Metro’s AI-enabled loyalty refresh shows how Singapore retailers can shift from discounts to real customer engagement with omnichannel, real-time rewards.
Most retailers don’t have a “loyalty” problem. They have a discount addiction.
If your retention plan is mostly vouchers, member-only sales, and bigger and bigger promo blasts, customers will behave exactly as you trained them to: they’ll wait, cherry-pick, and disappear until the next markdown.
That’s why Metro Singapore’s February 2026 loyalty relaunch is worth paying attention to—especially if you’re following our “AI dalam Peruncitan dan E-Dagang” series. Metro isn’t trying to out-discount everyone. It’s rebuilding loyalty as a relationship, using an AI-enabled loyalty engine (via Eagle Eye) to connect offline and online data, decide the next-best reward in real time, and support a more omnichannel way of shopping.
This matters in Singapore because the retail fight is intense: malls are saturated, eCommerce is habitual, and customer acquisition costs rarely go down. When retention becomes the growth engine, AI-driven customer engagement stops being optional.
What Metro’s loyalty refresh actually changes (beyond branding)
Metro’s new program (Treasured by Metro) looks familiar at first glance: a tiered loyalty structure with escalating benefits like birthday discounts, higher earn rates, free parking, and invite-only experiences.
The shift is how the program can now behave.
Instead of treating loyalty as a static set of rules (“Spend X, get Y”), the refreshed approach is built to support:
- Promotional earning windows (earn boosts during specific periods)
- Omnichannel earning and redemption across in-store and online
- Behavioural rewards (rewarding actions beyond purchase)
- Future expansion into more AI-powered personalisation
Here’s the stance I’ll take: tiers aren’t the innovation—decisioning is. A tier table on a website doesn’t create emotional stickiness. Timely, relevant recognition does.
Why “promotional earn” beats blanket discounting
Answer first: Promotional earning changes customer behaviour without permanently eroding your margin.
A discount lowers price now. A points multiplier can achieve a similar urgency while keeping your base price intact, especially if your redemption rules are designed carefully.
Practical examples Singapore retailers can copy:
- Double points on new-season categories to shift demand (beauty launches, CNY collections)
- Bonus points for weekday visits to smooth peak weekend traffic
- Accelerators for basket-building (e.g., +300 points when cart includes beauty + home)
This works because you’re not just paying for volume. You’re paying for the behaviour you want.
The real engine: one customer view + AI decisioning
Metro’s stated core technology move is straightforward: connect customer data from offline and online systems into a single loyalty engine, then use AI-driven rules/decisioning to decide:
- Who should receive which reward or message
- When they should get it
- Which channel should deliver it (in-store, email, app, etc.)
The customer doesn’t “see” the tech. They feel like the brand remembers them.
What “real-time decisioning” looks like in a retail loyalty program
Answer first: Real-time decisioning means your loyalty program reacts to customer behaviour as it happens, not weeks later after a monthly campaign calendar.
A few concrete moments:
- A member browses online but doesn’t buy: send a tailored points booster for the specific category they lingered on.
- A customer hits a tier threshold at the cashier: upgrade immediately and trigger the new tier benefits on the same trip.
- A lapsed member returns in-store: recognise them with a “welcome back” offer that’s not just 20% off everything.
This is where many teams get stuck. Not because AI is hard—but because data, identity, and operations are messy.
The unglamorous part: identity matching across channels
Answer first: Omnichannel loyalty fails when you can’t reliably recognise the same person in-store and online.
Before you chase fancy personalisation, make sure you can do these consistently:
- Match POS transactions to a member ID (phone/email/app barcode)
- Unify eCommerce accounts with loyalty IDs
- Handle edge cases (shared family phone numbers, staff purchases, marketplace orders)
In practice, this is the “AI business tools” story most leaders miss: the best results come from boring foundations.
Behavioural rewards: the next frontier (and the easiest way to stand out)
Metro notes it’s exploring behavioural earn, like rewarding customers for actions beyond transactions (e.g., sharing on social media after attending events).
Answer first: Behavioural rewards create loyalty signals that discounts can’t buy: advocacy, community, and habit.
If you run a retail or eCommerce business in Singapore, behavioural rewards are one of the fastest ways to differentiate because most programs still measure only spend.
Behavioural reward ideas that don’t feel gimmicky
Use actions that are measurable and tied to value:
- Event attendance + post-event feedback (earn points for checking in and completing a 30-second survey)
- Product education (earn for watching a short guide for skincare, appliances, baby products)
- Sustainable behaviours (earn for refill purchases, recycling drop-offs, or choosing consolidated delivery)
- Service adoption (earn for booking styling consults, alteration services, or click-and-collect)
The rule: reward behaviours that reduce your costs or increase lifetime value—not vanity metrics.
A quick note on Singapore compliance and trust
Answer first: Personalisation only works if customers trust how you handle their data.
Singapore shoppers are practical. They’ll trade data for value, but they don’t like surprises. Make it explicit:
- What data you collect
- What it’s used for (e.g., relevant offers, better service)
- How to control preferences
This is a brand issue as much as a tech issue.
What other Singapore businesses can learn from Metro (a practical playbook)
Metro is a large, established retailer, but the lessons apply to smaller brands too. The key is sequencing.
Answer first: Start with the loyalty mechanics that change behaviour, then layer AI personalisation once the basics are stable.
Step 1: Define the behaviours you want (not the offers you want to send)
Write this like a scorecard:
- Increase visit frequency from 1.2 → 1.6 trips/month (example target)
- Grow online-to-offline conversion (click-and-collect adoption)
- Reduce dependency on storewide discount events
- Improve repeat purchase in 60 days for selected categories
If you can’t describe the behaviour, AI won’t rescue you.
Step 2: Build an “earn and burn” structure that protects margin
A common local mistake is being generous on earn rates and vague on redemption.
A margin-safe structure typically includes:
- Clear redemption thresholds (avoid tiny redemptions that feel like noise)
- Category-based redemption rules (protect low-margin SKUs)
- Expiry and “points liability” tracking
- Promo multipliers aimed at incremental spend
Step 3: Get omnichannel right with two metrics
Answer first: Omnichannel loyalty is working when recognition and redemption are frictionless.
Track:
- Recognised rate: % of transactions tied to a known customer ID
- Redemption rate: % of members redeeming value (not just earning)
Low recognised rate means your data is incomplete. Low redemption rate means your program is irrelevant (or too hard).
Step 4: Use AI for next-best-action, not just segmentation
Many teams stop at “segments” (VIP, lapsed, bargain hunter). That’s 2016 thinking.
AI-driven loyalty platforms can help with next-best-action decisions such as:
- Offer type (points boost vs. experience invite vs. service perk)
- Timing (now vs. after second visit)
- Channel (app push, email, POS receipt)
The win isn’t “more personalisation.” The win is less wasted incentive.
Step 5: Treat loyalty like a product, not a campaign
Answer first: Your loyalty program should have a roadmap, not a promo calendar.
Borrow product habits:
- A/B test reward types
- Run small pilots by store or category
- Monitor tier progression and drop-off
- Review monthly: what behaviours improved, what margin did it cost?
That’s how you move from “marketing cost” to “growth asset.”
People also ask: What makes an AI loyalty program effective?
An AI loyalty program is effective when it consistently increases customer lifetime value with less discount spend. In practice, it needs three things:
- Unified customer data (in-store + online)
- Real-time decisioning (next-best reward/message)
- A value exchange customers understand (clear benefits + trust)
If any one of these is missing, you’ll feel it quickly: irrelevant offers, low engagement, or rising promo costs.
Where this trend is heading in 2026: loyalty built around lifetime value
Metro’s leadership frames the broader industry issue well: retail got trapped in a loop where incentives equal discounts, and customers learned to wait for sales.
Answer first: The 2026 direction is loyalty designed around customer lifetime value (CLV), not short-term sales uplift.
Expect more Singapore retailers to prioritise:
- Personalisation driven by actual behaviour (not guesswork)
- Experience-based perks (events, services, early access)
- AI decisioning that reduces manual campaign workload
- Loyalty integration across ecosystems (marketplaces, partners, cross-border)
If you’re operating in retail or eCommerce here, the competitive bar is being set by brands willing to build the data foundation and commit to a smarter program design.
What to do next if you’re planning your own loyalty refresh
If Metro’s story triggered a thought like “Our loyalty program hasn’t changed in years,” you’re not alone.
Start small but be decisive:
- Audit your data: can you identify customers across channels reliably?
- Replace one storewide discount with a points multiplier tied to a specific behaviour.
- Launch one behavioural reward that’s genuinely useful (event feedback, service booking, sustainable actions).
The forward-looking question for Singapore retail is simple: Will your loyalty program be a discount engine—or a relationship engine—by the next peak season?