How agencies run many stores from one place

An agency running fifteen stores has fifteen Google Ads accounts, fifteen Merchant Centers, fifteen feed tools, fifteen email tools and fifteen ways to report. Every client asks the same question: what did we get for the money? This post describes how an agency runs that book on Orkie: one private workspace per client, playbooks that carry from one account to the next, and reporting that reconciles to each store's real orders rather than to what each platform claims.

One workspace per client, walls between them

Each client gets a workspace. It is their environment: their pixel, their connected store, their ad accounts, their catalog, their inbox, their CRM. The agency's staff are users in it, with their own logins and their own roles. The client's team can be users in it too, with whatever access the engagement calls for.

Workspaces do not see each other. A connector authorized in one workspace reaches only that workspace. A report in one workspace reads only that workspace's data. When an assistant is connected through Workspace MCP, it reaches the workspace that issued the connection and nothing else.

That separation is what lets an agency put a client's own marketing lead into the workspace without worrying about what else they can see. It is also what lets the agency hand a workspace back if the engagement ends. The data, the pixel history and the feeds stay with the store.

Same modules, different mix

Every workspace installs from the same catalog of modules. One store might run the pixel, the store connector, Google and Microsoft Ads, Merchant Center, feeds, the inbox and the CRM. A fashion brand might add Meta, TikTok and Instagram. A B2B supplier might add deals and QuickBooks. The module is the same code in every workspace. The configuration is the client's.

For the agency, that means one set of things to learn. A tracking-template repair on Google Ads works the same way for every client. A feed output for Microsoft Merchant Center is built the same way for every client. A smart inbox with a returns skill pack is configured the same way for every client. Staff move between accounts without relearning the tool.

Playbooks that carry across accounts

A playbook is a written way of doing something that produces the same result on every store. Because the workspaces share modules, a playbook written once applies everywhere. Some the operator team uses:

Each of these is executed per workspace with that store's data. None of them requires a custom build. When a playbook improves, it improves for the next store, and the previous stores can be brought up to it.

Reporting that reconciles

The reporting problem for agencies is not a lack of dashboards. It is that every platform's dashboard reports its own results, and the client's finance team reports something else. Meta's number plus Google's number is more than the store's revenue, and the client notices.

Each workspace has one record: the pixel's sessions, joined to verified orders from the store, with ad spend counted once per channel. Reports read from that. So the monthly report says what the store took, what each channel was credited under the model the client agreed to, what was spent, and what the return was. The platform-versus-pixel comparison is on the same page, per campaign, so when Google's number and the pixel's number disagree the report explains the gap rather than hiding it.

Say an agency reports for a client doing $4 million a year. Under last click, Google Shopping gets $1.6 million and Meta gets $900,000. Under first click, Meta's share is larger and TikTok appears. The report shows both, with spend beside them. The client sees why the prospecting budget exists, and the agency does not have to defend a number the platform made up.

Reports can be built once and rerun with the period changed. Comparison cells show the previous period and the change. And because refunds are subtracted on the server, revenue is net of returns rather than gross.

Where the operator team fits

Orkie's model is operators, not consultants. The humans at Orkie run the work inside each workspace: installs, feeds, uploads, inbox setup, campaign builds. An agency can lean on that in two ways.

Some agencies keep strategy and client relationship and hand execution to the operator team. The agency's account leads read the same workspace the operators work in, and the client sees one set of numbers. Other agencies run execution themselves and use the operator team for the things their staff should not have to learn, such as a custom feed rule or a storefront widget, filed as a Projects task and built.

Either way, the work happens in the client's workspace, under the client's data, with the audit trail that comes with it.

Asking across a book of business

Within one workspace, an account lead can ask Claude or ChatGPT: "How did this store do last week against the week before, by channel, with spend?" or "Which products on this feed are seen a lot but rarely clicked?" and get an answer from that store's record. For the routine actions, a repeating task can be approved once and run on schedule: "Every Monday, set campaign 123 to this budget." Every exact action is recorded, and a change to the amount or campaign needs a fresh approval.

The assistant does not cross workspaces. To ask about a second client, connect to that client's workspace. That is a small cost for the guarantee that no client's data answers another client's question.

What an agency gets out of it

Fewer tools per client, one way of working across all of them, a report that reconciles to the store's books, and a team of operators who run the mechanics so agency staff can spend their time on the decisions. The price of that is discipline: pixel first, everything else reads from it.

Running more than one store? Talk to us to see a workspace per client.

---

Want this for your business? Talk to us How the pixel works