How we map where a brand’s Amazon pages can sell, market by market

A customer in one of your markets opens your product page on Amazon and finds nothing to add to the basket. Your own list still counts that market as covered, because the page loads. That gap decides where growth stalls and where it can start.

Many brands that sell through partners plan their next market from a page count. We build a different view across 12 Amazon marketplaces. Each product in each market gets one of four page states: offer active, no offer, not verified or no page. The map shows where pages can already sell and which step to try first.

What a brand can work on

On an open marketplace, any seller with genuine stock can offer your product on your page. Under EU trade mark law, goods first sold in the European Economic Area with the brand’s consent can generally be resold by anyone. Each reseller also sets its own prices, and EU competition law protects that freedom.

A page map respects both. It describes pages, not sellers, and it leaves prices to the people who set them.

What the brand can work on is its own side. It can check whether each page can sell in each market and whether the page itself is right. A page without an offer sends demand to other brands. From the outside, the brand sees only growth that stalls without an obvious cause.

So the working question is a practical one. Where can a customer buy your products today, market by market, and where does a page stand empty?

Why we start with brand-controlled pages

For brands with a partner network, we usually start with brand-controlled pages. Partners sell, while the brand looks after the pages and their content.

An open marketplace would bring coverage quickly and leave the content to whoever lists first. Selling direct from the brand’s own seller account keeps the content in hand but creates two new problems. It ties up capital, and it turns the brand into a competitor of its own partners.

Content carries more weight in Europe since new EU product safety rules took effect on 13 December 2024. Online offers must show the manufacturer, details that identify the product, and any warnings or safety information. Warnings belong in a language buyers in that market understand.

A brand that looks after its pages gets those details right once, and every partner who sells on the page benefits in every market. Good titles, images and translations work the same way.

The map makes that choice concrete, since nobody can look after pages they have not mapped.

How we work

We work from public marketplace information, and the map needs no access to your seller accounts. Each cell is read the way a buyer in that market would see the page. Availability can depend on the delivery address. Each cell is therefore read with an address in its own market.

Three questions settle a cell. Does the page load, is there a way to buy, and does a featured offer show? The answers and the date of the reading go into the cell. An old reading is therefore visibly old.

Variation families need care. Where a model sits in a family of sizes or colors, the state belongs to that exact variant, not to the family page. Where titles are ambiguous, the product photos decide which product a page shows.

The map’s output is a workbook with one row per product and one column per market. Each figure in it is regenerated from the underlying data rather than typed by hand. A correction therefore reaches every table at once. An internal consistency check runs before each release and catches mismatches between data and document.

Four states, not two

A presence count asks one question per market, and we ask two: does a product page load, and can a customer buy from it? Together they give four states. A product can have no page at all, or a page whose offer we have not verified yet. It can also have a page with no offer, or a page with an active offer. Each state needs a different next step and a different owner.

We keep the not-verified state deliberately. A loading page proves only that the product exists in that catalogue, not that anyone sells it. Counting it as coverage would flatter the map.

A cell stays not verified for a stated reason. The page loaded, but its offer has not been read yet. The link led to a different variant than the one in the range. Or the only offers are used goods, which a buyer of new products cannot use. The reason sits in the cell. Research then knows exactly what to check next.

An active offer is not always a featured offer either. Amazon can show offers on a page without a featured offer when none of them is eligible to be featured. The buyer then sees a button to view the options instead of an add-to-basket button.

That matters for advertising, too. Amazon’s own advertising guide says an ad will not display when the product is not presenting the featured offer. The map records whether a featured offer shows, so those cells stay apart from pages that sell well. It does not count them as pages without an offer.

A page is not an offer

The matrix separates two gaps that a page count merges. A product can be missing from a market, or its page can exist while nobody sells on it. Where pages exist but offers do not, the plan changes.

For those products the task is not to create pages but to bring an existing page together with a seller who has stock. New pages cost catalogue work and start without reviews. An existing page without an offer mostly needs supply. The first is a content job, and the second is a distribution job.

Supply gaps have a few usual causes. A partner may sell in one market and not ship to the next. A partner may hold stock but not offer it in that marketplace. Or no partner carries the model at all. Each cause needs a different conversation, and the brand knows its own partners best.

The map also shows the opposite case. Some pages have offers but weak content, such as a missing translation or an outdated image. There the job goes to the catalogue team, not to distribution.

The pipeline

Every pass through the data follows the same five steps.

1. Gather: we assemble public marketplace information on products and their page states in each Amazon marketplace.
2. Classify: we give each product in each market one of the four page states, with the date of the reading.
3. Verify: we compare product photos and details where titles are ambiguous or a variant is unclear.
4. Assign: we attach a next step and an owner to each finding. Every cell then has someone who acts on it.
5. Rebuild: we regenerate each output from the data and run the consistency check before release.

The pipeline turns on its judgement calls. A state we cannot verify is marked not verified rather than guessed. A product the matcher calls missing is checked again before anyone plans a launch for it.

The order of the steps matters. Classifying before verifying lets the easy cells settle fast. The effort then goes to the few cells where titles, variants or photos disagree.

One product, market by market

Spot checks show why the states matter. Follow one flagship model through each European marketplace, and the picture can be mixed. It may be buyable in some markets and show no featured offer in others. In one market it may be missing from the catalogue altogether.

A catalogue export would not show this. It lists pages without showing whether anyone sells on them. So we build a board for a product line, with one state per model and market. Where we cannot verify a state, the cell says not verified. We do not guess.

The board also keeps growing. Each new marketplace adds a column where a page can exist without an offer, for example the store Amazon opened for Ireland in March 2025.

Translated titles fool the matcher

We also ask the reverse question: which of the brand’s products have no page anywhere on Amazon? A matcher compares the product catalogue with the page titles in the data and flags a list of products as absent.

That first list overstates the gap. The matcher compares words, and a French or Italian title shares few words with an English product name. Some “missing” products are therefore there under translated titles, and a second check of the pages finds them before the list goes out.

Translation is not the only trap. A model name can use a Roman numeral in one market and a digit in the next. Accessories name the product they fit. A case or a spare part can then look like the product itself. Pack sizes can differ between markets. The second check settles each case by photos and product details, not by words.

We record those exceptions in a file that each rebuild respects. What remains are products that are truly unlisted, and those are real whitespace for a launch.

What the board changes in the weekly work

Once the board exists, it sets the agenda of the regular channel meeting. The team starts with the cells that changed since the previous board, and each changed cell already names its owner.

A page that lost its offer goes to distribution that week. A page that gained an offer gets its date recorded, so its first orders can be read later. Cells that stayed not verified go back to research with their reason.

The board can also be read by market or by product. By market, it shows which market is closest to selling the whole range, and so which to work on next. By product, it shows which models sell everywhere and which stand empty in most markets. The first view feeds the market plan, and the second feeds the range plan.

The economics of brand-controlled pages

The economics fit in one line. Marketplace sales are pages × offers × featured offer × conversion × price, and each driver can lose sales in its own way. Price is set by each seller, so the brand works on the other four.

When you plan the next step, activating an existing page is often the cheapest place to start. Its reviews carry over, while its sales rank has faded and comes back only with sales. Each activation is best treated as a test and measured. Sales are not profit either, and the goal is sales that the whole network can live with.

Guardrails

A map invites action, and acting on a wrong reading would send a team to the wrong market. A launch planned for a product that already has a page wastes catalogue work. The rules below hold for every map.

• Research only. The map changes nothing on Amazon.
• Pages, not people. Each cell describes a product page in one market.
• Dated readings. Every state carries the date of its reading. Nobody mistakes an old reading for today’s.
• Photos over words. A product counts as missing only after the photo and detail check.
• Rebuild, never edit. Each table is regenerated from data, and the consistency check must pass before release.
• Unknown stays unknown. A state we cannot verify is shown as not verified, never as absent.

Who decides

The system prepares the evidence, and people decide. Where pages already exist, activation comes before new pages. Distribution decides which partner to ask about a page without an offer. The catalogue team decides on content changes. Whether a new market opens is the brand’s decision, and questions that only the brand can answer go to the brand.

What a map unlocks

The map opens the work rather than finishing it, and each step builds on the one before.

A map earns its keep when it changes a decision. A page without an offer needs a seller and supply, not new content. A supply gap may close through a partner you already work with. A missing page is a brand decision about opening a market, with its own cost and timing.

To read whether an activation worked, record availability and the page state on the day the page is activated. Then watch orders alongside any stockout or change in the page state. If the product becomes unavailable, restore the buying path before you judge the result.

The next market starts faster, too. The board, the matcher exceptions and the checks already exist. A new marketplace adds a column rather than a new project.

Principles for another brand

1. Count offers, not pages. A page that loads is not a sale.
2. Read each market on its own, since a page can sell in one and stand empty in the next.
3. Activate before you create, because existing pages already carry reviews.
4. Keep control of the pages as a brand and let partners sell. Good content then works for every partner at once.
5. Treat an unverified state as unverified, and date every reading. A confident map built on guesses sends your team to the wrong market.

A page map is where a brand-controlled channel starts, and the next market begins from the same board. A simple test shows whether your brand needs one. Open your best-selling model in three European marketplaces. If the page loads in all three but one has nothing to add to the basket, you have found the first cell of your map.

Credits

Research, data engineering and design come from the Capcelerate team. Photo comparisons settle the ambiguous titles, and every release passes the consistency check before a brand sees it.