The owner of a 20-100-person company rarely has an HR analyst, but almost always has gut feelings: “something’s off in sales,” “new hires don’t stick around,” “we’re paying too much for what we get.” These feelings are often right, but decisions based on them come late and cost a lot.
This article covers seven questions a leader already asks every month, and how to get answers to them in numbers. Not forty textbook metrics, but seven answers, each of which leads to a specific decision. It also covers one event log from which all seven answers can be calculated without manual consolidation.
Why “gut feelings” no longer work in 2026
The 2026 labor market doesn’t forgive slow decisions about people. According to the “Labor Market Barometer” survey by the Ukrainian job site GRC.UA (as reported by Finance.ua; the original report has not been made publicly available), 82.6% of employers plan to raise salaries in 2026, more than 90% are continuing to hire, and retaining the team is named among the three main business priorities for the year. In other words, the market is paying more, and your people can see it.
In this situation, decisions based on gut feeling come too late: a leader notices a problem when the person is already looking for a job, not when they first started to get worn out. Numbers notice it earlier, if you look at them regularly.
There is a second reason. When an owner asks “what’s our turnover for the last six months?” and the answer takes three days of merging spreadsheets, people stop asking. Data that is hard to get doesn’t get used. That’s why the seven questions below were chosen so that the answers can be available at all times.
Here are the seven questions: how many people leave and from where; whether new hires stay; how the company is growing; what the team costs; how many people are absent; whether people are satisfied; and what needs attention today. Each one is worth looking at by department and by manager, not just for the company as a whole.

Question 1. How many people leave, and from where
Turnover fits in one line: the number of people who left during a period, divided by the average headcount for that period, multiplied by 100. Average headcount is the average of the headcount at the start and at the end of the period, not today’s staff count. Example: a company had 44 people at the start of the year and 56 at the end, so the average is 50; 8 people left during the year, so annual turnover is 16%.
The company-wide average doesn’t tell you much. The useful breakdown is different: by department, position, office, and, above all, by manager. If company turnover is 16% but two departments are at 5% and 35%, the problem isn’t “the market,” it’s a specific place. The breakdown by manager often shows what everyone in the company suspects but no one says out loud.
The second useful breakdown is the reason for leaving, recorded at the time the person leaves. Voluntary or the company’s decision; pay, the manager, workload, or relocation. Without a reason, the turnover number shows the scale of the problem but doesn’t suggest what to do. If your tool doesn’t have such a field, a separate column in the event log (more on that below) is enough.
There is no public Ukrainian data on “normal” turnover by industry, and benchmarks from other countries vary widely by field. So compare yourself with yourself: this quarter against the previous one, this department against the others.
Question 2. Do new hires stay
New-hire retention shows what share of the people who started in a given period are still working after 3, 6, and 12 months. If eight of the ten people hired in the first quarter are still working three months later, three-month retention is 80%. The group is defined by start date, so you need only two dates per person: start date and departure date.
This number answers a question nobody usually owns: is your onboarding working, and are you hiring the right way? If people leave in the first three months, the problem lies in the expectations set during the interview or in their first weeks at the company. If they leave within the first year, the cause is more often the manager, pay, or lack of growth.
A reference point from the US: in a 2018 Jobvite survey (1,500 job seekers), about 30% of job seekers had left a new job within the first 90 days at least once. That’s the share of people over a career, not the share of hires per year, and it isn’t Ukrainian data, so treat the number as a sense of scale, not as a norm. In your own company, you can know this number precisely.
What to do with the result: if three-month retention is low, look at onboarding and at what was promised in the interview. If 12-month retention is low, look at pay relative to the market and at the manager.
Want to see turnover by department and manager without merging spreadsheets? Request a demo.
Questions 3-5. How the company is growing, what the team costs, how many people are absent
Three questions about team size and cost that an owner has to answer to finance, to investors, and to themselves.
Question 3. How the company is growing. Headcount at the end of each month, the number of hires and departures per month, and the trend over the year. These three series of numbers show whether growth is real or the company is just replacing those who left. A company that hired 20 people and lost 18 didn’t grow; it spent a year hiring. Where the hires came from is a separate question, covered in a separate article on the hiring funnel.
Question 4. What the team costs. Total payroll and payroll by department, plus median salary by department and position. The median matters more than the average: two high salaries for managers distort the average, but not the median. The median by position, next to the salary ranges in your own job postings, shows whether you’re paying new people more than those who have been with you for a long time. This is a common reason experienced people leave, and it’s rarely seen until someone does the math.
Where to get salary data: in small companies it often lives with the accountant or in a payroll service, not with HR. For this question, it’s enough to copy every salary change, with its date, into the event log once a month.
Question 5. How many people are absent. Absence rate: days of absence divided by working days in the period, by department and absence type (vacation, sick leave, personal reasons). Example: a department has 10 people and 22 working days, for a total of 220 person-days; 11 days of absence give a rate of 5%. A department with twice as much sick leave as the others is often signaling a problem with workload or the manager, not with health. Vacation left unused at the end of the year sends a separate signal: people aren’t taking time off, or are afraid to.
Three questions, three tables that finance and HR usually put together separately and in different ways. When headcount, salaries, and absences are calculated from one source, at least they match each other.
Questions 6-7. Are people satisfied, and what needs attention today
Question 6. Are people satisfied. The simplest place to start is eNPS: one question, “How likely are you to recommend our company to a friend as a place to work?”, on a scale from 0 to 10. The score equals the share of people who gave 9-10 minus the share who gave 0-6.
On its own, the score says little; what matters is how it changes between survey rounds. If the score in a department with a few dozen responses dropped 20 points in a quarter, that’s a signal to check with a conversation, not with the next survey. The breakdown by department matters here as much as it does for turnover, with one limit: the survey is anonymous, so results for groups with fewer than 7-10 responses aren’t shown, and small departments are combined. Otherwise the responses stop being anonymous, and people stop responding.
Question 7. What needs attention today. This is the question an owner asks every morning: what will I hear about late in a chat, even though I could have known in advance? Who is out today, whose probation period ends this week, which vacation requests are waiting for a decision, and which onboarding step each new hire is at. For a 50-person company, that’s five to ten lines a day.
In a spreadsheet, this is a single “Today” tab that HR or the office manager updates each morning in five minutes from the absence calendar and the list of new hires. This summary most often becomes a leader’s first way into HR data: not a quarterly report, but a short list for today, from which the habit of looking at everything else gradually grows.
How to get answers without merging spreadsheets: the event log
You don’t have to gather the answers to the seven questions if they come out of everyday processes. A vacation request, a new hire’s first day, a departure, a survey, a change in position or salary: each of these events is data. If an event is recorded once, at the moment it happens, all seven answers can be calculated from it with filters, without manual consolidation.
The opposite situation is familiar to most 20-100-person companies: vacations in one spreadsheet, salaries with the accountant, candidates in email, surveys in a separate service. Then every answer to the owner’s question turns into half a day of manual work, and after a few months people stop asking.
The minimal structure that makes all seven answers possible is one event log with seven columns: date; employee; event (start, departure, position change, salary change, absence, survey result); value (new position, new salary, type and number of days, score); department; manager; reason (for departures). Turnover then becomes a filter on the “departure” event and department, new-hire retention becomes two events per person, and absence rate becomes the sum of values for the “absence” event.
The tool is secondary here. A spreadsheet where this log is kept with discipline is better than any system where it isn’t kept. The difference between a spreadsheet and an HR management system lies elsewhere: the system records the event at the moment a person submits a request or takes a survey, so the log doesn’t need someone to maintain it, and the answers don’t go stale.
Frequently asked questions
What turnover rate is considered normal?
There are no public Ukrainian norms by industry, and benchmarks from other countries differ by a factor of two depending on the field, so quoting them as a norm would be misleading. The practical answer: compare yourself with yourself. This quarter’s turnover against the previous quarter’s, and each department against the others. A department where turnover is twice the company average needs attention regardless of any “norm.”
How do you calculate new-hire retention?
Take everyone who started work in a given quarter. Count how many of them are still working after 3, 6, and 12 months, and divide by the number who started. You’ll get three percentages for each group of hires. All you need is each person’s start date and departure date, meaning two events in the log.
How many metrics does a 50-person company need?
The seven questions in this article cover what an owner actually decides: who to retain, how to pay people, where there’s a problem with a manager, whether onboarding works. Add more metrics only when a decision comes up that can’t be made without them. A metric with no decision behind it stops getting looked at within two months.
How often should you look at the data?
Review the “what needs attention today” summary daily: it takes a minute. Turnover, retention, absences, and headcount once a month. eNPS once a quarter or once every six months: any more often and people stop responding. Payroll and medians at every salary review, usually once or twice a year.
Do you need a dedicated HR analyst?
Not for the seven questions in this article. All of them can be calculated with simple arithmetic from one seven-column event log. An HR analyst is needed when a company wants to test hypotheses across hundreds of people: what exactly predicts departures, how performance reviews relate to productivity. For a 20-100-person company, that’s premature.
How is HR analytics different from an Excel report?
A report answers a question that was asked last month and goes out of date the day it’s made. Analytics answers the question being asked today, because the data is updated by events, not by manual consolidation. A spreadsheet can be analytics if the event log in it is kept daily; a report in a system can be dead if events don’t make it in.
Try tracking absences, onboarding, and surveys in one place, so the event log fills itself in. Free for 14 days.



