Skip to content
← Insights
Nov 14, 20264 min read

Online retail can grow while your store gets squeezed

National sales, search interest, and customer payback measure different things. Follow comparable cohorts before blaming the whole online market.

A pastel parcel on a desk beside a magnifying glass, a small sales chart and grouped customer-order cards.

A growing online retail market can contain a shrinking store. That is why I would hesitate to explain a merchant's difficult year with a chart of national e-commerce sales, or explain the whole market with falling searches for one marketplace.

Those charts answer different questions. One estimates money spent across a defined retail population. Another measures relative search interest. Neither tells you what happened to a particular store's returning customers, product mix, or cost of acquiring its next order.

A post-pandemic slowdown needs a defined measure too. A store might mean fewer orders, slower revenue growth, or a longer wait to recover acquisition spending. Those changes call for different evidence.

Read the release notice before the number

The U.S. Census Bureau's quarterly e-commerce page illustrates a less glamorous problem with current-looking statistics. Its August 18, 2026 release estimated second-quarter seasonally adjusted online retail sales at $340.2 billion. The estimate was not adjusted for price changes.

As checked on November 14, the page also carries a notice that those estimates ceased to be the most up to date on September 28. It says revised estimates will appear in the third-quarter release scheduled for November 19. I would cite the August number as a dated release estimate, not present it as the final current total.

Even without a revision, a rise in dollar sales is not the same as a rise in units purchased. Prices, basket size, product mix, and the number of transactions can move differently. Seasonal adjustment addresses recurring calendar patterns; it does not remove inflation.

For a merchant deciding what to change this month, national growth is context. The store still needs its own comparable period, category, and customer records. Selling home-office furniture and selling repeat-purchase pet food expose a business to different buying cycles.

Search interest cannot fill the revenue column

Google's explanation of Trends data says the service uses a sample of searches, normalises them for the selected time and geography, and scales relative interest from zero to 100. A value of 50 is not a count of customers or half a company's sales.

A marketplace can receive fewer branded searches while existing customers open its app directly. The opposite is possible too: a controversy can produce searches from people who have no intention of buying. Those are possible explanations to investigate, not findings about any named company.

Changing the search window can also change the scale against which a point is displayed. Save the query, region, date range, and search type when sharing a chart. Without those settings, two attractive lines may not describe a comparable measurement.

I would use such a chart to ask where attention is moving. I would not use it to rank marketplace revenue, calculate an individual store's market share, or conclude that online shopping is disappearing.

Follow a customer cohort past the first order

A store's acquisition report needs a different unit of analysis. Group customers by when and how they first bought. Follow their orders, refunds, and repeat purchases for the same elapsed period. A cohort acquired six months ago has had more opportunity to return than one acquired last week.

Shopify's current acquisition guide distinguishes acquisition cost, contribution margin, and payback period. These are useful accounting concepts, although a platform's general guide cannot supply the right target for every merchant.

Consider an illustrative campaign costing $2,000 that brings 100 new customers. Acquisition cost is $20 per customer under that attribution rule. If their first orders leave $1,500 after the variable costs included in the calculation, the cohort has not recovered that campaign spend from first orders. It still needs $500, before any overhead excluded from the example.

An assumed future repeat purchase is not money already received. Follow the actual cohort long enough to observe those purchases and later returns. Keep the attribution rule stable, since counting a returning customer as newly acquired changes the denominator without creating another customer.

Choose the evidence for the decision

The ONS internet-sales dataset offers another reminder about scope: it reports Great Britain and separates store types. It cannot be substituted for a U.S. market total because both sources use the word retail.

Before changing a store's acquisition budget, I would put three records together: comparable store sales, customer cohorts observed for equal periods, and the costs those cohorts have actually recovered. National releases can help explain the environment. Search data can suggest a channel to investigate. The decision to buy another thousand visits should rest on what happens after this store's customers arrive.

E-commerceAnalyticsBusiness