What the IBM-NRF 2026 Study Tells Singapore SMEs About How E-Commerce Is Changing

The shopping funnel has quietly moved upstream, and most Singapore SMEs have not adjusted for it.

That is the practical reading of the IBM-NRF "State of the Consumer 2026" study, which surveyed over 18,000 consumers across 23 countries in Q3 2025. The headline number — 45% of consumers now use AI as part of their buying journey — is significant on its own. But the part that matters for SMEs is where in the journey AI is showing up.

It is not at checkout. It is at the start.

Key Takeaway: A growing share of e-commerce decisions are now being made inside AI assistants — ChatGPT, Gemini, Perplexity, Copilot — before a customer ever reaches your website, your ads, or your store. If your product information, reviews, and pricing are not legible to AI, you are increasingly invisible at the moment buyers are choosing what to buy. The old SEO playbook still matters, but it is no longer enough on its own.

Written by Derek Chua, digital marketing consultant and founder of Magnified Technologies. Derek works with Singapore SMEs on SEO, AI search visibility, and how shifts in consumer behaviour translate into practical changes in marketing strategy.

What the Study Actually Found

The IBM Institute for Business Value, working with the National Retail Federation, ran two surveys in Q3 2025: a consumer survey of more than 18,000 people across 23 countries, and an executive survey of 200 senior leaders across retail, consumer goods, and e-commerce.

The numbers worth paying attention to:

• 45% of consumers now use AI in some part of the buying journey.
• 41% use AI to research products before buying.
• 33% use AI to interpret reviews.
• 31% use AI to hunt for deals and compare prices.
• 72% still shop in physical stores — but the decision of what to look for increasingly forms before they walk in.
• 30% say they want a smart home with an AI personal shopper handling routine purchases.
• 54% of brand executives admit to persistent data challenges that prevent them from adapting.

The study segments respondents by engagement style, price sensitivity, and AI literacy. The consistent thread is that AI is not replacing shopping — it is reshaping the pre-purchase phase.

This matters because the pre-purchase phase is where SMEs traditionally compete on SEO, content, and ads. If buyers are now asking ChatGPT "what's the best accounting software for a small Singapore F&B business?" before they ever Google anything, your visibility in Google is a downstream metric of your visibility in AI.

Why This Is Bigger Than "Another AI Trend"

It is easy to read a stat like "45% of consumers use AI" and mentally file it next to other futurist headlines.

But this study is different in two ways.

First, the sample is large and recent. 18,000 respondents across 23 countries in Q3 2025 is not a vendor whitepaper — it is one of the largest pieces of consumer behaviour research published this cycle. The signal is real.

Second, the behaviour change is happening faster than the last one did. The shift from offline to online shopping took ten to fifteen years to fully play out. The shift from search-first to AI-first research is happening in twelve to twenty-four months. Multiple analysts, including eMarketer and Morgan Stanley, are now projecting that AI shopping agents will account for a meaningful share of online spending by 2030 — with Morgan Stanley estimating around 25%.

Whether the exact number lands at 25% or 15% is not the point. The point is that even modest adoption rewrites how customers find you.

Three Things That Change for Singapore SMEs

If you sell to consumers, or you sell to other businesses whose buyers do their own research, the implications are concrete.

1. Your product pages now have two audiences

Until recently, your product pages were written for two readers: humans, and Google's crawler. That is no longer the full picture.

Now your pages also need to be readable by large language models that summarise, compare, and recommend. That changes the writing in subtle but important ways:

• Specifications must be explicit, not implied. A product page that says "great for small offices" is invisible to an AI being asked "what laptop do I buy for a 5-person office in Singapore?" A page that says "designed for teams of 3 to 10, with multi-user licensing from S$X per seat per month" can be cited.
• Comparisons should exist on your site, not only on competitor blogs. If the only "Product A vs Product B" content on the internet is written by your competitor, the AI's summary of you will be your competitor's framing.
• Structured data matters more, not less. Product schema, FAQ schema, review schema — these are the cleanest way to feed AI assistants accurate facts about your product.

2. Reviews have become more important, not less

It would have been reasonable to predict that AI summaries reduce the importance of individual reviews. The IBM-NRF data points the other way.

33% of consumers now use AI specifically to interpret reviews — to summarise sentiment, extract themes, and surface trade-offs. That means the content of reviews is being read more carefully than ever. AI does not skim. It reads everything and then weights it.

For an SME, the practical implication is:

• A handful of five-star reviews with thin content are weaker than a moderate volume of detailed reviews that mention specific use cases, problems, and outcomes.
• Negative reviews you have responded to thoughtfully are signals of trust, not weakness — AI assistants surface that pattern.
• Review velocity matters. Reviews that are six months old are aged out of most AI summaries.

If you have not asked your last 50 customers for a review, that is the highest-leverage marketing task this month.

3. Deal-hunting has been automated

31% of consumers use AI to compare prices and find deals. This is not new behaviour — it is old behaviour at higher speed.

What is new is that the comparison is happening across competitors you do not know about. Voucher sites, marketplace sellers, regional alternatives, and overseas options that a manual buyer might never have found are now one prompt away.

For Singapore SMEs, this raises three uncomfortable questions:

1. If a customer asks an AI assistant to find the cheapest version of what you sell, will the answer be you?
2. If the answer is not you, do you have a clear reason — service, speed, local support — that the AI can also surface?
3. Is that reason written down, on your site, in language an AI can quote?

The businesses that will hold pricing power are the ones whose differentiation is concrete and discoverable, not the ones whose differentiation lives only in the head of the founder.

What This Does Not Mean

A few things this study does not prove, despite how some commentators are framing it.

It does not mean Google is dead. Google still drives the majority of e-commerce traffic in Singapore. AI search is a layer on top of, not a replacement for, traditional search — and Google's own AI Overviews are now the dominant interface for many queries. If you have been thinking about how AI Overviews change SEO, the answer is to do both, not to choose.

It does not mean physical retail is over. 72% of consumers still shop in stores. The shift is in decision-making, not in transaction. People are doing their research with AI and then buying offline, online, or both.

It does not mean every SME needs to adopt agentic commerce tomorrow. The fully autonomous AI shopper — the agent that buys on your behalf without confirmation — is still a small share of actual purchases (around 13% in the study). The bigger near-term shift is the advisory AI: the assistant that recommends, summarises, and shortlists. That is the audience to write for now.

What to Actually Do This Quarter

If you take the IBM-NRF findings seriously, here is the smallest set of things worth doing in the next 90 days.

1. Audit how AI describes you today. Ask ChatGPT, Gemini, and Perplexity — separately — "What does [your company] do, and who is it for?" Then "What are alternatives to [your company] in Singapore?" The answers tell you what AI thinks of you, what your competitive set looks like in AI's eyes, and where the factual gaps are.

2. Make your product pages legible to AI. Specifications, pricing tiers, use cases, FAQs, and comparisons in plain text on the page. Add product schema and FAQ schema. The goal is for an AI summarising your page to be able to extract correct facts without inferring them.

3. Refresh your reviews. Ask recent customers for detailed reviews on Google, on your site, and on any industry-specific platforms (e.g. Carousell, Lazada, Shopee for retail; Clutch or G2 for B2B). Aim for volume and specificity.

4. Write your own comparison content. "[Your product] vs [main competitor]" pages, written honestly. If you do not write them, your competitor will, and that becomes the source AI assistants quote.

5. Track AI-driven traffic separately. GA4 will not split out AI assistant referrals cleanly yet, but referrals from chat.openai.com, gemini.google.com, and perplexity.ai are a leading indicator of how visible you are. Set up a custom segment.

None of these are large projects. They are the quiet, compounding work that distinguishes businesses that adapt from businesses that get displaced.

The Bigger Picture

The IBM-NRF study is one data point, but it sits in a broader pattern. eMarketer, Morgan Stanley, commercetools, and Search Engine Land have all published similar conclusions in the last six months. The directional story is consistent: AI is becoming the layer between your product and your customer.

The businesses that thrive in this shift will not be the ones with the most AI tools internally. They will be the ones whose products, pricing, and reputation are easiest for an AI to understand and recommend on behalf of a buyer.

That is a writing problem, a structured data problem, and a customer review problem — long before it is an AI problem. Most Singapore SMEs can make meaningful progress on all three in a quarter, without changing their tech stack at all.

The window to do this work while it is still uncrowded is open now. It will not be open in twelve months.

Working With Magnified

Magnified is a Singapore-based digital marketing agency. We help SMEs across professional services, e-commerce, and B2B services adapt their content, structured data, and review systems for AI search visibility — alongside traditional SEO and paid media.

If you are not sure how visible your business is inside ChatGPT, Gemini, and Perplexity today, we offer a free 30-minute audit. We will run the queries with you, show you what AI says about your business and your competitors, and give you a candid view of what to fix first.

Frequently Asked Questions

How is AI search different from regular Google SEO? Traditional SEO optimises a page to appear in the ten blue links. AI search optimisation makes a page legible to large language models that summarise answers rather than list pages. The fundamentals overlap — clear content, structured data, authority — but AI search rewards specificity, comparisons, and well-organised facts more than keyword density.

Should Singapore SMEs invest in agentic commerce now? Probably not as a primary channel — the share of purchases completed autonomously by AI agents is still small. But the advisory layer (AI assistants recommending what to buy) is already material at 45% adoption. Optimising for that is the higher-leverage move in 2026.

What is the fastest way to see if I'm visible in AI search? Open ChatGPT, Gemini, and Perplexity. Ask each one: "What are the best [your category] companies in Singapore?" and "What does [your company name] do?" Within five minutes you'll know whether you're cited, mis-described, or invisible.

Does this affect B2B SMEs or just consumer e-commerce? Both. The IBM-NRF study focused on consumer behaviour, but B2B buyers are consumers too — and they increasingly use AI to shortlist vendors before any sales conversation. If anything, the shift is faster in B2B because professional buyers adopt productivity tools quickly.

How long will this trend take to play out? The IBM-NRF data and adjacent research suggest 12–24 months for the advisory AI shift to fully reshape pre-purchase behaviour, and 3–5 years for autonomous agentic commerce to reach meaningful scale. SMEs that start adapting in 2026 will be well-positioned. SMEs that wait until 2028 will be playing catch-up against competitors whose AI visibility is already established.

---

If this guide was useful, you may also find these helpful: Google AI Overviews Already Appear on Half of All Searches and The 2026 Digital Marketing Checklist for Singapore SMEs.