HR Audit: How to Assess Your HR Processes and Decide What to Fix First

Most business owners find out about people problems once they’re already visible: a top salesperson leaves, a vacancy enters its third month, department heads argue over who’s to blame for a bad hire. Everyone has an explanation — “the market is overheated,” “HR keeps sending the wrong people,” “young people don’t stay.” Sometimes the explanation is right. But when decisions are made on the first theory, a company risks raising salaries where people are actually leaving a manager, or replacing a recruiter when the real issue is that candidates get a misleading picture of the job.

An HR audit lets you test those explanations before you spend money on solutions. It isn’t a document inventory or a hunt for someone to blame. It’s a structured review of how well hiring, onboarding, performance management, rewards and line management support the company’s goals.

This article walks through a logic you can use to run a first diagnostic yourself:

• Signal: what management has noticed.
• Data check: whether the numbers confirm the signal and where it’s concentrated.
• Root cause: which hypotheses explain the problem and which one holds up.
• Impact: what the problem is costing the business.
• Priority: what to fix first.
• Plan: the action, the owner, the deadline, the metric.
• Follow-up: whether anything changed after implementation.

Which type of HR audit do you need?

“Audit” can mean very different reviews. The choice determines what data you collect and what you get out of it.

These reviews complement each other but don’t replace one another. Flawless contracts won’t explain why new hires quit after a month. And a precise diagnosis of your hiring won’t protect you from a fine for an unregistered contract.

Signs your HR processes need a closer look

Some situations are almost always worth investigating:

• resignations keep happening in the same roles or the same department;
• people leave within their first months;
• vacancies take a long time to fill, or the time varies wildly;
• managers regularly complain about the quality of new hires;
• headcount is growing, but team output isn’t;
• responsibilities overlap and nobody fully owns the result;
• bonuses are seen as unfair or unrelated to performance;
• HR reports on the number of tasks completed, but its impact on the business is unclear;
• people decisions are made case by case, without shared rules or data;
• the company is preparing to scale, open a branch or launch a new line of business.

Below are four common signals. For each: what management sees, what could be behind it, and how to check.

New hires leave in their first months

What management sees: hiring never stops, yet the team doesn’t grow.

Possible causes: the job was described differently in the interview; nobody onboards the new hire; the line manager gives no feedback; candidates are screened against the wrong criteria; starting pay turned out lower than expected.

What to check: the share of hires who leave within 90 days, tracked by hiring cohort and broken down by department and manager; who initiated each exit; what was promised in the job ad and interview; whether new hires have a plan for their first weeks and someone responsible for it.

The trap: leavers often give the safe answer — “I got a better offer.” If exits cluster under one manager while pay is the same across teams, money is unlikely to be the real reason.

Vacancies take too long to fill

What management sees: a department runs short-staffed for months and everyone else carries the extra load.

Possible causes: the role requirements were never agreed; few candidates on the market have that experience; the offer is below market; a hiring manager is slow to review candidates; candidates drop out at the final stage.

What to check: at which stage time is lost — from request to first candidates, from interview to decision, from offer to start date; how many candidates declined and why.

The trap: the knee-jerk reaction is to change recruiters or sourcing channels. If most of the time is lost on internal approvals, a new agency won’t change a thing.

Headcount grows, output doesn’t

What management sees: the team got bigger, but sales or production stayed flat.

Possible causes: roles overlap and there isn’t enough work for the new people; there are no clear goals or success criteria; the manager can’t keep up with a larger team; output is limited not by headcount but by process or demand.

What to check: output per employee over time; how tasks and authority are distributed; the manager’s span of control and how often they actually work with their reports.

The trap: concluding that “people aren’t performing” doesn’t explain why. Weak team results are more often down to how work is organized than to individual employees.

Bonuses are seen as unfair

What management sees: complaints about bonus calculations, disputes, strong performers leaving.

Possible causes: the formula is unclear or keeps changing; bonuses depend on things the employee can’t influence; strong and weak performers get the same; people in identical roles are paid differently for no clear reason.

What to check: whether employees can calculate their own bonus; the spread of payouts within the same role and how it relates to performance; how payouts have changed over recent periods.

The trap: increasing the bonus pool won’t fix a fairness problem if the formula stays the same.

In many companies in Uzbekistan, especially small and mid-sized ones, the HR function is limited to personnel administration and recruitment, while people decisions are made by the owner or department heads. At 20–30 employees, this works: the owner knows everyone. As the company grows, informal arrangements stop working, and problems surface in hiring, pay and management all at once. This is one of the most common triggers for a first review of HR processes.

What an HR audit covers

A full audit usually spans eight or nine areas. Not all of them matter equally for every company: focus on those with the biggest impact on quality of hire, team performance, workforce stability and how manageable the business is. Use the table as a map — mark the areas where you see warning signs and start there.

Other areas — learning and development, internal communications, engagement, employer brand, workforce planning — are included when they relate to the question the company is asking. If the goal is to understand why salespeople leave, a deep dive into the accounting team’s training program isn’t needed.

For the documentation area, keep in mind that employment contracts and any amendments must be registered in the Unified National Labor System (ENST), and a paper contract within five working days of signing. The process is covered step by step in CA Consulting’s guide on how to hire an employee in Uzbekistan.

How the processes connect

You can’t assess these areas in isolation: a problem in one often shows up in another. Take high turnover in a sales team. It could be driven by:

• recruitment — hiring people with no outbound sales experience because the role has to be filled fast;
• the job description — candidates are quoted earnings a new hire can’t realistically make in the first months;
• onboarding — new hires get a full quota from day one with no training;
• the manager — one team lead loses people noticeably more often than the others;
• compensation — the bonus depends on a metric a new hire can’t influence.

Each cause calls for a different fix. Raising base pay won’t help if people are leaving because of their manager. A new onboarding plan won’t help if the wrong candidates are being hired. That’s why an audit doesn’t draw conclusions about causes until several hypotheses have been tested.

What data to collect

Data is there to test hypotheses, not to gather “everything we have.” Before you start, frame the management question — for example, “why are new sales hires leaving in their first two months?” That question determines what data you need.

A minimum dataset for a small company

A company of 10–30 people doesn’t need HR software for a first diagnostic. Four sources are usually enough:

• A staff spreadsheet covering the last 12–24 months, with start and exit dates, department and manager.
• A vacancy log for the same period, even if you have to rebuild it from emails and chats.
• The current pay and bonus scheme, with actual payouts for a few roles.
• Five to eight interviews: with the owner, department heads and two or three employees with different tenure.

In companies where employee data lives in 1C, spreadsheets and work chats all at once, simply putting this dataset together is often the audit’s first finding: people decisions are being made without a single picture.

When numbers are enough and when you need interviews

Every diagnostic method has its limits.

Numbers are enough when the question is about scale: where turnover is higher, at which stage time is lost. Interviews are needed when the question is about causes and the numbers allow more than one explanation. A reliable conclusion rests on at least two different types of evidence. If interviews say “people leave because of the manager” but the data shows turnover in that team is no higher than elsewhere, the hypothesis needs rethinking. How to triangulate interviews, surveys and HR data is covered in detail, using resignations as the example, in SPECTRE’s article “Analysing why employees leave”.

Confidentiality and how to announce the audit

Employees should know an audit is taking place and understand why. Announce it in advance: the company is reviewing processes, not looking for someone to blame. Day-to-day work continues; participants just need to set aside time for an interview or a survey.

An interview isn’t anonymous, but it can be confidential: only the interviewer sees the answers, and the report contains only aggregated findings that can’t be traced back to anyone. Promise only what you can actually guarantee. Employee information is personal data: restrict access to it, use anonymized data in analysis spreadsheets, and check with a lawyer before choosing a survey tool that will collect personal data.

How to measure hiring effectiveness

Hiring has two dimensions: speed and quality. They’re easy to confuse. A role filled in two weeks is a poor result if the person leaves a month later. A role filled in two months can be a good result if it’s a scarce specialist who’s now in their third year.

Here are five metrics you can track in an ordinary spreadsheet.

Time to fill

What it shows: the number of days from an approved hiring request to offer acceptance or the new hire’s start date. Pick one end point and always measure against it.

How to read it: compare within the same job family and against previous periods. Break the time down by stage — sourcing, interviews, decision, offer, start — to see where it’s being lost.

What skews it: roles that sat open without active sourcing; requests where the requirements changed midway; seasonality.

What you can decide: if time is lost on approvals, speed up internal decision-making; if it’s lost on sourcing, revisit the requirements, the offer or the channels.

Share of roles filled on schedule

What it shows: how predictable hiring is. For the business, this often matters more than average speed: managers plan project launches around the promised date.

How to read it: a low share alongside a reasonable average means high variance — some roles close fast, others stall.

What skews it: unrealistic deadlines set when the role was opened.

What you can decide: agree deadlines based on the job family and the market, not “as soon as possible.”

Pipeline conversion rates

What it shows: how many candidates move from application to interview, from interview to offer, and from offer to start.

How to read it: a sharp drop at one stage points to a problem at that stage. Many declined offers suggest the offer isn’t competitive or the process puts candidates off. Few candidates passing the first interview suggests the job ad is attracting the wrong people.

What skews it: small numbers. With five applicants, percentages mean nothing — look at the individual cases instead.

What you can decide: rewrite the job ad, revise the offer, cut the number of stages or change the selection criteria. How to build a selection process around evidence of competencies is covered in “How to assess a candidate in an interview”.

Early attrition and probation outcomes

What it shows: how many new hires leave within 90 days — of their own accord or at the company’s initiative. As a general rule, probation in Uzbekistan lasts up to three months, or up to six for certain senior roles, provided it’s set out in the employment contract.

How to read it: calculate it by hiring cohort, not by the calendar month of exit — otherwise the numerator and denominator describe different people. Voluntary exits say more about onboarding and working conditions; terminations by the company say more about selection quality.

What skews it: small groups and incomplete cohorts that haven’t yet reached 90 days.

What you can decide: if voluntary exits are high, check whether the job matches its description and how onboarding works; if the company often terminates during probation, review the selection criteria.

Hiring manager assessment of quality of hire

What it shows: how well a new hire meets expectations after three to six months.

How to read it: ask the manager the same two or three scaled questions about every new hire: are they reaching expected performance, would you hire them again, how do they compare with strong performers in the same role? Cross-check this subjective view against actual performance data where it exists.

What skews it: personal bias, and the fact that the manager made the hiring decision.

What you can decide: if ratings are consistently low for one channel or one interviewer, that’s what needs to change.

There are no universal benchmarks for these metrics. A “normal” time to fill for a call center agent and a CFO differs several times over, and competition for experienced specialists in IT, finance and sales is tougher than for frontline staff. Analyze the metrics over time, by job family and in the context of your own business.

How to assess retention and performance management

These are two different tasks, but they’re linked: people often leave when they don’t know what’s expected of them or how their results will be judged.

Employee retention

An overall turnover rate tells you very little. Twenty-five percent a year can mean normal churn among frontline staff — or losing half your key specialists. Break it down:

• by department and manager — where turnover is consistently higher than in comparable groups;
• by tenure — whether it’s new hires or experienced staff who leave;
• by who initiated the exit — voluntary resignations vs. terminations by the company;
• by value to the business — whether your best performers are leaving;
• over time — whether there were spikes after a change of manager, schedule or bonus scheme.

The reason someone gives when they resign is a hypothesis. People tend to name the reason that’s safe to say out loud. Test it against other types of data: a survey of current employees in the same group, timesheets, payroll records, team comparisons. How to run a first diagnostic of staff turnover is covered in “Why employees really leave”.

Not every resignation is an HR failure. A weak performer leaving, or someone moving to another city, is a normal part of company life. And falling turnover isn’t always good news: it also drops when a company stops hiring, or when the only people left are those with nowhere else to go.

Performance management

To find out whether your performance review system works, answer four questions:

• Do employees understand what results are expected of them? Ask several people in the same role to name the three main metrics for their job. If their answers don’t match each other or their manager’s view, expectations haven’t been defined.
• Are review criteria linked to job responsibilities? Ratings on generic traits like “responsibility” or “initiative,” with no examples, say little about results.
• Do employees get regular feedback? Ask each person you interview when they last discussed their work with their manager.
• Are review results used in decisions? Take the last 10 decisions on bonuses, promotions or dismissals and check whether they relied on review results.

Signs of a box-ticking review: almost everyone gets the same high scores; forms are filled in once a year before a deadline; results don’t affect pay or development; managers can’t back up ratings with examples.

From findings to decisions

Findings on retention and performance lead to different changes:

• new hires leave because the job was described differently — rewrite the job ad and the interview script;
• new hires lack proper onboarding — a 90-day onboarding plan with an assigned mentor;
• turnover is concentrated under one manager — feedback to that manager based on aggregated data, and management skills development;
• employees don’t understand expectations — a short list of metrics for each key role;
• bonuses aren’t linked to results — rework the formula, not just the amount.

HR audit checklist: a 30-minute self-assessment

Answer each question for your company: “yes,” “partly,” “no” or “not enough data.” The checklist won’t give you a verdict on your HR function as a whole. It shows which processes to look at first.

How to read the results. “No” and “not enough data” tell you where to look, but they don’t prove a process is failing. A 15-person company can manage without a formal review system if the owner personally and regularly discusses everyone’s work with them. “Not enough data” is often more telling than “no”: it means decisions in that area are being made blind. If you have many negative answers spread across different areas, a full diagnostic makes sense. If they cluster in one area, start there.

How to prioritize what you find

Once the data is in, you’ll have a list of problems. Trying to fix everything at once is a sure way to never know what worked. Rate each problem as high, medium or low on five criteria:

• Business impact: what the company is losing — revenue, time, customers, key people.
• Urgency: whether losses will grow if nothing is done for another quarter.
• Scope: how many employees, departments and processes are affected.
• Confidence in the cause: whether the cause is confirmed by different types of data or is still a theory.
• Cost and feasibility of a fix: what can change within a month with no budget, and what requires redesigning a system.

Based on the ratings, problems fall into three groups:

• critical — high impact and growing losses; need action within weeks;
• priority — noticeable impact without a sharp rise in losses; addressed in the next planning cycle;
• systemic — affect many processes and require sustained work over several months.

If confidence in the cause is low, the first step isn’t a change — it’s verification. An expensive fix based on an untested hypothesis is the biggest risk at this stage.

Example: symptom vs. root cause

This is an illustrative example. In a wholesale distribution company, seven people left an 18-person sales team in six months — six of them within their first two months. The sales director asks for a pay rise: “the market is overheated, competitors are poaching our people.”

The data tells a different story. Experienced salespeople aren’t leaving, and their earnings are in line with the market. In exit interviews, new hires say their income turned out lower than promised. At interview, they were quoted the average earnings of experienced reps, while new hires carry the same sales quota as everyone else from month one — so they earn almost no bonus in their first months.

The root cause isn’t the level of base pay. It’s how earnings are described during hiring and the lack of a reduced quota during ramp-up. Raising salaries across the team would be expensive and wouldn’t touch the cause. The place to start is an honest description of first-months earnings and a quota that ramps up to full target in stages.

What an action plan looks like after the audit

A good HR audit doesn’t end with a list of shortcomings but with a few well-founded priorities. For each one, the plan answers seven questions: what the problem is, what evidence supports it, what the likely cause is, what will change, who owns it, by when, and which metric will show the result.

Quick wins and systemic change

Some measures deliver results within weeks and cost nothing: rewriting a job ad, setting deadlines for candidate feedback, assigning a mentor to new hires, properly documenting contract changes. Others require training managers, redesigning the reward system or changing the structure — these are rolled out in stages, starting with a pilot in one department.

How to track results

• Set a baseline before any changes: the target group’s metrics over several previous periods.
• Compare the same group in the same season: if the change was in sales, look at early turnover in sales, not across the company.
• Use more than one metric. Early exits may fall simply because the company paused hiring. Track them alongside quota attainment and managers’ ratings of new hires.
• Watch for side effects. A reduced quota for new hires may frustrate experienced reps if it isn’t explained.
• Schedule the follow-up measurement when you launch the change. For early turnover, that’s when the next cohorts reach 90 days.

In-house review or external expert?

An in-house review has real strengths: an HR professional knows the context, the history of decisions and the people, and will be the one implementing the changes anyway. The limitations are just as clear. Someone who works inside a system every day gets used to its flaws. It’s hard for an employee to tell the owner that some problems stem from leadership or from HR itself. And people aren’t always candid with a colleague who has a say in their career.

An external expert adds value when there isn’t enough data to draw conclusions, the problem spans several connected processes, people explain the causes differently, or leadership struggles to separate symptoms from root causes. They bring a view of how other companies handle the same issues, and employees tend to speak more openly with them. But independence doesn’t automatically guarantee accuracy: an outside specialist needs time to learn the context, their conclusions depend on access to data, and ownership of the changes stays inside the company. How to choose the right HR model for your business — an in-house HR manager, an HR director or an external consultant — is covered in “HR manager, HR director or external consultant”.

The bottom line

An HR audit is worth doing when it ends in decisions. Its value isn’t in describing the state of HR but in answering three questions: where the company is losing people and money, why it’s happening, and what to fix first. You can start small: a staff spreadsheet with dates, a vacancy log, a few interviews and the checklist above. That’s often enough to see which process deserves a closer look.

When there are several signals, the data is contradictory or you need an independent view, a SPECTRE HR audit can help. We review structure and role allocation, recruitment, onboarding, performance management and training, HR metrics, and the workplace climate based on an employee survey. You receive findings, a risk assessment and a prioritized change plan, with implementation support if needed. If the diagnostic shows you need to build specific processes, such as a recruitment or retention system, these are part of our HR services.