Managing SEO across several stores — the agency method

Running SEO for one store is a job; running it for twenty is another one. The problem isn't technical difficulty — the levers are the same — it's the repeated load: twenty catalogs to understand, twenty editorial lines to hold, twenty reports to produce. Without a method, every new client degrades the quality served to the previous ones.

Here's how the agencies that last organise themselves.

The golden rule: standardise the process, not the content

The fatal mistake in multi-store SEO is uniformising content to go faster: the same articles recycled from one client to the next, the same generic descriptions. Google sees it, and duplicate content sinks both stores involved instead of one.

What you standardise is the process: how you find keywords, frame a brief, review, publish, measure. The content itself stays specific to each catalog. The good news: the standardisable part is precisely the time-consuming one.

There's one case where the question really bites: two competing clients in the same niche. The rule is simple and worth writing into your process — never the same target keyword for two stores in the portfolio. If both want it, one takes the main query and the other a longer intent variant. Competing against yourself is the only certain way to lose both.

The four steps of a portfolio under control

1. Map each store once. For every client: the categories that make money, the commercial-intent keywords, the pages that already exist.
2. Set the cadence. A sustainable rhythm, identical for everyone (say 4 articles + 20 product pages per month per client), beats a heroic effort you can't reproduce.
3. Publish without friction. This is where multi-client work breaks: juggling twenty back offices, twenty passwords. Remote publishing over a signed API removes that cost — no client credentials to store.
4. Measure and show. One report per store, comparable over time, turns the engagement into a demonstrable result.

A portfolio dashboard: four columns are enough

With twenty stores the temptation is to track everything. In practice, four indicators per store are enough to decide where to spend your half-day:

• Is the store answering? Is the module connected, is the catalog sync recent? A store silent for three weeks is a client receiving nothing, and they'll find out before you do.
• Is measurement wired up? Without Search Console connected, you're flying blind and the report becomes an opinion.
• How many pieces published in 30 days? That's the contractual commitment, and the only line the client checks.
• The 28-day click trend against the previous 28. Not absolute traffic: the variation. That's what separates a store taking off from a store falling asleep.

Those four columns give you an order of priority, and that order is almost always counter-intuitive: the store that needs you isn't the one whose owner calls most often.

The multiplied back-office trap

Add it up: twenty stores, each with its credentials, its updates, its PrestaShop version (from 1.6 to 9). Logging into each one to publish is half a day lost per week — and a security risk with every password that circulates.

That risk deserves naming, because it's rarely quantified: holding the back-office credentials of twenty merchant stores means holding access to the orders and customer data of twenty businesses. If there's an incident at your end, the responsibility travels back up. A publishing account limited to content, over a signed API, reduces the surface: nobody shares a password, and the technical scope of your access matches your engagement exactly.

The way out: a module installed once at the client, then everything driven from one account. The client keeps their back office; you keep the production. That's exactly what Publium allows, and it's the heart of our agency page.

Getting client approval without creating accounts

On a portfolio, approval is the second bottleneck after review. The classic routes — a shared document per client, email back and forth, screenshots — don't scale: at twenty clients you spend more time chasing than producing.

What works: a preview by link, which the end client opens with no credentials, where they read the content as it will appear and answer in one gesture (approve / request changes, with a comment). You get a dated decision, they have nothing to install or remember. And the day they don't answer, your contractual approval deadline applies without debate — see how to bill and bound the scope.

What changes with AI tooling

Across a portfolio, AI-assisted generation doesn't just speed things up: it makes the rhythm reproducible. Every article starts from the client's real catalog, links their product pages, and you keep only the value-adding work: framing and review.

A word on what the tooling does not change, because the opposite promise circulates a lot: it doesn't replace your judgment on what should be published, nor your knowledge of each client's market, nor the relationship that keeps you hired. It removes the handling. That's already most of what stopped you serving twenty stores properly.

The mistakes that cost the most across a portfolio

• One calendar for everyone. Stores don't share a seasonality. A calendar per client, however rough, stops you publishing a Christmas buying guide in February.
• The generic report. A report showing overall traffic without saying what was done that month proves nothing. Date the publications, show their effect on the pages concerned, compare with the rest of the site — carefully: a correlation isn't proof, and saying so strengthens your credibility.
• Forgetting versions. A real PrestaShop portfolio mixes 1.6, 1.7, 8 and 9. Choosing tooling that only covers recent versions rules out half the market — and the old base is precisely the worst served, so the easiest to win.
• Uniform production. Concentrate volume where measurement moves; keep the rest on upkeep.

In short

A portfolio of stores is run like a chain: identical process, unique content, centralised publishing, comparable measurement. Four indicators per store give you the order of priorities; a preview link unblocks approval; a signed API removes the passwords in circulation. SEO technique doesn't change scale — organisation decides whether you serve twenty clients well or five badly. To talk it through on your portfolio, book a demo.