RRetail AI
Research
← retailresearch.eu
Retail AI Research · Field Note 001

The delivery promise is a product feature, not a delivery policy

13 November 2026 · Published without any participating retailer. No shop data was used. Sources and limits below.

Why this note exists. We have no clients, so we can publish things a consultancy with clients cannot. A consultancy cannot publish a finding that embarrasses a chain, because the chain is the client. We have no such constraint. This is the first attempt to show what that means in practice: an analysis that uses no privileged data, says clearly what it does not establish, and is worth a shopkeeper's ten minutes.

The question

Belgian retailers are told that delivery is the number-one driver of online conversion. What does that mean for an independent shop, and what should a shop actually do about it?

The decision this is meant to inform

Not “should you offer faster delivery” — that is a budget question, and we cannot answer it for a shop we have never met.

The decision is: how should an independent retailer treat its delivery promise as part of the product rather than as a shipping setting?

What we used

SourceDateWhat it gives
Eurostat, E-commerce statistics for individualsextracted Feb 2026, covering 2025EU-wide consumer problems and online purchase behaviour
KBC Economics, Retail vacancy rate in Belgium at record high (Johan Van Gompel)20 Oct 2026Belgian retail vacancy, retail density, enterprise birth rate
Landmark Global, Top 10 Essential Facts About Belgium Ecommerce2026 editionBelgian consumer expectations on delivery cost visibility

No shop data. No proprietary dataset. No retailer contact.

What the evidence shows

1. Delivery is the most commonly reported online purchase problem in the EU, by a wide margin.

Established

In 2025, 63% of EU online buyers reported no problem at all. Among those who did report problems, “delivery speed slower than expected” was named by 20% — the single most frequent issue. Next came a website that was hard to use (12%), a wrong or damaged delivery (10%), difficulty finding information about guarantees and legal rights (8%), and costs higher than indicated (5%).

Eurostat, isoc_ec_ib20, extracted February 2026.

2. Belgian retail is under structural pressure that has nothing to do with AI.

Established

Belgian shop vacancy reached 11.9% at the beginning of 2026 — the highest since records began in 2008 — covering roughly 23,000 premises. Of that vacancy, 32% had been empty for more than three years and 20% for more than five. The strongest recent increase is in Flanders, where the rate rose to 10.8% and then 11.9%, ending a long-standing gap with the national figure. The Netherlands, by comparison, sits at 7%.

KBC Economics, 20 October 2026, citing Locatus inventory data.

3. Belgium has an unusually low rate of new retail business.

Established

The enterprise birth rate in Belgian retail was 5.8% in 2023, against 8.8% for the EU as a whole; only Italy and Cyprus scored lower. Belgium also has retail density one and a half times higher than the Netherlands.

KBC Economics, 20 October 2026, citing Eurostat.

4. In Belgium specifically, delivery cost transparency is the leading delivery-related preference.

Probable

60% of Belgian shoppers want delivery costs shown before checkout, and 85% follow their parcel through tracking notifications.

Landmark Global, 2026 edition. This is a commercial publication and the underlying survey methodology was not visible to us. We treat it as indicative, not established.

5. For an independent shop, the delivery promise is more likely to be a conversion constraint than a logistics cost.

Probable

Reasoning: the EU evidence shows delivery is the most common failure point, while the Belgian consumer research indicates cost visibility before checkout is the dominant preference. Both point at the same mechanism — the customer decides, and the decision is made partly on delivery terms. A shop that treats delivery as a courier setting is treating the most-frictioned part of the purchase as its most adjustable.

We rate this Probable rather than established, because it rests on consumer-side evidence about problems and preferences, not on measurement of how delivery terms actually affect conversion in Belgian independent retail. That measurement does not exist publicly, to our knowledge.

6. The right move for most independent shops is to make one delivery promise and keep it, rather than to offer a menu of options.

Possible

Reasoning: the third most common EU complaint is a wrong or damaged delivery, and the fourth is difficulty finding information about guarantees and legal rights. Both point at uncertainty rather than at speed. A single, clearly stated promise reduces the number of decisions a customer must make at the moment of purchase, and reduces the number of promises a small operation has to keep.

Rated Possible. It is a plausible mechanism, not a measured one.

7. “Add AI to fix this” is not supported.

Contradicted

Any framing in which the core problem is a lack of AI capability is contradicted by the structure of the evidence: the leading issues are about the clarity, predictability and consistency of a promise, and about cost visibility. These are operational and merchandising problems. A recommendation engine that chooses a better delivery option for a customer who was already going to buy changes nothing about whether the parcel arrives when promised.

What we deliberately did not conclude

Plausible next steps, ranked by consistency and cost to test

Explanation for why independent shops struggleConsistent with evidence?What would distinguish it?Cost to test
Customers abandon because delivery terms are unclear or surprising at checkoutYes — directly matches the “costs higher than indicated” and delivery-speed complaintsChange one site's delivery presentation, measureLow: one page, two weeks
Customers abandon because the shop cannot reliably meet a promise it makesPossible — matches the wrong/damaged delivery complaintAudit actual delivery performance against stated terms for one monthLow to medium: needs data the shop already has
Customers abandon because the promise does not fit the purchase moment (e.g. a gift needed next week)Possible — not directly tested in the sources we usedAsk customers which delivery option they needed, per orderLow: a question at checkout
Delivery is a symptom, not a cause, and the real issue is assortment or priceCannot be excluded with public dataRequires shop-level conversion dataHigh: needs participant data

What we would need to go further

DataPeriodWhy it would change the answer
Order-level delivery cost as presented at checkout vs. as charged, for one shop3 monthsTests explanation 1 directly; this is the gap in our reasoning
Share of orders where stated delivery date was missed3 monthsSeparates “unclear promise” from “broken promise”
Customer-stated reason for abandonment at checkout1 monthWould test whether delivery is even the binding constraint
Order value and margin before and after a delivery-presentation change6 weeksThe only way to estimate anything commercial

Method and limits

Method. We collected public statistics on EU online purchase problems and Belgian retail structure, then applied the scope rule and finding-status discipline from our diagnostic template: each statement carries a status, evidence and an alternative that is not excluded. No model was trained and no prediction was made. This is a structured reading of public evidence, not a study.

Limits. This note cannot establish any causal link. It uses EU-wide consumer data applied to a Belgian context, one commercial publication of unverified methodology for the Belgian-specific delivery-preference figures, and no shop-level evidence whatsoever. It contains no AI system and no AI finding. Its value is that it is cheap to check and easy to disagree with — if you think the reasoning is wrong, say so, and we will correct the note or withdraw it.

Confidence. We are confident in statements 1, 2 and 3, which rest on official and well-documented published sources. We are moderately confident in statement 4 and explicitly uncertain about 5 and 6. We are confident that 7 is contradicted by the structure of the evidence rather than by direct measurement.

Evidence checked on: 13 November 2026.

Sources
  • Eurostat, E-commerce statistics for individuals (data extracted February 2026) — eurostat.ec.europa.eu
  • KBC Economics, Retail vacancy rate in Belgium at record high, Johan Van Gompel, 20 October 2026 — kbc.com
  • Landmark Global, Top 10 Essential Facts About Belgium Ecommerce: 2026 Edition — landmarkglobal.com

If you run a shop, and you read this far

You will notice that this note does not tell you what to do. It cannot: it has never seen your shop.

If you want the version that does, the offer is simple. You describe one problem — one decision, one period, one unit — and we send back a written analysis in your language, free, with the method and the limits shown, and no follow-up sales pitch. We do not take your customer data. If a diagnostic shows something we believe generalises, we may write it up as a Field Note, and you decide whether you are named, described anonymously, or not mentioned.

Bring us one retail problem ↗