Setting up an online shop now takes a few days. Finding buyers takes months. The whole difficulty sits in that gap, and it is where most projects stall: the site is ready, the catalogue is live, and the traffic never arrives. The rule I keep coming back to is the simplest one to remember: your revenue is a direct function of your qualified traffic. Without visitors who actually want what you sell, no amount of product-page polish will rescue the launch.
Before you open, and ideally before you pick a CMS or commission a logo, three pieces of research are worth doing: the real demand in your market, the words your buyers actually type, and the level your competitors have already reached. Together they represent a few days of work, and they prevent mistakes that are paid for in years. What follows is how I run each of them, and the traps I see most often.
A mature market, not a saturated one
Some framing first. The clearest published figures I work with are French — treat them as French data, not as a description of your own market. In its annual review, published by Fevad on 11 February 2026, online sales in France reached €196.4bn in 2025, up 7% year on year. The detail matters more than the total. The number of transactions grew faster than revenue (3.2bn, +10%) while the average basket fell to €62 (−3%). French shoppers, in other words, are buying more often and spending less each time. They come from a panel of payment platforms and leading sites, not from an exhaustive census.
The second piece of framing is less comfortable, and again French. The 2026 edition of Fevad's key e-commerce figures, reported in July 2026 by Ecommerce Nation, counts 158,200 active merchant sites in France and indicates that sites above €10m in annual turnover represent a little over 1% of players but concentrate more than 78% of the market. That market is not saturated; it is highly concentrated. You are not entering an empty space but one where a few hundred players capture most of the attention. Whether the same ratios hold in your own market or not, the lesson travels: the three checks below exist to find the ground large players do not cover.
Research 1: how much demand actually exists
The first question is not "is my product good?" but "how many people look for it each month, and with what intent?". A market can be fascinating and tiny. Another can be dull and deep. Only data settles it.
I always start with the Google Ads Keyword Planner, which provides estimated monthly search volumes — estimates, often shown as ranges, not measurements. I cross-check them against Google Trends to see seasonality and the direction of travel over three to five years: a slowly declining market shows up very quickly on a long curve. Then I read autocomplete suggestions and related questions, which reveal the vocabulary buyers use rather than yours.
One point I repeat on every project: separate informational queries from buying queries. "How to choose boxing gloves" and "12oz boxing gloves" are not worth the same thing. The first feeds a blog and an audience; the second feeds a till. A catalogue is built on the second kind, an editorial content plan on the first. Both are useful, but not at the same moment and not with the same budget.
Research 2: the keywords that shape your catalogue
The second piece of research is choosing a head term and deriving the site architecture from it. On a sports equipment project, the priority expression was "boxing equipment": broad enough to carry the whole shop, precise enough to match the catalogue. Everything else — categories, subcategories, filters — follows from the words buyers genuinely use, not from the supplier's internal nomenclature.
That is the most expensive mistake I see: categories get named the way the wholesaler names them, and nobody searches for those labels. A tree built on demand can be fixed in a day before launch. Afterwards it means redirects, URL changes and positions to win back.
Do not confuse volume with accessibility, either. A term with 20,000 monthly searches held by five national chains will earn you less in year one than a family of longer queries at 200 searches each that nobody has written a decent page for. Deciding which of the two you can realistically attack is what a properly scoped audit is for.
Research 3: measuring the gap with your competitors
Then comes the part founders avoid most willingly: looking coldly at who already occupies the first page of Google for your target queries, and how far away you are.
On the sports equipment project mentioned above, the reading I took in 2020 with Semrush gave a very telling scale: more than 26,000 referring domains for a national retail chain, roughly 570 for a specialist shop that had been trading for several years, and about a hundred for the newcomer. Those numbers are a personal, dated reading from one tool: every link analysis platform maintains its own index and reports different totals. The order of magnitude is the useful information. Competing with the national chain made no sense. Going from a hundred to five or six hundred referring domains through two years of serious link acquisition work was, by contrast, achievable, and was enough to close in on the direct competitor.
The second indicator is catalogue depth, and here a correction to my earlier advice is needed. I used to recommend the site: operator to count indexed pages:
site:example-competitor.com
site:example-competitor.com/collections/
site:example-competitor.com/product/Google now documents explicitly that this operator does not necessarily return every indexed URL and that the list is not exhaustive on large sites. Use it as a rough probe on a competitor, never as a measurement. For your own site, the only reliable source is the page indexing report in Search Console, completed by the performance report, which shows the queries you already appear on. If pages are missing from that report, the fix is usually upstream: getting pages indexed deserves its own pass.
| Research | Indicators observed | Decision it supports |
|---|---|---|
| Market | Estimated monthly volumes, three to five year trend, seasonality | Is there enough demand to live off this catalogue? |
| Keywords | Head term, buyer vocabulary, buying versus informational intent | How to name and organise categories and product pages |
| Competition | Referring domains, catalogue depth, quality of the pages ranking first | Which queries to enter now, which ones to postpone |
Turning the three checks into a work plan
None of this is worth anything if it ends up in a document nobody opens again. I convert the findings into three written decisions: the list of categories to create and their exact labels, the volume of content to produce over the first twelve months, and the number of external links to earn to match the direct competitor — not the leader.
One conclusion still holds: when the catalogue gap is wide, broadening the range is often more effective than rewriting the same hundred pages again. That only works if every added page has its own purpose. A thin or duplicated page contributes nothing and may never be indexed.
The last point separates projects that last from projects that fade: measure. The Afnic study "Réussir avec le web", run among nearly 2,500 French micro-businesses and SMEs between September 2024 and August 2025 and presented by France Num on 18 March 2026, reports that 61% of the French businesses surveyed do not really monitor the performance of their online acquisition and retention activity. The same self-reported study shows that 24% of them sell through their own website. I have no equivalent UK figure and will not invent one, but the habit the study describes is common enough to be worth guarding against.
A simple dashboard is enough: Search Console, your analytics, revenue by category, consolidated in Looker Studio if you want it in one view. Six months in, you will know which categories are lifting, which are flat, and whether the three checks proved out or exposed a wrong assumption — far cheaper to learn in month six than in year three.
Common questions
How long should these checks take before opening a shop?
Allow three to five working days for a mid-sized catalogue: a day to frame demand, one or two to build the keyword list and the site tree, one or two to analyse the competitors already ranking. That is negligible next to rebuilding a category tree after launch, which means redirects and lost positions. Refresh the exercise once a year: buyer vocabulary shifts.
Are the search volumes shown by keyword tools reliable?
They are estimates, not measurements. The Google Ads Keyword Planner often shows ranges rather than exact values, and third-party tools such as Semrush or Ahrefs report different numbers from one another. Use them to compare terms against each other and to establish orders of magnitude, not to build a revenue forecast.
Does the site: operator tell me how many pages a site has indexed?
Not reliably. Google's documentation states that the site: operator does not necessarily return every indexed URL and that the list is not exhaustive on large sites. For your own site, the page indexing report in Search Console is the only source worth acting on; for a competitor, the operator gives an order of magnitude at best.
Should I target high-volume terms or niche queries?
Both, but not at the same time. A heavily searched term occupied by national chains takes years of work and a substantial link budget. A family of longer queries, individually smaller but poorly covered, produces sales far sooner and funds the next stage. Build on the second group, then go after the first.
What should I look at first on a competitor's shop?
Three things: the number of referring domains pointing at it, the depth of its catalogue, and the actual quality of the pages ranking first for your target queries. The first measures accumulated popularity, the second the editorial effort needed to catch up, the third the space left for you. Benchmark against the direct competitor, not the sector leader.
Can you run an online store without doing any SEO?
Yes, if you have another acquisition source: paid advertising, marketplaces, social platforms, an existing customer base. Those channels stop when the budget stops, whereas organic search keeps producing visits. Most profitable shops combine at least two and measure what each contributes.