Notes for owners · CRM, automation and RevOps
Measuring an online store: the five numbers an owner should read each week
Orders, conversion, cost per order, return share and repeat share, read side by side against your own baseline, tell an online store owner far more than revenue does. What each means and how to read them together.
The GullySales team · Updated 17 Oct 2026 · 6 min read
On this page
Sales were up last week, yet there is less money in the account. An online store owner explains that by reading five numbers together, each week, against the store's own last few weeks: orders, conversion, cost per order, return share and repeat share. Revenue on its own counts cash-on-delivery parcels that will come back, discounts that ate the margin and ad spend that doubled to get there. The five, read side by side, show which of those happened. None of them needs software beyond the store admin, the ad accounts and the courier panel.
Why revenue on its own misleads an online store
A shop counts money in the drawer. An online store counts orders placed, and a placed order is a promise, not a sale. A COD parcel can be refused at the door and travel back at your cost. A size return arrives ten days after the week it was counted in. An ad campaign can lift revenue while each extra order costs more than it earns.
So the figure on the store dashboard is the most cheerful number you have. Read it, then read the five that explain it.
The five numbers, and where each one comes from
| Number | How to work it out | Where it comes from | What it gets confused with |
|---|---|---|---|
| Orders | Orders placed, counted separately for the website and each marketplace | Store admin, seller panels | Revenue |
| Conversion | Website orders divided by website sessions | Your analytics | Add-to-cart rate |
| Cost per order | Total ad spend divided by store orders | Ad accounts plus store admin | The cost per purchase an ad platform reports |
| Return share | Returns plus refused or undelivered COD parcels, divided by orders shipped | Courier or shipping panel | Customer returns alone |
| Repeat share | Orders from customers who had ordered before, divided by all orders | Store customer list, matched on phone number | Total customers |
Two details make the sheet honest. Split return share into customer returns and return-to-origin parcels, because they have different causes. And match repeat customers on phone number, not email, because one buyer may use three email addresses and one mobile.
Read them in pairs, never one at a time
A single number moving tells you almost nothing. Two moving together point at a cause.
- Orders up, conversion down. You bought more visitors, and they were the wrong ones. Look at which ad, creator or post sent the new traffic.
- Orders up, cost per order up faster. Each extra order is costing more. The last slice of ad spend is buying orders you would not pay that much for if you saw the bill one at a time.
- Orders flat, return share up. Something changed between the page and the parcel: a new courier, a new region taking COD, or a product page that promises a colour or fit the parcel does not deliver.
- Orders up, repeat share down. New buyers are arriving through ads and the old ones are not coming back. The work sits after the first delivery, in how ecommerce businesses can encourage repeat purchases.
One more line is worth working out every time: ad spend divided by orders that were kept, not orders placed. If a fifth of orders come back, your real cost per order is a quarter higher than the sheet says.
Use your own baseline, and know when a week is too small
Fill the sheet backwards for the last eight normal weeks before you judge anything. That is the baseline. A blogger's average conversion rate is built on stores with other prices, other categories and other shares of COD, and it will make you either complacent or panicky.
Mark the abnormal weeks in a notes column: a sale, a festival, a courier strike, a creator's post going round. Compare a Diwali week with last year's Diwali week, not with the week before it.
And be honest about volume. If one order more or less moves your conversion figure visibly, a single week is too small to read. Add up four weeks and read the total instead, until orders grow.
Imagine an umbrella brand in Kozhikode
Imagine a Kozhikode brand selling folding umbrellas and raincoats on its own site and one marketplace. The first week of heavy rain arrives in Kerala, revenue jumps, and the owner doubles the ad budget on the strength of it. The new ads reach states further north, and COD is opened there too. The brand and every figure below, rupees included, are invented.
| Week | Orders | Conversion | Cost per order | Return share | Repeat share | Note |
|---|---|---|---|---|---|---|
| First heavy-rain week | 180 | 2.1% | ₹210 | 9% | 31% | Ads at normal budget |
| Two weeks later | 260 | 1.6% | ₹290 | 22% | 19% | Budget doubled, COD opened to new regions |
Revenue was higher in the second row. The sheet said otherwise. Conversion fell because the wider ads reached people in places where the rain had not yet arrived: browsing, not buying. Return share more than doubled, almost all of it refused COD parcels from the new regions. Repeat share fell because the regulars were now a smaller part of a bigger, worse crowd.
The owner's first move was a confirmation message on every COD order before dispatch, with unconfirmed ones held; our note on confirming a COD order on WhatsApp has the wording. Ads went back to the regions where parcels were accepted, and moved north only as the rain did. Past buyers heard about the new season's stock before any new ad ran. It was a partial fix: the refused parcels had already cost about ₹17,000 in shipping both ways, and return share only fell to 13% over the next fortnight.
What these five numbers will not tell you
They do not tell you profit. For that you need product cost, shipping, payment charges and returns per order, which is a separate sheet. They do not tell you why a buyer left the cart, which needs you to read the product page on a phone as a stranger would. And they are only as true as the tracking behind them: if the purchase event fires twice, conversion doubles and nothing real changed.
Open a sheet, put the five columns and a notes column in it, and fill the last eight weeks from the store admin, the ad accounts and the courier panel. Measurement sits under every stage of selling online, and website analytics and conversion tracking is the service for the tracking itself. If the numbers disagree with each other, tell us what you need.
