The Loyalty Argument in 2027: Six Named Voices on AI and Data

Spending plans are tilting back toward retention next year. Loyalty and CRM are quietly taking over half the marketing budget on average, and the finance review will ask the obvious question before January is out: what does the next unit of loyalty spend actually buy. More loyalty, or more data. The two are not the same thing, and most boards have never had to tell them apart.

I went and read what the people who do this for a living are telling the market about 2027. Six names worth listening to, with their positions on record, and one position that most agencies will not put in writing but you need to hear anyway. The views do not line up, which is why the piece is worth your time. If every voice in a trend survey agrees, somebody is nodding along to be in the picture.

Three decades of watching commerce has taught me that the quiet lesson of a budget cycle like this one is almost never the headline figure. The headline is loyalty share rising. The quiet lesson is that loyalty is now where first-party data is kept, and that is a different budget line than engagement. Treat it as the same and you buy a points scheme that fattens a ledger nobody uses.

The money is moving back into retention

Antavo’s 2026 Global Customer Loyalty Report, a survey of 3,000 marketers and 10,000 consumers published 3 February 2026, puts loyalty and CRM at an average of 51.5% of the total marketing budget, with 92.7% of program owners reporting positive ROI at an average of 5.3× (Antavo GCLR 2026).

Attila Kecsmar, CEO and co-founder of Antavo, puts it this way in the launch note: “Another year of rising satisfaction, ROI, and investment only goes to highlight the degree to which brands need to be laser-focused on their loyalty programs.”

Open Loyalty’s Loyalty Trends 2026 read, built on 170 loyalty and CRM leaders and updated 24 April 2026, lands in the same place from a different angle: 59% of loyalty professionals name customer lifetime value as this year’s primary KPI, 42% expect gamification to deliver the biggest impact, and 35% report explicit internal pressure to prove loyalty ROI to the business.

So the budget line is up and the ROI case is defensible on paper. What is less clear is which part of the retention economy the money is actually paying for.

What the published forecasts say about 2027

Three positions you can put in front of a board without caveat, each from a named, linkable source.

Brad Jashinsky and Halle Stern, Gartner Market Guide for Loyalty Program Vendors (September 2023): one in three businesses without a loyalty program today will establish one by 2027, specifically to secure first-party data collection and retain high-value customers. The forecast is widely cited in operator press coverage, including ebbo’s reading of the Market Guide.

Deloitte 2026 Retail Industry Outlook Survey (January 2026, 330 retail executives surveyed October–November 2025): 44% of global retail executives expect generative AI to weaken brand loyalty this year by shifting consumer choice toward value and fit over brand recognition, and 50% expect a larger impact in 2027 or beyond. Figures and framing per EMARKETER’s reading of the Deloitte release.

Troy Townsend, CEO at Zitcha, in a published Mediaweek piece: “By effectively harnessing first-party customer data, brands can significantly enhance their return on investment and foster greater customer loyalty.” Older position, but it is the clearest published statement of the data-first argument from an operator who sits inside the retail-media stack.

These three positions do not conflict. They describe three different things: the market will add programs, AI will undermine brand-level stickiness, and the operational asset is the data rather than the points. Together they describe a 2027 that will spend more on loyalty while loyalty itself gets structurally harder to buy.

The dissent that needs to be in the room

The strongest service I can do a marketing or e-commerce lead planning this spend is not to hand them the industry’s own PR. It is to put the serious sceptic in the room, and there are two worth listening to.

Jon Lombardo, Global Lead of The B2B Institute at LinkedIn, writing in “The Loyalty Lie”: “what drives growth isn’t customer loyalty, but customer penetration.” His piece argues that Frederick Reichheld’s 1997 foundational case for loyalty economics rested on arithmetic errors and reconstructed spreadsheet data, and that the Ehrenberg-Bass tradition has been more right about the mechanics ever since.

Dr Byron Sharp, Director of the Ehrenberg-Bass Institute for Marketing Science, has made the empirical case for two decades: loyalty programmes produce slight loyalty effects and do practically nothing to drive growth, and brands grow by acquiring more new customers rather than squeezing more frequency from the ones they already have (Ehrenberg-Bass summary of the position is on the Institute’s own site, with the longer argument in How Brands Grow).

Neither of them is saying your loyalty programme is worthless. They are saying it is unlikely to be a growth engine, and that if you buy it as one you will spend into a result that will not materialise. That is exactly the mistake a 51.5% budget share makes easy.

Safe bet, and outrageous prediction

Distilling the published positions above into the two columns Econsultancy’s forecast pieces run on, so a reader can take them to a planning meeting rather than paraphrase them from memory.

Safe bet. Loyalty budget share holds or rises again, following Antavo’s trend line. The data case for a programme becomes the primary business case, following Gartner. AI assistants chip at category-level brand recognition this year, following Deloitte’s own executives. Programmes keep running whether they drive growth or not, because the budget is already allocated. The quiet one, from Sharp and Lombardo, is that the growth report will keep being filed and the growth will keep coming from penetration, not from the programme.

Outrageous prediction. Half the enterprise retailers in the Deloitte sample carry an AI-visibility line in their 2028 loyalty P&L, measuring whether an assistant surfaces the brand at all. If that reads as too much, re-read Deloitte’s 50% expecting larger impact in 2027 or beyond. And the CFO question that kills more loyalty budget next year is not “what is the ROI” but “where does the data this programme produces actually get used, and by which system”. Programmes that cannot answer that lose their renewal.

Stop, and start

Across the published positions above the stop column and the start column are strikingly consistent, which is worth the reader’s attention: a trend survey where every voice says the same thing on execution and disagrees only on prognosis is one you can plan from.

Stop. Treating programme engagement metrics (points earned, tier movement, redemption rate) as proof of commercial outcome. Antavo’s own 2026 research puts about one quarter of US programme points as unspent, roughly $10 billion in theoretical consumer savings that never leave the ledger. That is engagement without redemption, and it inflates every dashboard built on earned-points telemetry. Stop measuring loyalty on humans only: the analytics stack runs on JavaScript in a browser, AI crawlers do not execute JavaScript, and the first surface where a brand is losing recognition is the one your dashboard cannot see.

Start. Modelling the programme as a consented first-party data pipeline with a loyalty UX on top, rather than the other way round. Deciding, before the next tier redesign is commissioned, which downstream system (ESP, retail media connector, personalisation engine, customer service) the data is actually feeding, and what breaks when it stops arriving. Reporting programme ROI alongside the acquisition cohort for the same quarter, so the board can see whether growth came from the programme or from acquisition the programme sat next to. If Sharp and Lombardo are right that penetration drives growth, that is the comparison that keeps the budget honest.

What this means inside a WooCommerce store

Strategy that is not shippable is not advice. Here is what the argument above looks like when the ticket actually lands in a build team’s queue.

A loyalty programme on a WooCommerce estate is three systems stacked on each other, and the stack is more honest about them than most dashboards.

• The programme itself. A plugin surface (points, tiers, rewards, referral codes). WooCommerce Points & Rewards for a thin build, LoyaltyLion or Yotpo for the heavier spec, YITH as a middle. This is what the customer sees.
• The data it produces. Order history already lives in the WooCommerce orders tables; the programme writes member identity, point balances, tier transitions and redemption events into its own tables and into user meta. Every one of those rows is first-party, consented and purchase-linked, which is the Gartner argument in concrete form.
• Where the data goes next. Klaviyo, Mailchimp, Mautic for email; a CDP for the serious segmentation case; a feed exposed through the Store API or a product endpoint so that an AI assistant can read your catalogue and your membership rules at the same time. If this third layer does not exist, the programme runs the first two and the Gartner case does not land for your estate.

What this costs to ship properly on WooCommerce for a mid-market retailer, in rough order of effort: a day or two to install and configure the loyalty plugin, two to five days to clean the identity layer so that guest-checkout orders attach back to the right customer record, five to ten days to wire the data pipeline into whichever downstream system actually uses it, and an open-ended governance line for the policy that decides what event triggers what message. The last line is where most programmes break: the plumbing works, the policy was never written, and the ESP ends up sending a 7% discount trigger to a customer who placed a full-price order the same morning.

A sanity test before any of that goes near the brief: can the person asking for the programme name, in one sentence, which downstream system the data feeds and what the trigger is. If they cannot, the programme is not a programme. It is a tier page with a backend.

What this does not resolve

The sceptic’s case, that programmes do not materially drive growth, does not have a satisfying counter in the published 2026 data. Antavo’s 5.3× ROI headline is the industry’s own measurement of its own output, which is not nothing but is also not independent evidence, and the loyalty-lie reading is one that no vendor report will adopt. The honest position for a buyer is to treat the programme as a data and retention asset that will defend margin, to measure that defence against the acquisition cohort from the same quarter, and to plan the budget on that basis rather than on a promise of extra growth the independent research does not support.

If AI assistants do become the first screen between a brand and a shopper, as Deloitte’s 44% suggests they already are becoming for this year, the next thing to put on the loyalty roadmap is not another tier. It is instrumenting whether the programme reaches the surfaces where the shopping conversation now happens, and whether the data it produces is readable by the systems that will decide which brand the assistant surfaces at all. That is a 2027 question, and it is the one worth planning for now.

If you want help turning that into a build on your own estate, with the data pipeline specified rather than hand-waved and the KPI picked before the plugin is installed, that is the conversation we would like to have.