the change is contractual, whether it needs consultation, and how long it takes to land. Easy, moderate or hard, with the reasons one click away.
Can we save and share a scenario?
Yes. Save as many as you like, comment on them with colleagues, set one as your target, export the CSV, print it, or present it full screen. A share link is read-only.
Signals and aims
What if we’ve never decided where we want to sit?
Most organisations haven’t, and you don’t need to before you start. Everyone begins on market. If you deliberately lead in one category or sit below in another, set that aim in Settings in a minute, and lumi reads that category against it.
How do the signals decide what to tell us?
A signal fires when something crosses a line: your percentile moves band, or a gap to the market widens. They are ranked by distance from the aim you set, and come as a weekly digest by default.
Commercial
What is the founding offer?
Join by 31 December 2026 and your access is free until 31 March 2027. From 1 April 2027, £5,000 + VAT covers the next two years. After that, membership renews at the standing rate of £5,000 + VAT a year. Read-only access has no free period.
Do you benchmark salaries?
No, deliberately. lumi benchmarks the rest of reward. It runs alongside the salary survey you already have.
We are a group with several businesses. Is that one membership?
Yes. Each business becomes a unit of your membership with its own data, peer group and benchmark, and your team switches between them from one login. Units are never counted as each other’s peers. Up to five are included.
What happens to our data if lumi goes away?
Export it any time: every benchmark, register, dashboard and scenario downloads as CSV, Excel or PDF. Never sold, never used to train AI models.
Not answered here? Ask us.