Building a Unified Operating Model for a Multi-Site Retail Portfolio
Client Profile
A retail real estate operator. Saudi Arabia.
Portfolio of more than a dozen malls and plazas. Three in-house marketing functions: digital marketing, customer experience, and events. Marketing officers stationed at individual properties.
Position at engagement: over a quarter of a million monthly visitors across the portfolio, arriving at disconnected digital properties with no shared operating system behind them.
Client identity, property names, programme dates, and exact figures withheld at client request.
Executive Summary
The operator's marketing teams were not underperforming. That was the problem.
Every function was active. Every individual was working hard. Yet execution kept getting heavier. Social decisions took too long. Event coordination required constant alignment. Customer engagement drifted between properties.
The presenting problem was speed. They believed better tools and tighter coordination would restore it.
The real problem was that coordination had replaced structure. Work moved only when someone pushed it. Nothing was broken, which is precisely why nothing was visible.
The obvious answer was to push authority down to the people standing in the malls. That answer is right, and it is where most programmes fail.
We did not decentralize the decision. We built the capability to hold it, then decentralized the decision.
The Challenge
1. Coordination had become the operating system. Nothing progressed on its own. Every handoff required a person to carry it. Meetings were not where decisions were made. They were where work was manually advanced. That scales with headcount and never with output.
2. Shared responsibility had produced no responsibility. Three functions overlapped across the same tasks. Everyone was accountable for the outcome. Nobody owned a stage. When something stalled, there was no single person whose job it was to notice.
3. The centre was writing about rooms it could not see. Content for each property was decided centrally. The centre could not observe the floor, so messaging described the mall as it had been briefed a week earlier rather than as it was. The people who could see it had no authority to say anything.
4. The audience already existed. The infrastructure did not. The portfolio's flagship properties each drew over a hundred thousand organic visits a month on separate domains with middling technical health and no unified property behind them. This was not a demand problem. Demand was arriving and landing on nothing.
5. The edge had authority-shaped work but no vocabulary. The marketing officers were capable operators who had never been given the language of the discipline. This never appears in a brief. It is the reason delegation programmes quietly fail.
Discovery
We do not begin with platforms or features. We begin by observing how teams actually work.
We mapped the operation as it ran, not as it was documented:
- The real path of a piece of content from idea to publication, including every undocumented handoff
- Where work stopped moving, and who intervened to restart it
- Stated ownership against actual ownership for each function
- The technical and traffic baseline of each digital property, crawled and recorded before any work began
- What the on-site officers knew that the centre did not, and what the centre knew that they did not
The finding cut both ways. The centre was deciding about conditions it could not observe. The edge could observe them and could not describe them.
The Reframe
The conventional fix for slow multi-site marketing is to centralise. Standardise everything, enforce it downward. It makes output uniform and timing wrong.
The fashionable fix is the opposite. Push authority to the edge and call it empowerment.
Both fail, and the second fails worse, because authority without capability is not empowerment. It is abandonment.
Decentralisation is not a permission you grant. It is a capability you build, and only then a permission you grant.
The System
The Centre-Edge Model
Two rules govern the operating model.
- The centre owns how work moves. Structure, standards, ownership, and rhythm are designed once and are not negotiable.
- The edge owns what gets said. The people standing on the property decide the daily output within that structure.
Tight on the system. Loose on the content. Most organizations invert this and end up with the worst of both.
The Mandate Split
Ownership was rewritten so each function held a distinct stage rather than a share of everything.
Digital marketing owns overall social strategy, the content calendar, and the publishing standard across all properties. Customer experience owns monitoring and response to inbound customer contact. Events owns planning, execution, and post-event analysis.
The test applied to every task: exactly one name. Where two functions could claim it, the mandate was wrong and was redrawn until only one could.
Shared accountability sounds collaborative. In practice it is the mechanism by which work stalls without anyone being at fault.
The Capability Stack
This is the part that makes edge execution survive contact.
Authority alone would have failed. So the delegation shipped with four things underneath it.
A single method. One documented sequence from idea generation through creation, review, approval, scheduling, engagement, and post-campaign analysis. Six stages, each with a named owner and a named tool. Not a philosophy. A procedure.
Written SOPs, held where the work lives. Standard operating procedures were written for each team and loaded into the same system that tracks the work, rather than filed somewhere separate. Adherence is visible because the procedure and the task occupy the same surface.
A shot list. Video production was reduced to a checklist an officer can execute on a phone between other duties. Establishing shot, storefront, product, offer, edit. The checklist is what converts intent into consistent output.
A glossary. Seventeen terms, defined plainly, each with a worked example. Engagement rate. Reach. Impressions. Click-through. Bounce rate.
The glossary is the least glamorous item in the programme and the one most responsible for it working. You cannot delegate a decision to someone who lacks the vocabulary to make it. Every firm that skips this step ships authority into a vacuum, watches it fail, re-centralises, and concludes that the edge could not handle it.
The Tooling Layer
Project management, content operations, and social scheduling and analytics were consolidated into a defined stack, with each stage of the method bound to a specific tool. Reporting became an output of the system rather than a manual task.
Ideation runs on a custom AI tool built by Arbaaa and deployed into the client's stack. Officers enter the campaign context: the occasion, the audience, the offer. The tool returns content directions aligned to the brand. It exists because the first stage of the method was the one most likely to stall an officer with a camera and no brief.
Adoption was uneven by design and by reality. The project management layer carried the steepest learning curve and required supplementary sessions and materials beyond the original plan. That was surfaced during the programme rather than after it, and the schedule absorbed it.
Pilot Before Portfolio
Daily publishing authority was first handed to on-site officers at two flagship properties, out of a portfolio of more than a dozen.
A model that survives contact at two sites can be scaled. A model rolled out everywhere at once cannot be tested, only defended.
Evidence
The pilot has concluded, and the model is live. Baselines were established before the programme. Post-implementation figures are held by the client and are not published here.
Baseline, recorded before work began. Technical health and organic traffic were crawled and logged for each flagship property. Both drew six-figure monthly organic sessions on mediocre on-page scores. These are starting positions, not outcomes.
Observed operational outcomes. Teams operate within a defined workflow rather than around one. Cross-functional coordination improved, particularly between digital marketing and customer experience. Dependency on manual follow-up fell. Each person can state what they own and how it connects.
Observed content outcomes. At the pilot properties, publishing moved from centrally scheduled to locally originated. Content reflected live conditions rather than week-old assumptions.
Structural outcome. The model is positioned to extend across the remaining portfolio without redesign.
Reported by client teams and by Arbaaa observation. Cycle time, decision latency, organic traffic, and engagement to be measured against the recorded baselines in the next phase. No performance figure is published until it carries a baseline and a date.
The Shift
Before. Three committed teams and a quarter of a million monthly visitors, held together by constant coordination.
After. One structure that moves work on its own, with output decided by the people who can see what they are deciding about.
Close
Every organization that decentralizes does the same two things. It grants authority and announces trust. Almost none of them build the vocabulary, procedure, and checklist that allow the authority to be used. Then the output disappoints, the centre takes it back, and the lesson recorded is that the edge was not ready. The edge was ready. It was never equipped.
Arbaaa builds the infrastructure behind reputation. Published with client permission. Client identity withheld at client request.