More orders than last year. Has the shop grown?

Imagine the owner of an online gardening shop. He has closed September, checked the orders and messaged his business partner: “A quarter more than last year.”

“Good. How much did we spend on ads?”

“More as well. Let me check.”

Then he remembers: they added garden hoses in spring. Last autumn, they had none at all. And this year they bought extra stock of several watering cans that had sold out quickly the year before.

There really are more orders. But what has grown, and what should guide the next stock purchase, is not yet clear. Let’s work through it with a fictional shop and invented figures.

First, decide which two Septembers you are comparing

The owner takes 1–30 September 2025 and 1–30 September 2026. Two complete months, one currency and one source of order data.

Then he checks order statuses. If last year’s figure includes only completed orders, while this year’s still includes cancelled and unpaid ones, the comparison starts with a mistake. Returns also need to be treated consistently. Recent data may still change; say so alongside the result.

In our example, we count completed orders that had not been fully returned by the reporting cut-off. Product revenue is after discounts and known returns, excluding VAT and delivery charges. The definition is the same in both years; later returns may still change the total.

Fictional figures for the whole shop across two complete Septembers; this is neither an ad campaign report nor a client case study.

Orders rose by 25%: 300 ÷ 240 − 1. Revenue rose by 37.5%: 16,500 ÷ 12,000 − 1. Revenue per order increased by 10%.

Those numbers fit together: 1.25 × 1.10 = 1.375. But they do not yet explain why each order was worth more. Customers may have bought more units, chosen different products or received a smaller discount. Prices may have changed too. That is the next question to ask about the contents of the orders.

Separate new products without losing the overall result

The owner makes a second comparison: products carried in both years, including those with no sales in one of the periods. He separates out the new hoses and also notes models that disappeared. Nothing is removed from total revenue: it is all part of the shop’s result.

This lets him answer two different questions: how sales of the existing assortment changed, and how much of today’s revenue came from new products. Not every hose sale can be called additional growth: some customers may have chosen a hose instead of another product, or bought it alongside one.

Conditions were not the same for the old assortment either. A watering can was in the catalogue in both Septembers, but last year customers could not order it for two weeks. You can compare its sales; you cannot attribute the difference to growing demand without considering availability.

So he adds out-of-stock dates, price changes and significant changes to delivery alongside the models. You do not have to reconstruct a perfect history for every item at once. Start with the products that account for a sizeable part of the difference.

Open a separate report for the ads

Now the owner looks at Google Ads. For this example, suppose September’s spend rose from €1,000 to €1,500, and the purchase value recorded by the ad system rose from €5,000 to €7,500. Measurement settings and the selected campaigns are comparable in this example.

Both metrics increased by 50%. ROAS remained at five: 5,000 ÷ 1,000 and 7,500 ÷ 1,500.

A correct entry is: “Spend and recorded purchase value rose by the same proportion; ROAS held.” Writing “the extra budget caused all the growth” needs separate evidence. Perhaps the budget was raised because orders were already improving. Perhaps a product came back into stock at the same time. Matching percentages do not identify the cause.

That value attributed to ads is a separate metric. You cannot add it to the shop’s revenue: purchases the system attributes to ads may already be included in the shop’s orders.

When comparing, also check whether the reports are complete. Google describes conversion delay: some purchases do not appear in the ad report straight away. Keep that in mind when comparing last year’s final result with a recent one that is still being updated.

Do not blame the season on the strength of one line

September against September is a useful starting point. But the same month does not make the weather, promotion dates, mix of customers or the shop’s offer identical.

Look at the neighbouring months in both years for the same product group as well. If purchases simply shifted from August to September, September alone will suggest a picture that the whole period does not support. The boundaries of this additional comparison must also match across the two years.

Keep ad impressions alongside the figures as a clue to investigate. But a budget change can alter how often ads are shown. More impressions alone therefore do not prove that customers need your products more.

Return to the next stock purchase

Instead of one message saying “we grew by a quarter”, the owner now has a clear breakdown. Orders increased, and so did revenue per order. New products joined the assortment, while some existing ones were available for longer. Ads spent more; their separate contribution has not yet been established.

The next action is to examine the models behind the difference, check their costs and the terms for reordering. Revenue growth does not tell you how much money remains after all expenses or how much of it is available to buy stock.

You can finish your own comparison with three sentences: what definitely changed in the figures; which conditions changed alongside them; and what you will check before the next investment. Last year will then help you make a decision, rather than simply give you an attractive percentage to message your partner.