Gulf retail groups run some of the region’s largest loyalty programmes. Majid Al Futtaim’s SHARE programme, Alshaya’s multi-brand loyalty ecosystem, Landmark Group’s Shukran — these are not small card-and-points schemes. They are data-heavy, multi-brand, multi-currency platforms with millions of enrolled members and a contact centre volume that reflects it.
When a loyalty programme at that scale runs into friction — a missing points credit, a redemption that doesn’t go through, an account that shows the wrong balance — the member does not wait. They message. On WhatsApp, through the app, via the call centre, at the store. Often more than once. The contact centre team that handles the overflow has to find the member’s record, locate the transaction, understand the programme rules, and respond — in Arabic or English, in the tone appropriate to the brand, within a service-level window.
That is the contact-centre CX problem for Gulf retail. Not whether AI can personalise outbound marketing — that is a separate question. The question that brings retail CX leaders to Agentforce is: can we handle more of these interactions, faster, without adding headcount proportional to programme growth?
What the retail contact-centre interaction looks like
The interactions landing in a Gulf retail contact centre fall into recognisable categories. Loyalty queries account for a significant share: points not credited, redemption caps, programme tier questions, balance discrepancies after a return. Order management queries follow: delivery status, exchange or return initiation, missing order confirmation. Account management queries round out the volume: profile updates, linked cards, duplicate accounts after a platform migration.
None of these require a human agent to make a judgement call on the customer’s behalf. They require a human agent to retrieve accurate information, apply the right programme rule, and frame a response that is clear, on-brand, and in the right language.
The delay in the system is almost always retrieval and drafting time, not decision time. A well-designed Agentforce integration removes most of that delay: the agent arrives at the interaction with the member’s record pulled, the relevant transaction located, the programme rule applied, and a draft response ready to review before sending.
The three interaction types that fit best
Loyalty balance and transaction queries. When a member contacts the brand about a missing points credit or a balance discrepancy, the contact agent needs to locate the transaction in the CRM or loyalty platform, verify whether the points should have been credited, and explain the outcome clearly. Agentforce, integrated with the loyalty data source and CRM, can retrieve the transaction, apply the programme logic, and generate a draft response that the human agent reviews and sends. This pattern handles the highest-volume category of retail contact-centre interactions with the most consistent outcome.
Return and exchange initiation. When a customer requests a return or exchange through a digital channel, the interaction follows a structured process: verify eligibility, locate the original order, generate a reference or label, confirm the next steps. Agentforce can handle the retrieval and response-drafting steps, presenting the human agent with a pre-populated resolution draft that confirms eligibility, provides the reference, and outlines what happens next. The human agent reviews and confirms before the message sends.
Post-interaction case summarisation. Retail contact-centre interactions frequently involve multiple touchpoints before resolution — a WhatsApp exchange, a call, a follow-up message. At the end of a resolved interaction, writing the case note takes time that experienced agents could spend on the next queue. Agentforce can generate a structured summary of the issue, what was done, and the outcome, which the agent reviews and saves. In high-volume contact centres, the recovered capacity from this step is material.
The boundary with AI-in-marketing
Retail CX projects involving AI tend to attract a scope question: should we use this to personalise outbound marketing as well?
The short answer is that it is a different implementation, a different data governance requirement, and a different risk profile. AI-assisted inbound CX — where a human agent reviews and approves every response before it reaches the customer — sits in a different category from AI-driven outbound personalisation, where the system selects and sends without a human in the loop for each message.
The governance requirements are different. The consent and data processing basis is different. The failure mode — an AI sending a commercially inappropriate message to a customer at scale, without a human reviewer — is substantially harder to contain than an AI drafting a response that a human agent catches before it sends.
This does not mean outbound personalisation is wrong for Gulf retail. It means it should be a separate project evaluation, not bundled into a contact-centre AI deployment. The two workstreams have different timelines, different stakeholders, and different readiness criteria.
Keeping the scope clear at the outset prevents the governance review from landing late and splitting the project in two.
The Gulf retail governance context
Gulf retail groups handling loyalty programme data sit inside the UAE PDPL and, for members enrolled in Saudi Arabia, the Saudi PDPL framework. The relevant questions for an Agentforce deployment are not dramatically different from BFSI — where is the member data processed, what is the audit trail for AI involvement in a response, and how does the deployment handle cross-border data for programmes that operate across multiple GCC markets?
Two areas need specific attention in the design phase.
Loyalty data as personal data. A loyalty member’s purchase history, tier status, and redemption record is personal data under PDPL. When an Agentforce integration accesses this data to generate a draft response, the data processing basis and any applicable retention or deletion requirements need to be established before the integration is designed, not retrofitted after it is built.
Cross-brand data in multi-brand programmes. Some Gulf loyalty programmes aggregate purchase data across multiple brands in a group. When the AI integration accesses that aggregated data to handle a query, the design needs to specify which data is in scope for a given interaction type and which is not. A query about a missing points credit at one brand should not surface purchase data from another brand unless the programme’s consent framework explicitly covers it.
Both of these are questions that a Discovery engagement maps in the data and integration assessment. They are not obstacles to deployment — they are design inputs. Getting them right at the start prevents the compliance review from stopping the pilot.
Why retail projects stall
Retail Agentforce projects stall for one of three reasons. The loyalty data and CRM are not integrated at the level the deployment requires — the contact-centre tool cannot see the loyalty record without a manual lookup. The scope expands mid-project to include outbound personalisation, adding governance complexity that was not in the original timeline. Or the use cases that were designed for the pilot do not match the interactions that account for the actual contact-centre volume.
The first is a data integration problem that a Discovery assessment maps precisely. The second is a scope problem that a clear boundary prevents. The third is a use-case problem that an accurate contact-type inventory at the start of Discovery solves.
What Discovery produces for a retail client
At the end of a Discovery engagement with Emerge, a Gulf retail group has four things.
A contact-type inventory that identifies the interaction categories suited to AI-assisted handling and the interaction categories that should remain entirely human — with the rationale for each boundary stated explicitly.
A data and integration readiness assessment that maps the current state of CRM and loyalty platform integration and identifies what is required before a pilot is viable.
A governance requirement log for the personal data and consent questions relevant to the specific programme and the GCC markets it operates in.
A phased pilot scope — typically two to three contact types, limited channels — that produces a measurable reduction in handling time within eight to twelve weeks and provides the evidence base for a broader rollout decision.
The goal is not a loyalty transformation roadmap. It is a specific, time-bounded answer to whether Agentforce can reduce handling time in your contact centre, what it takes to reach production, and what the first measurable outcome is.
If you are a retail CX or digital leader evaluating that question, this is the conversation Discovery is designed to have.