Will Brands Become Redundant? How AI Turns D2C Brands into Manufacturers

Abstract

AI assistants such as ChatGPT are changing how people shop. They compare afresh with every request and, depending on the brief, look for the cheapest offer or the best value for money. For direct-to-consumer (D2C) brands, which sell straight to shoppers, that can mean losing regulars and the pull of the brand, and drifting towards being manufacturers, where quality and price decide. The obvious escape route, advertising inside ChatGPT, is risky on thin margins: you buy impressions and clicks, not sales. This article sets out the developments with sourced figures, separates fact from judgement, and explains why affiliate marketing, where you pay only when a sale happens, with publishers as AI's information source and exclusive, permanent voucher codes as the reason to buy, is the sturdier sum for D2C shops.

1. Introduction: the situation and the problem

A brand used to be a promise: if you knew the logo, you trusted the product and bought again. That promise rested on habit, advertising and recognition. AI assistants challenge the foundation. They run on large language models (LLMs): programs trained on vast amounts of text so that they can answer questions in full sentences. Ask one "Which magnesium supplement is best for the money?" and the answer rests on data, tests and mentions, not on habit. And because the answer is already in the answer, people click through less: on Google searches with an AI summary, US users clicked a link in 8% of visits, against 15% without one (Pew Research) [28].

There is a second squeeze: advertising is getting dearer. Meta reports that its average price per ad rose 14% year on year in the third quarter of 2026 (a global figure, not limited to online shops or to the UK) [11]. And OpenAI, the company behind ChatGPT, began selling advertising inside its assistant in 2026.

This article looks at four questions:

1. Do AI assistants make shoppers less loyal and more price-conscious?
2. Where do brands survive regardless?
3. Is advertising in ChatGPT a workable answer for D2C shops?
4. Which sum adds up on small margins?

Dollar amounts are US figures, converted at the European Central Bank reference rate of 2 November 2026 (US$1 = £0.745, €1 = US$1.1216 = £0.8358) [31]. Where we have no reliable source, we say so and label the statement as our assessment.

2. How AI changes buying behaviour

2.1 AI buys on facts, not on logos

An assistant has no feeling for a logo. It reads what is written about a product: price, ingredients, dimensions, reviews, delivery time. How these systems weigh sources shows in an analysis of 3,312 AI answers about electronics, wearables and household goods by the provider cloro: the most-cited sources were YouTube and Reddit (19% of answers each) and the review site RTINGS (16%). Manufacturers' own sites barely registered. The study comes from a vendor, relates mainly to English-language queries, and did not examine prices or specifications separately [21].

OpenAI itself is now betting that merchants will feed product data and offers directly, so that their products are "fully represented", and it lets them keep their own checkout [20]. A closer look also shows the limits of the thesis. A research preprint (not yet independently peer-reviewed) tested AI agents shopping. Its finding: how much price, ratings and reviews count differs sharply from model to model, and the agents show a strong bias towards products in the top positions [22]. "AI buys purely on facts" is therefore too simple. It is truer to say that AI buys on the information it finds, and that information comes less from the brand itself than from third parties.

2.2 Advertising evaporates on the way to the purchase

This point is our assessment, not the result of a study: a user sees an ad, then asks the assistant for a cheaper or better alternative, and a different seller gets the sale. The brand paid for the attention; a competitor collected the revenue. The easier it becomes to compare by asking an AI, the more often this can happen.

2.3 Products become interchangeable

Where a product can be described through data, such as supplements, accessories, household items or basics, an AI can compare it with alternatives. This too is an assessment. The less a product sells on emotion and story, the easier it is to replace.

2.4 Competition goes global

AI also finds the manufacturer in Asia or a retailer's own label. Market data show how strong these sellers already are:

• In the UK, Temu reached 28.4 million adults a month according to Ofcom, up from 20.1 million a year earlier. Shein reached 13.9 million (May 2025 data, reported by ChannelX) [16].
• In Germany, 51% of online shoppers have already ordered from Temu or Shein, and 33% rule out using them in future (Bitkom, survey of 1,120 people including 1,072 online shoppers) [14]. The German Retail Association (HDE) puts deliveries from the two platforms at 460,000 parcels a day. That is a study by an interest group, and the report does not give its method [17].
• Since 1 July 2026 the EU has charged a flat €3 duty on small parcels, which the report describes as "per product category" [18]. In the UK, the relief for goods up to £135 ends in March 2029 at the latest, according to Retail Gazette, with an earlier date under review [19]. These sellers' price advantage is shrinking, but it will not vanish overnight.
• In groceries, own-label products account for 52% of units and 44% of value in both the UK and Germany, according to Circana. That is grocery retail, not D2C shops, but it shows how far shoppers are willing to switch [15].

2.5 Loyalty falls

If the AI searches afresh every time, habit stops being a reason to buy. For the UK, EY has data from a survey of 1,000 consumers: value is now twice as important as brand in purchasing decisions. 42% have switched to own-label or cheaper products, and 67% say own-label meets their needs as well as brands do [12]. These are survey answers, not measured buying behaviour. For Germany, a European survey (Shopfully, 1,110 respondents in Germany) finds that product quality, at 73%, is named as decisive just ahead of price at 69% [13]. Price matters, but it is not everything.

3. Where brands survive regardless

Five reasons argue against the end of the brand:

1. AI recommends what is mentioned often and well. Reviews, tests and mentions with publishers form a kind of body of evidence. Muck Rack analysed more than 25 million links from answers by ChatGPT, Claude and Gemini. Around 27% of cited sources were journalistic articles (an analysis by a PR software provider) [25]. A brand then means: how often and how well are you mentioned?
2. Emotional products stay branded products. In fashion, beauty, lifestyle and sport, people also buy belonging. That is our assessment.
3. Users name brands directly. Someone who asks "Does brand X make product Y?" has made the choice themselves. This too is an assessment.
4. AI does not always look for the cheapest. It looks for what the user specifies, and that might be quality, suitability or delivery time.
5. A direct line to the customer bypasses the AI. Subscriptions, newsletters and communities do not need an assistant as go-between.

People's trust in AI recommendations is also limited. In a survey of 2,216 UK adults (ThoughtSpot and YouGov, reported by Resultsense), only 10.6% trusted an AI recommendation when buying, while 53% distrusted it. The report itself points to unclear figures [23]. According to a study by the advertising body IAB (600 consumers, country not stated), 46% fully trust AI recommendations and 89% cross-check them [24]. In Germany, 43% flatly refuse to let an AI do their shopping alone, while 21% could imagine it [14].

4. The middle dies

From both sides we draw a synthesis, which we label as our assessment: two types will survive.

• Cost leaders: the best quality at the lowest price, with lean distribution.
• Community brands: brands with real attachment that reach their buyers directly.

Whoever sits in between has the biggest problem: an average product, a big advertising budget and no real attachment. From now on, a brand largely means this: how often and how well am I mentioned where AI reads?

5. The margin question: is there still room for an ad budget?

If price and quality decide in a comparison, there is little room for expensive advertising. Every pound of marketing has to show up in a sale. A simple example with our own assumptions, not market data: a basket of £60 of which £17 is left after goods, shipping and returns costs. An order that cost £110 in advertising cannot be carried; one that costs £6 to £9 in commission can. So the question is not whether marketing is needed, but how to pay for it so that the sum adds up.

6. Why advertising in ChatGPT is not the answer

6.1 What is documented (as of 3 November 2026)

• Launch and availability. Advertising in ChatGPT launched in the US in February 2026 [1]. Since June 2026 a beta of the ads manager has been running in the UK, including buying by click [2][3]. For Germany and other European countries a launch was announced in August 2026, and ads are shown only to users on the free and low-cost plans [4]. How far self-service is yet available in Europe, we could not check.
• Prices have fallen. The price per 1,000 impressions (CPM) fell, according to Digiday, from a launch figure of US$60 (£44.70) to as low as US$25 (£18.60) in places. The agency Jellyfish reports an average of around US$45 (£33.50). The minimum budget fell from US$250,000 (£186,000) to US$50,000 (£37,300) [1].
• Click prices vary widely. The cost per click (CPC) ranged, according to Search Engine Journal, from US$3 to US$22 (£2.20 to £16.40), with no benchmarks from OpenAI. OpenAI recommends a maximum starting bid of US$3 to US$5 (£2.20 to £3.70); that is not a market price. An advertising software provider (Synter) measured an average of US$5.10 (£3.80) in the UK and US$10.62 in the US [5]. One case study of a shop with a high basket value gave US$4.41 (£3.30) per click and a return of 3.3 to 6.8 times spend [5]. That shows individual campaigns can pay, but these are individual results.
• Dearer than Google. Terence Einhorn of the measurement firm Measured says the cost per click in ChatGPT is around six times Google's. That is an interview remark, not a study. He says results so far have been "pretty bad" and most sales were not additional [6].
• Few clicks. According to Similarweb, the click-through rate in ChatGPT was 0.50% in June 2026 (0.68% in May), against around 6.4% for Google search ads. The figures are for US desktop and, according to Similarweb, not directly comparable, because users keep chatting after the ad [7][8].
• Growing competition. OpenAI cites more than 100,000 advertisers that have run campaigns, many on small test budgets. The market tracker Sensor Tower counted more than 1,400 individual advertisers in August 2026. The figures are not comparable [6][9]. Advertising spend reached US$109 million worldwide in April 2026 (£81 million), according to BIScience [10].
• Measurement. Reporting shows impressions, clicks, spend and conversions, but no competitive data such as share of the auction [5]. OpenAI says advertising does not influence the answers [9]. We have seen no independent test of what an impression is worth.

We could not verify the claims "2.6 times as expensive as Google search ads", "20 times as expensive as the Google Display Network" or per-click billing even for conversion campaigns. They are therefore not in this text.

6.2 The sum

An example with an assumption of ours: at US$3 to US$5 per click (£2.20 to £3.70) and a 2% purchase rate, an order costs US$150 to US$250 (£112 to £186). At a 5% purchase rate it is US$60 to US$100 (£45 to £75). On a £60 basket, there is little or no margin left in either case.

There are reasons to expect it to get dearer. That is an assessment: placements are awarded in an auction weighted by relevance [5], more advertisers push up bids, and big brands can outbid small shops.

6.3 Putting it in context

ChatGPT advertising is the old solution in a new channel: buy attention, pay up front, carry the risk yourself. For brands with big budgets it can be a testing ground. On small margins, it is in our assessment the wrong tool.

7. Affiliate marketing as the answer

7.1 Pay only when you sell

In affiliate marketing, a partner (a publisher) earns a commission when a purchase comes through their recommendation. No clicks, no impressions, just commission on a real sale. On a small margin, it is the only sum that reliably adds up.

7.2 The sums side by side

The figures are worked examples using the assumptions above, not market averages.

7.3 Publishers become AI's most important information source

AI does not invent offers, prices or vouchers; it takes them from deal sites, newsletters, creators, communities and guides. Muck Rack, cloro and Glass Digital show that AI answers lean heavily on third-party content [21][25][26]. The UK agency Glass Digital found that 420 of 479 cited, commercially classifiable editorial pages carried affiliate links (87.7%). The study says explicitly that it shows no cause, and the sample is small [26]. AI companies are now also paying for content, for example Amazon for its Rufus shopping assistant [30].

Every mention with a publisher is also a possible recommendation by AI. That benefits the shop without paying for every impression. From our own day-to-day practice in affiliate marketing we also know that AI assistants pick up voucher codes that publishers post and name them in their answers. That rests on practical experience, not on a study.

7.4 Vouchers are the reason to buy

AI finds publishers' voucher codes and can cite them in its answers. The shop appears when the user asks for the best offer, and the user buys from the shop rather than from the competition. The list price stays, the brand stays premium, and the shop still wins the price comparison. How much reliable codes matter is shown by a US test from Coupons.com (a study by a voucher provider): only 24.5% of the codes ChatGPT, Gemini and Claude gave for 101 retailers were valid [27]. AI does not check codes itself. A current, permanently valid and exclusive code has a better chance of being passed on correctly.

7.5 Exclusive codes that run permanently

Each partner gets their own code, and it stays active. Only a permanent code can establish itself with AI and search engines. With a dedicated code you see exactly what each partner delivers, even without a click, for example when an AI completes the purchase.

7.6 Data protection

Safari deletes script-set cookies after seven days without user interaction and caps cookie lifetime at 24 hours after link clicks that carry an identifier [32]. In the UK, cookie tracking depends on consent, and the Data (Use and Access) Act 2025 allows only a few narrow exemptions; beyond them, consent is required [33]. In our assessment, attribution by code is therefore sturdier.

7.7 Deal communities, the manufacturer model and brand through mentions

• Deal communities. In the UK, HotUKDeals and cashback sites such as Quidco or TopCashback; in Germany, for example, mydealz. They live on voucher codes and do not depend on Google. We could not verify current user numbers and quote none.
• It fits the manufacturer model. Anyone becoming a manufacturer needs a sales channel that costs money only on success. Affiliate marketing is outsourced selling, paid per sale.
• Brand through mentions rather than ads. An ad disappears after the click; a mention stays and strengthens the body of evidence that AI looks at.

7.8 selecdoo takes your shop to the publishers

Through the selecdoo network, your shop and your voucher codes reach newsletters, creators, communities and deal sites, which is exactly where AI gets its answers.

8. Differences between the UK and Germany

United Kingdom. Shoppers are strongly value-driven according to EY, and switch to own label [12]; Temu has a large reach [16]. ChatGPT advertising has been running there since June 2026 as a beta with click buying [2][3], and click prices average US$5.10 (£3.80) [5]. Google's AI checkout for merchants is due to reach the UK later [29]. Cashback sites and deal communities have a long tradition here.

Germany. For readers who also sell into Germany: online shopping is near universal (96% bought online in the past twelve months, 92% often through large marketplaces) [14]. Quality ranks just ahead of price [13]. ChatGPT advertising is announced for Germany, and we could not check its availability [4]. Data protection makes cookie measurement harder.

9. Recommendations for D2C shops

1. Count per order, not per click. Test every channel against what an order costs and compare it with your margin.
2. Pay for results. Where margins are thin, favour channels that charge only on a sale.
3. Give each partner their own exclusive, permanently active code. It is your reason to buy in the price comparison, and it shows you what each partner delivers.
4. Be mentioned where AI reads. Newsletters, creators, communities, deal sites and guides are your body of evidence.
5. Build the direct line. Newsletter, subscription and community do not depend on AI.
6. Do not rely on one channel. Neither Google nor ChatGPT advertising should carry you alone.

10. Conclusion

AI does not automatically make shoppers disloyal, but it compares afresh with every request. That hits brands in the middle hardest: an average product, a big advertising budget and no real attachment. Advertising in ChatGPT is the old solution in a new channel and is risky on small margins, because you pay for impressions and clicks. Sturdier is a sales channel that costs money only on success, through publishers AI draws on and exclusive, permanent codes as the reason to buy.

The question is no longer just "How do I get into the ad?" but "Who recommends me when people and AI look for the best offer?"

Sources

1. Digiday (17 April 2026): Everything is coming down: ChatGPT ads are getting cheaper
2. Digiday (19 June 2026): OpenAI expands Ads Manager to the U.K., adds CPC
3. Digiday (9 July 2026): OpenAI set to expand ads to France, Germany and Ireland
4. Search Engine Roundtable (19 August 2026): ChatGPT ads coming to Europe
5. Search Engine Journal (8 September 2026): 6 months into ChatGPT ads, advertisers still don't know what good looks like
6. Modern Retail (9 October 2026): Why ChatGPT ads won't likely impact holiday sales in a major way this year
7. Similarweb (3 September 2026): ChatGPT ads reshaping paid search
8. Search Engine Roundtable (5 May 2026): ChatGPT ads click-through rates
9. The Next Web (21 October 2026): ChatGPT ads reach 89 countries, 100,000 advertisers
10. BIScience (18 May 2026): AI ads are becoming a real competitive channel
11. Meta (21 October 2026): Meta Reports Third Quarter 2026 Results
12. EY UK (20 May 2025): Brand loyalty declines as financial concerns persist
13. e-commerce magazin (2 June 2025, in German), on Shopfully "The State of Shopping 2025": Kaufverhalten: Produktqualität und Preis am wichtigsten
14. Bitkom Research (July 2026, in German): Studienbericht Digitaler Handel in Deutschland
15. Grocery Gazette (14 April 2026), on Circana: Private label hits record 50% unit share across Europe's biggest grocery markets
16. ChannelX (12 December 2025), on Ofcom "Online Nation": Ofcom Online Nation report reveals UK top 10 retail sites
17. e-commerce magazin (18 May 2026, in German), on an HDE study: Handelsverband: Temu und Shein schaden der Wirtschaft
18. Euronews (1 July 2026): EU slaps €3 duty fee on Shein, Temu and AliExpress imports
19. Retail Gazette (18 June 2026): Government considers speeding up Shein and Temu import tax crackdown
20. CNBC (24 March 2026): OpenAI revamps shopping experience in ChatGPT after Instant Checkout
21. cloro (26 May 2026, updated 15 September 2026): What AI shopping assistants recommend
22. Allouah et al. (2025), preprint: What Is Your AI Agent Buying?
23. Resultsense (8 September 2026), on ThoughtSpot/YouGov: UK shoppers and AI recommendation trust
24. IAB (28 October 2025): AI ranks among consumers' most influential shopping sources
25. Muck Rack (as updated 7 May 2026): What Is AI Reading? (May 2026)
26. Glass Digital (2 September 2026): Do affiliate mentions help AI visibility? We tested three UK categories
27. Coupons.com (16 September 2026), press release via Yahoo Finance: Coupons.com study reveals AI chatbots and discount codes
28. Pew Research Center (22 July 2025): Google users are less likely to click on links when an AI summary appears in the results
29. Google (20 May 2026): Shopping updates from Google Marketing Live
30. Press Gazette (as updated 2 November 2026): News publisher AI deals and lawsuits
31. European Central Bank (2 November 2026): Euro foreign exchange reference rates
32. WebKit: Tracking Prevention in WebKit
33. Stevens & Bolton LLP (6 May 2026): The Data (Use and Access) Act 2025: ICO guidance on cookies and the use of storage

Note: all sources were retrieved on 3 November 2026. Conversions use the ECB reference rate of 2 November 2026 (€1 = US$1.1216 = £0.8358; derived US$1 = £0.745), rounded. Dollar figures are US figures.