Every retailer eventually asks the same question: is our conversion rate good? It is a reasonable question with an unhelpful honest answer — it depends on what you counted, how you counted it, and which stores you are standing next to. The number itself is easy arithmetic. Making it comparable is the real work.
Here is the formula, the definitions that decide whether your figure means anything, and how to build a benchmark you can actually defend instead of borrowing one from a blog post.
The basic formula
Conversion rate = transactions ÷ people who entered × 100
In a store where 1,200 people entered and 180 purchases were recorded, that is 180 ÷ 1,200 = 0.15, or 15%. The numerator comes from the POS. The denominator comes from a traffic counter or a manual count. Simple — until you look closely at both sides.
Define three things before you calculate anything
Most disagreements about conversion rates are not disagreements about arithmetic. They are disagreements about definitions that were never written down.
1. What counts as an entry
Does a person who walks in and immediately leaves count? Yes, in a standard definition — they entered. Does a group of four walking through one door count as four entries or one? As four, because four people came in and one purchase from the group is still one purchase against four opportunities. Does a courier dropping a parcel count? Practically, most systems cannot separate them, which is why the honest approach is to note the distortion rather than pretend it does not exist. Does a staff member returning from lunch count? They should not, which is why staff exclusion is a configuration decision worth making deliberately.
2. What counts as a transaction
This is the side retailers usually overlook. Decide and document:
- Returns and exchanges. A refund is not a purchase. If your POS counts the return as a transaction, an efficient store that processes a lot of returns will look better than it is.
- Multi-tender and split payments. One customer paying half by card and half in cash may appear as two transactions.
- Click-and-collect and online orders picked up in store. If these land in the same POS feed as in-store sales, your conversion rate is measuring two different journeys as one.
- Wholesale or B2B sales at the counter. A single bulk invoice can distort a day.
Pick a rule for each and apply it consistently. A store with a slightly imperfect but consistently applied definition is far more useful than one with a perfect definition applied only when someone remembers.
3. The attribution window
Traffic and transactions must be matched over the same period. Daily matching fails in low-volume stores where a shop sells ten items a day and the timing of a single large transaction swings the figure by several points. For stores below roughly a few dozen transactions a day, calculate weekly and compare week to week; use hourly data only to understand peak staffing, not to judge performance.
Why published benchmarks disagree so much
Search for a "good" retail conversion rate and you will find numbers that differ by a wide margin, including between sources that both sound authoritative. That is not because one is lying. It is because they are measuring different populations with different definitions:
- Store format. A grocery store, a jewelry store, a footwear shop, a pharmacy, and a convenience store have structurally different conversion rates. High-frequency, low-consideration purchases and high-consideration purchases behave nothing alike.
- Location type. A street-front store with passing pedestrians who happen to enter has a different entry-to-purchase dynamic than a store inside a mall, where shoppers arrive already in a buying mode and often browse several stores before deciding.
- Time window. Annual figures, weekday figures, and peak-season figures do not belong on the same chart.
- Definition of the denominator. Benchmarks that count entries per device, per manual count, or per device-based estimate are not comparable.
- Channel mix. A store that fulfils large volumes of online orders from its back room has a sales figure that has nothing to do with its front door.
The practical conclusion: a cross-industry average is a curiosity, not a target. Treat any published figure as a sanity check on whether your number is plausible, never as the score you are trying to hit.
How to calculate it properly, step by step
- Fix the entry count. Cover every customer entrance. Confirm that doors left open, side doors, and staff entrances are handled deliberately — an uncovered second door silently inflates conversion.
- Exclude staff reliably. Configure the counting system to remove staff, and spot-check it. This is one of the largest silent errors in retail traffic data, because staff move in and out constantly.
- Extract transactions under your documented rule. Apply the returns, split-payment, and channel decisions above.
- Match the periods. Same store, same hours, same time zone. Watch for POS records that close after midnight and land in the wrong day.
- Compute weekly first, then decompose. Get the weekly number stable before you look at hours. A weekly figure with a stable definition is a management tool; an hourly one computed from eleven transactions is noise.
- Record the method alongside the number. One line — entries source, staff exclusion, transaction rule, period — turns the figure from an argument into a baseline.
Reading the number: four diagnoses
The value of conversion rate is not the number itself but what it separates. Compare it against your own recent baseline and against your own store cohort, and you get four useful situations:
- Traffic down, conversion stable. Fewer people came, and those who came still bought at the usual rate. Look outward — weather, road works, a mall promotion elsewhere, a competitor opening, seasonality. The store is not failing; the opportunity count is lower.
- Traffic stable, conversion down. The same number of people arrived and fewer bought. That points inward: staffing at peak hours, queue length, out-of-stocks, pricing changes, a display reset that broke the entrance path, or a service problem.
- Both down. This is where a single figure would have hidden the problem. Check whether the traffic drop is real before chasing a conversion problem that may not exist. A counter that stopped reporting, or a second entrance left uncovered, produces exactly this pattern.
- Traffic up, conversion down. Often capacity rather than demand: more people, longer queues, more abandonment. Also common after a promotion that brings in browsers. Check queue analytics and service levels before concluding that staff are underperforming.
A single number cannot produce this table. The pair of numbers can.
Setting a benchmark that holds up
Instead of importing an external figure, build one from your own estate:
- Establish a baseline period of eight to twelve weeks with a fixed method. Short windows in retail are dominated by weather, paydays, holidays, and promotions.
- Rank your comparable stores. Same format, similar size, similar catchment. The median of that group is a far better reference than any published average.
- Segment before you compare. Weekday versus weekend, and trading hours versus quiet hours, behave differently enough that a blended rate hides both.
- Set targets as deltas, not absolutes. "Lift conversion by two points in this store over the next two months against its own baseline" is actionable. "Reach a specific industry average" usually is not, because no one in the room can reproduce where that average came from.
- Re-baseline after material changes. A new layout, a new entrance, or a change of counter configuration resets the measurement. Note the date, or you will spend the next quarter comparing two different definitions.
Common ways the figure goes wrong
- An uncovered second door. The most common single cause of a conversion rate that is quietly too high.
- Staff traffic left in the denominator. Staff entering ten times a day across forty staff members adds up fast.
- Non-shoppers from nearby units. Shared corridors, deliveries, and building access can land in the count if the counting zone is not drawn tightly.
- Event days mixed into the baseline. A mall event or a local festival produces a traffic spike that converts poorly. Keep those days flagged, not deleted.
- Denominator from a biased source. Device-based estimates and camera counts are not interchangeable. Mixing them across stores makes the comparison meaningless — the differences in the sensing method are larger than the performance differences you are looking for.
- Attributing online-influenced sales to in-store traffic. Customers who research online and buy in store are a genuine conversion, but a store fulfilling web orders is not.
FAQ
What is a good conversion rate for a retail store? There is no single number that applies across formats and regions, and any source offering one without stating its definitions should be treated cautiously. The useful benchmark is your own store cohort measured the same way. If you need an external reference point for a board discussion, insist on one that documents its store mix, geography, and entry-counting method.
Do I divide by entries or by unique visitors? Use entries for the standard retail conversion rate. Unique visitors, where a system can derive them, are more relevant for loyalty and repeat-visit analysis than for daily conversion, and the two figures should never be presented as the same metric.
Should staff be excluded from the entry count? Yes, for conversion analysis. Staff represent zero purchasing opportunity and known, repeating traffic. Exclude them at the counting configuration level and verify periodically, since a change in uniform or a new shift pattern can silently break the exclusion rule.
How do I calculate conversion without a traffic counter? Run a manual count for a defined set of days — a full trading day is better than a partial one — and pair it with POS transactions from the same hours. This produces a rough baseline and, in most stores, an immediate appetite for automation, because manual counting cannot cover every door every day.
Can I compare conversion rates between my stores? Yes, but only if the counting method, transaction definition, coverage of all entrances, and time window are identical. If they are not, compare each store against its own trend and treat the cross-store ranking as indicative at best.
---
If you are starting from manual counts and want to see what automated measurement looks like across retail, mall, and public-building deployments, the solutions overview shows how counting is deployed in each environment. For the measurement background behind this article, browse the blog archive, which includes our guides on counting methods and accuracy. Have a specific store layout or estate to size up? Contact our team and we will help you define the method before you buy the hardware.