Attrition Analysis: Calculate, Diagnose and Cut Turnover

Attrition Analysis: Calculate, Diagnose and Cut Turnover

Most Indian SMBs feel attrition before they measure it. A key engineer resigns in the week after appraisal letters go out, a sales team in Indore loses three people in a month, and suddenly the founder is asking HR "why is everyone leaving?" Attrition analysis is the discipline that turns that anxious question into a set of numbers, segments and root causes you can actually act on. It covers how you calculate your employee turnover rate, how you split it into meaningful categories, how you find out what is really driving exits, and how you put a rupee value on the problem so leadership takes it seriously. This guide walks through the full process, from the attrition rate formula to a 90-day plan, with an Indian SMB lens throughout.

Why Attrition Analysis Matters More for SMBs

A 5,000-person IT services company can absorb a bad quarter of exits. A 60-person D2C brand in Jaipur cannot. When headcount is small, every departure is a larger share of your capacity and client relationships, and the pipeline to replace them is usually one HR generalist and a referral WhatsApp group. The upside is that SMBs can move fast once they know what is wrong: a pay band or a team structure can be fixed in weeks. Attrition analysis tells you which lever to pull.

Attrition Rate Formula: How to Calculate Employee Turnover Rate

Before you can diagnose anything, you need a consistent number. The core attrition rate formula is simple, but the details (what counts as a "separation", which headcount you divide by, what period you use) trip up many HR teams. Get these decisions written down in a one-page definitions document so that the rate you report in April is comparable to the one you report in October.

The basic attrition rate formula

At its simplest:

Attrition rate = (Number of separations in the period ÷ Average headcount in the period) × 100

Average headcount is usually (opening headcount + closing headcount) ÷ 2. If your headcount swings a lot within a month (common in seasonal businesses or during a campus-hire intake), use the average of daily or weekly headcounts instead, which most HRMS platforms can compute automatically.

Monthly attrition rate

Monthly attrition is your early-warning signal. It is noisy in a small company (one exit in a 40-person firm is 2.5%), so look at it as a trend line rather than a single figure. Plot it for the last 12 to 24 months and you will often see the appraisal-cycle spike and the post-Diwali bonus exits jump straight off the chart.

Worked example: a 120-person Pune SaaS firm starts March with 118 employees, ends with 122, and sees 4 people leave during the month. Average headcount is 120, so monthly attrition is (4 ÷ 120) × 100 = 3.3%.

Annual attrition rate

Annual attrition is what boards, investors and benchmarks refer to. Calculate it two ways and be clear which one you are using:

• Trailing twelve months (TTM): separations in the last 12 months ÷ average headcount over those months. This is the most honest number and the one most benchmark surveys use.
• Annualised monthly: monthly rate × 12. A quick projection that exaggerates seasonal spikes; annualising a 3.3% March rate to 40% is misleading if March is always your worst month.

Voluntary vs involuntary attrition

Voluntary attrition is when the employee chooses to leave: resignation, moving abroad, joining a competitor, going back to studies. Involuntary attrition is when the company initiates the exit: performance termination, redundancy, contract end, disciplinary action, and in some definitions, retirement and death.

Track these separately because they tell opposite stories. High voluntary attrition points to problems with pay, managers, growth or culture. High involuntary attrition points to problems with hiring quality, onboarding, or performance management. Rolling them together hides both.

Regretted vs non-regretted attrition

This is the single most useful split for an SMB, and the one most companies skip. A regretted exit is someone you would have wanted to keep: a high performer, a person in a critical role, or someone whose knowledge is hard to replace. A non-regretted exit is someone whose departure the manager and HR agree was neutral or even positive.

The mechanics are simple. At the time of resignation, the manager and HR jointly tag the exit as regretted or non-regretted in the HRMS, before the notice period starts and before anyone's memory gets rosy. A company with 20% total attrition but 5% regretted attrition has a very different problem from one with 12% total and 10% regretted.

Early attrition (first-90-day and first-year)

Early attrition measures people who leave within a defined window after joining. Two windows are worth tracking:

• First 90 days: exits here almost always point to a mismatch created in hiring or onboarding: wrong role description, oversold offer, no laptop on day one, a manager who did not know they were joining.
• First year: exits between month 4 and month 12 often point to broken expectations around growth, learning, or pay revision timing.

Early attrition rate = (Joiners who left within 90 days ÷ Total joiners in the cohort) × 100

Note that this is a cohort measure (based on joiners), not a headcount measure. A 90-day early attrition rate of 15% means that of every 100 people you hired, 15 were gone within three months.

Attrition rate formulas at a glance

Common mistakes in calculating employee turnover rate

• Counting resignations instead of exits. A person who resigns in March and serves a 60-day notice leaves in May. Decide which date you count on, then be consistent.
• Ignoring contract and intern separations. If they are in your headcount denominator, their exits must be in the numerator too.
• Reporting only the total. A 22% blended rate tells the founder almost nothing. A 4% regretted rate in engineering and a 38% first-year rate in inside sales tells them exactly where to look.

Segmenting Your Attrition Analysis: Where Turnover Actually Lives

A single company-wide attrition rate is an average of very different realities. The real insight comes from segmentation. Your HRMS already holds the fields you need: department, reporting manager, date of joining, location, grade or level, employment type and source of hire. The job of attrition analysis is to cut exits by each of these and compare against the base population.

The core segments

How to read a segmented view

Suppose "Meridian Foods", a fictional 200-person packaged-foods company in Nagpur, has an annual attrition rate of 24%. Segmented:

• Sales (field): 41%
• Plant operations: 18%
• Finance and admin: 9%
• Corporate office: 11%

The blended 24% suggests a company-wide culture issue. The segmented view says the problem is almost entirely in field sales. Cutting field sales by manager shows that two of five regional managers account for most exits, and by tenure that most leavers are in months 4 to 9, right after the probation-linked incentive scheme kicks in. That is a solvable, specific problem.

Attrition Analysis Root-Cause Diagnosis: From Numbers to Reasons

Segmentation tells you where. Root-cause analysis tells you why. The mistake most SMBs make is to rely on a single data source, usually the exit interview. Exit interviews are useful but heavily biased: people leaving on good terms say "better opportunity", and people leaving on bad terms often say nothing honest because they want a clean relieving letter. Triangulate across at least four sources.

1. Exit data (interviews and exit surveys)

Run a short, structured exit survey through your HRMS on the day the resignation is accepted, not on the last working day when the person has mentally left. Keep it to eight to ten questions with a fixed reason taxonomy (compensation, manager, growth, role fit, workload, commute or relocation, personal, higher studies, health) plus one free-text box. Fixed reasons let you trend; free text lets you understand.

Then do a human exit conversation with regretted exits only, ideally by someone other than the direct manager. Ask about the moment they first started looking, not why they are leaving. The first-look moment is where the real cause lives.

2. Engagement and pulse data

If you run an engagement survey, join it to attrition data at the team level (never at the individual level, to protect anonymity). Teams whose scores on "my manager supports my growth" or "I understand how my pay is decided" were lowest six months ago are usually the teams with the highest regretted attrition today. Engagement is a leading indicator; attrition is a lagging one. If you do not run surveys, start with a quarterly five-question pulse; after three cycles you have a trend.

3. Manager span and manager quality

Pull two numbers per manager from your HRMS: their span (number of direct reports), and their team's regretted attrition over the trailing 12 months. Add a third if you have it: the percentage of one-on-ones completed on time. A team lead with 14 direct reports cannot do career conversations or timely feedback. When attrition rises steeply with span, your org structure, not the individual manager, is the problem, and the fix is a restructuring conversation rather than a blame conversation.

4. Pay competitiveness

You do not need a paid compensation survey to spot a pay problem. Use three practical signals:

• Offer-to-current ratio for leavers. In exit conversations, ask what percentage hike the new offer represented. If regretted leavers are consistently getting 40% or more, and your appraisal budget averages 8% to 10%, the market has moved and your bands are stale.
• Compression. Compare what you paid the last three external hires at a level with what existing employees at the same level earn. If new joiners are paid more than two-year veterans, the veterans will find out; they usually do.
• Counter-offer frequency. If you find yourself making counter-offers every month, your bands are wrong. A counter-offer is a symptom, not a strategy.

5. Workload and burnout signals

Your HRMS and payroll system already contain workload data if you look for it:

• Leave balances that keep accumulating (people who cannot take leave)
• Frequent unplanned leave or sick-leave spikes in a specific team
• Overtime, late log-outs or weekend timesheet entries
• Comp-off balances that are never used

Overlay these with the team-level attrition view. A team with high leave accrual, high weekend hours and rising attrition is a burnout problem, and no pay adjustment will fix it on its own.

6. F&F settlement data as a diagnostic source

Full and final settlement data is a surprisingly rich and underused source for attrition analysis in India. Each F&F record tells you:

• Whether notice was served, bought out or waived. A high rate of notice buyouts (where the employee pays to leave early) signals leavers had a competing offer with a tight joining date: a pay-and-market signal. Company-waived notice suggests managers were happy to see people go, which should be reconciled with the regretted tag.
• Leave encashment amounts. Large encashment payouts at exit mean the person could not take leave: a workload signal.
• Disputes and pending reimbursements. If leavers routinely dispute F&F, it points to documentation or sign-off problems that also erode current employees' trust.
• Recovery of bonds, training costs or joining bonuses. If you are frequently recovering these, your early-attrition problem is confirmed and the bond is not preventing it.
• Time to settle. A slow F&F process damages your employer brand among leavers, who talk to your current staff and future candidates.

7. Notice-period behaviour

Indian notice periods for permanent staff commonly range from 30 to 90 days, and how they play out is a diagnostic in itself. Track for each exit the contractual notice, the notice actually served, whether it was bought out, and whether the employee "absconded" (stopped coming without formal separation). Rising absconding rates, especially in early tenure or contract populations, usually mean your hiring pipeline is bringing in people who never intended to stay, or that onboarding is poor enough that people do not feel they owe you a formal goodbye.

For each high-attrition segment, condense all of this into a one-page diagnosis: the numbers, what exit and engagement data say, what structural data says, your root-cause hypothesis, and the intervention with the metric that will prove it worked. That page is what you take to the founder.

India-Specific Attrition Patterns Every SMB Should Model

Attrition analysis frameworks from global HR literature are useful, but the Indian SMB context has its own rhythms. Build these into your analysis explicitly.

Appraisal-cycle attrition spikes

Most Indian companies run annual appraisals with letters going out between March and June, and many pay variable or annual bonuses around the same time or around Diwali. Employees who are unhappy with their revision often already have offers lined up and resign within days of receiving the letter; others wait for the bonus credit and then resign. This creates one or two predictable spikes per year.

Your dashboard should show attrition by month with the appraisal-letter date and bonus-credit date marked. Then analyse the spike separately: were the leavers below-average ratings (a performance-management issue), or above-average ratings who got below-market revisions (a compensation issue)? The intervention is completely different.

A related pattern: employees who have decided to leave often time their resignation so the notice period ends just after the bonus payout. Tracking this timing tells you how much of your spike is "waiting for the money" versus "reacting to the letter".

Campus hires and the first-year cliff

Campus hires (from engineering colleges, B-schools, polytechnics or ITIs) are a large part of many SMBs' junior pipeline. They tend to show a distinctive curve: low attrition in the first six months (still learning, often bound by a service agreement), then a sharp rise between months 12 and 24 as they gain market-ready skills and as peers leave for higher studies or larger brands.

Track campus cohorts separately from lateral hires and compare batch-on-batch retention at 6, 12, 18 and 24 months. If a specific batch retains far worse than others, look at the onboarding and mentoring they received, not just the college.

Tier-2 and Tier-3 hiring dynamics

Many SMBs have shifted hiring to Tier-2 cities (Indore, Coimbatore, Bhubaneswar, Nashik, Vizag and others) for cost and loyalty reasons. Employees there are often more likely to stay for stability and proximity to family, so voluntary attrition can be lower. But when a large company opens a centre in the same city, or a remote-work offer from a metro employer arrives, the local pay ceiling is suddenly exposed.

Tag relocation-driven exits ("moving to Bengaluru or Hyderabad") separately from dissatisfaction-driven exits, and segment by location, so you can see whether your Tier-2 attrition is a pay-competitiveness problem or simply the natural flow toward metro opportunities.

Contract vs permanent staff

Contract, third-party-payroll and gig staff are common in operations, support, warehousing and field sales. Track their separations separately: contract-end exits are planned and should not inflate voluntary attrition, but contract staff who resign mid-term are a real signal, often about pay parity, benefits, or the belief that conversion to permanent will never happen. Conversion rate and post-conversion retention tell you whether your contract model is a genuine pipeline or a revolving door.

Statutory context

Notice periods, gratuity, leave encashment, bonus applicability and the treatment of contract labour all interact with attrition and F&F data. Statutory rules change from time to time and vary by state, so verify current government rules and your own employment contracts before relying on any assumption in your analysis.

Cost of Attrition: A Worked Illustrative Example

Founders respond to rupees more than to percentages. Modelling the cost of attrition converts your analysis into a business case for the interventions you want to fund.

Components of replacement cost

The cost of replacing one employee has five broad components:

1. Separation costs: HR and manager time on exit formalities, F&F processing, and any notice buyout you fund for the replacement.
2. Vacancy costs: lost productivity while the seat is empty, overtime or contractor cover, delayed deliverables, lost revenue in client-facing roles.
3. Hiring costs: portal and agency fees, referral bonuses, interviewer time (often the biggest hidden cost), background verification, offer-drop losses.
4. Onboarding and training costs: induction, laptop and set-up, formal training, buddy and manager time.
5. Ramp-up productivity loss: the period when the new hire draws full salary but delivers partial output. Usually the largest component for skilled roles.

Add a sixth, harder-to-quantify component for regretted exits: knowledge loss, client-relationship risk and the ripple effect on team morale and workload.

Illustrative worked example

The following example is entirely illustrative. The figures are invented for a fictional company to demonstrate the method; they are not benchmarks and should not be used as such. Plug in your own numbers.

"Kaveri Analytics" is a fictional 120-person data-services firm in Pune. It loses a mid-level analyst (CTC ₹9,00,000 per year, roughly ₹75,000 per month) in a regretted exit. Its HR team models the replacement cost as follows.

In this illustration, replacing one ₹9 lakh analyst costs roughly ₹3 lakh, or about four months of that person's salary. If Kaveri Analytics loses 20 people a year, of which 8 are regretted, the regretted exits alone represent an estimated ₹24 lakh of cost, most of it invisible in the P&L because it shows up as slower delivery and busier managers rather than as a line item.

How to build your own cost model

• Use actual CTC by level from payroll, and time-to-fill and time-to-productivity from your ATS and managers; do not guess.
• Be conservative. A defensible ₹2 lakh estimate is more persuasive than an aggressive ₹8 lakh one that a CFO can pick apart.
• Report cost of attrition by segment. "Field sales attrition is costing us an estimated ₹35 lakh a year" is far more actionable than a company-wide figure.
• Refresh annually. Ratios like "months of salary per replacement" vary widely by role, industry and city; do not import someone else's multiplier.

Building an Attrition Dashboard from HRMS Data

An attrition dashboard is the operational heart of attrition analysis. It should refresh automatically from your HRMS, payroll and ATS data, and it should answer the five questions leadership asks most often: how bad is it, where is it, is it getting better or worse, who is at risk, and what is it costing.

Step 1: Fix your data foundation

Before any dashboard, confirm that these fields are populated accurately for every employee, current and former: joining, resignation and last-working dates; separation type; regretted flag; primary exit reason; department, manager, location and level; employment type and source of hire; latest rating and revision percentage; and notice contractual, served and buyout flag.

If your HRMS does not have a regretted flag or a reason taxonomy, add a custom field. This is a 30-minute configuration task that unlocks most of the value of attrition analysis.

Step 2: Define the core metrics

Build the formulas from the table earlier in this guide as saved calculations, not as spreadsheets someone rebuilds each month. Add notice-buyout rate, absconding rate, average tenure of leavers versus current staff, and estimated cost of attrition by level.

Step 3: Design the views

A practical dashboard for an SMB has four screens:

1. Executive summary: TTM attrition (total and regretted) with a 24-month trend line, appraisal and bonus dates marked, and year-to-date cost estimate.
2. Segment heat map: department × tenure band, coloured by regretted attrition rate. This is where problems jump out visually.
3. Manager view: span, team attrition, regretted attrition and one-on-one completion. Share privately with each manager and their skip-level.
4. Early-warning view: current employees who match risk patterns, for HR follow-up.

Step 4: Add leading indicators

Lagging metrics tell you what has already happened. Leading indicators let you intervene. Practical ones from HRMS data include:

• Employees whose latest revision was below the company median with no promotion in two or more years
• Employees whose manager changed in the last six months
• Employees with high accrued leave and low leave taken
• Employees in teams where two or more colleagues resigned in the last 90 days (attrition is contagious)
• Employees approaching a common exit tenure (for example month 11, if your first-year cliff is at month 12)

None of these is a prediction on its own. Together, they give a small HR team a short, prioritised list of stay conversations to have each month.

Step 5: Establish a review rhythm

• Monthly: HR reviews the dashboard, updates regretted tags, works the early-warning list and logs actions.
• Quarterly: HR presents the segment view and cost estimate to leadership; outlier managers get a structured conversation.
• Annually: the full attrition analysis feeds compensation planning, org design and the hiring plan.

One caution: dashboards contain sensitive data. Restrict manager-level views to HR and the relevant leadership chain, and use risk flags only to prompt supportive conversations, never to make adverse decisions about an individual.

Setting Attrition Targets and Benchmarks

The most common question HR gets is "what is a good attrition rate?" The honest answer is: it depends, and anyone who gives you a single number without context is guessing.

Why benchmarks vary so much

Attrition benchmarks vary enormously by industry (IT services, BPO, retail, manufacturing and start-ups all have very different norms), by role (field sales and support churn faster than finance and engineering), by city, by company size and by the economic cycle. They also depend on how each survey defines attrition, which, as we have seen, varies. If you want an external reference, look at industry-body reports for your sector, talk to peer companies in your city, treat any figure as a rough range, and verify the definitions behind it before comparing.

Set targets against your own history

For most SMBs, the most useful benchmark is your own trailing performance. A practical approach:

1. Establish your TTM baseline for total, voluntary, regretted and early attrition.
2. Set a target for regretted attrition first. This is the number that matters most and the one most within your control.
3. Set a target for 90-day early attrition, because it is the fastest to improve through better hiring and onboarding.
4. Set segment targets for your two or three worst segments rather than a company-wide target that averages away the problem.
5. Leave total attrition as a monitored metric, so that managers are not incentivised to hold on to non-regretted leavers or avoid necessary involuntary exits.

Not all attrition is bad. Some non-regretted turnover brings in fresh skills, opens growth paths and corrects hiring mistakes. The goal is not zero attrition; it is to drive regretted and early attrition down and to plan for the rest rather than react to it.

A 90-Day Action Plan to Reduce Employee Turnover

Analysis without action is a report nobody reads. Here is a 90-day plan that a small HR team can execute alongside their day job, structured so that each month builds on the last.

Days 1–30: Measure and diagnose

By day 30 you should be able to say, with evidence, "our problem is concentrated in these teams, at this tenure, for these reasons, and it is costing us roughly this much".

Days 31–60: Quick wins and structural fixes

Days 61–90: Embed and communicate

What good looks like at day 90

You will not have cut attrition in half in 90 days; notice periods alone mean exits already in motion will still complete. What you should have is a reliable measurement system, a clear picture of where and why turnover happens, a cost model leadership accepts, a live early-warning list, and the first structural fixes in place. The trend line typically starts to move in the second and third quarter after this work begins.

One pitfall to avoid: do not let the dashboard become a report. The value is in the monthly action list, not the chart. That is what attrition analysis looks like when it is working: fewer surprises, better conversations, and a founder who asks "what does the dashboard say?" instead of "why is everyone leaving?"

Frequently Asked Questions

What is the difference between attrition rate and employee turnover rate?

In everyday Indian HR usage the two terms are interchangeable, and both usually mean the percentage of average headcount that separated from the company in a period. Some practitioners distinguish them: turnover includes all separations, while attrition refers to exits that are not backfilled, as when a company is deliberately shrinking. For SMB attrition analysis, the practical advice is to pick one term, define it in writing, and separately track voluntary, involuntary, regretted and early exits, since those splits matter far more than the label.

Should I count resignations or last working days in attrition analysis?

Both are useful, for different purposes. Resignation date is the better early-warning measure because it tells you when people decided to leave, which matters when you are studying appraisal-cycle spikes. Last working day is the better capacity and cost measure because that is when the seat actually becomes empty. Many SMBs report headline attrition on last working day and track a separate "resignations received" count for early warning. With 60- or 90-day notice periods common in India, the two views can differ significantly within a quarter.

How should I handle contract employees and interns in the attrition rate?

Track them separately from permanent staff. Contract-end separations are planned events and belong in their own category rather than in voluntary attrition, otherwise they inflate the rate and hide the real signal. Contract staff who leave before their term ends are a genuine retention signal and deserve their own rate. For interns, measure conversion to full-time and post-conversion retention instead. If you must report a single blended rate, footnote how contract and intern exits were treated so the number stays comparable over time.

What is a good attrition rate for an Indian SMB?

There is no universal answer. Benchmarks vary widely by industry, role type, city, company size and the state of the job market, and different surveys define attrition differently. Rather than chasing an external number, establish your own baseline, then set targets for regretted attrition and 90-day early attrition, which are the metrics most within your control. If you want an external reference, seek out sector-specific industry reports and peer conversations in your city, verify how those figures were calculated, and treat them as rough ranges rather than as targets.

How can F&F settlement data help with attrition analysis?

Full and final settlement records carry signals that exit interviews miss. Notice buyouts indicate leavers with competing offers and tight joining dates, which is a market-pay signal. Large leave encashment at exit points to people who could not take leave, which is a workload signal. Frequent disputes suggest documentation or manager sign-off problems. Recoveries of bonds or joining bonuses confirm an early-attrition problem that the bond is not solving. Slow settlement times damage your employer brand. Pull these fields from payroll into your dashboard alongside the HRMS exit data.

How do I reduce attrition after the appraisal cycle?

Start by analysing last year's spike: were leavers low-rated (a performance-management issue) or high-rated with below-market revisions (a compensation issue)? Then prepare before letters go out: calibrate ratings across managers, give managers clear talking points on how revisions were decided, and be transparent about the logic. Identify high performers at risk before the cycle and have stay conversations early. Fix pay compression where new hires out-earn tenured staff. Finally, do not rely on counter-offers; they treat the symptom and signal to everyone else that resigning is the way to get a raise.

Conclusion

Attrition analysis is not a once-a-year report. It is a working system: a consistent attrition rate formula, a set of segments that show you where turnover actually lives, a diagnostic process that triangulates exit data with engagement scores, manager span, pay competitiveness, workload signals and F&F records, a cost model that gets leadership's attention, a dashboard that turns all of this into a monthly action list, and a 90-day plan that starts moving the numbers. For Indian SMBs, layering in the appraisal-cycle spike, campus-cohort cliffs, Tier-2 dynamics and contract-versus-permanent patterns makes the analysis far more accurate and far more actionable. Remember that benchmarks vary and statutory rules change, so verify both before you rely on them. If you would rather not build all of this from spreadsheets, CozyHR's HRMS gives you attrition dashboards, regretted-exit tagging, F&F data and early-warning views out of the box. Start a free trial and see your first segmented attrition view within a week.