Choosing HR software: the question set for any vendor
Echipa HR 365 · reviewed 2026-09-02 · 5 min read
Feature lists look identical across vendors, because they are written to look identical. The real differences appear in twelve uncomfortable questions, four of which usually eliminate half the options before the second demo.
The four questions that eliminate
They come first, at the first demo, and they shorten the process by weeks. Ask to see, not to be told — for each of them, the right answer is a demonstration on screen, not a promise.
- What is adapted to the reality I work in, and what is not? Leave types, the timesheet form, the identification code, the currency. An international system that “can be configured” usually means you will configure it.
- What happens to people with no work email? In production, retail and hospitality, that is half the team. If the answer is “we create accounts for them”, ask how somebody logs in when they have nowhere to receive the password.
- How do I get all my data out if I leave, in what format and in what time? An evasive answer here is the clearest alarm signal in the whole set.
- Who can see salaries, and what is left behind when somebody opens them? If there is no access log, sensitive information is visible with no trace.
The rest of the set
| The question | A good answer | An alarm signal |
|---|---|---|
| How long does importing a headcount of 200 take, and what stays manual? | An estimate in days and a concrete list of what does not import | “We will import everything”, without having seen the file |
| What comes across from history: balances, tenure, documents? | X comes across, Y does not, with the reason | “Anything can be imported” |
| What is the real annual price at my headcount, and what changes it? | A figure and a list of what moves it | Price on request only, after three conversations |
| What happens if headcount rises or falls by 20%? | The billing rule, written down | “We will discuss it then” |
| Who answers a problem and within what time, in writing? | A deadline and a channel, in the contract | “We are very responsive” |
| What happens to leavers’ data, and after how long? | A configurable retention policy | It has never come up before |
| Which integrations work today, at a real customer? | A concrete, verifiable example | A list of logos with no detail |
| What have you refused to build in the last year, and why? | A real example, with the reasoning | “We do whatever the customer asks” |
The last one is the most useful in the whole set. A vendor who has never refused anything either has no customers or says yes to everything and delivers half. A good answer sounds like a product decision, not an excuse.
Who has to be in the conversation
Three roles, three different moments
- HR — leads the evaluation and tests the daily flows — not the impressive features, but the leave request and the monthly timesheet
- A team manager — joins the second demo and tries exactly what they will do; if it looks complicated to them, the team will not use the system
- IT or the data owner — receives questions 3 and 4 from the set, in writing, and the answers in writing too
The second row is the best predictor of whether the implementation succeeds. A system chosen by HR alone, however good, hits the managers who never log in by month two — and then the data empties out and everything goes back to email.
What you do in a trial in order to learn something
- Do not test with invented data. Take ten real people, with their complicated situations, and see whether they fit into the system without manual exceptions.
- Run a full flow end to end: a leave request submitted by a real employee, approved by their manager, landing in the records.
- Close a month of timesheets. It is the best test, because it touches the most corners of the system at once.
- Ask somebody from the shop floor to open what they need to open, on their own phone, with no explanation from you beforehand.
The last point is the harshest test and the most rarely run. A system that HR understands perfectly and that somebody in the warehouse cannot open on their own will not have correct data — however good it looks at the demo.
What goes in the contract, not just in conversation
Good answers at the demo are worth nothing if they stay verbal. Four things are worth writing down, and they are exactly the ones a serious vendor accepts without argument:
- The deadline and format in which you receive a full export of your data, including at contract termination.
- The response time for a problem that blocks an operation, distinct from the one for an ordinary question.
- The billing rule when headcount varies, in both directions, with the thresholds written out.
- What happens to the data after termination: how soon it is deleted and what confirmation you receive.
The fourth point is the most often omitted and the most awkward to recover later. A vendor who cannot say in writing what they do with your data after you leave is telling you they have not thought about it — and that is information about the rest of the product.
How many vendors do I compare?
Three is the practical number. Below three you have no basis for comparison; above five, the demos blur together and the decision is made on the last one seen, not the best one.
Is one system that does everything better than several specialised ones?
It depends how many flows touch each other. If timesheets, leave and the personnel file use the same data about the same people, a single system removes the manual syncing — which is usually the work you were trying to eliminate.
A demo in which we go through all twelve questions together
Thirty minutes on your real flows, including the questions we answer least well — so you also learn what we do not do, not only what we do.
Or create an account directly: every module for 7 days, no card required.