The AI Purchase Channel does not reward the marketing tactics that worked in display, search, social, and shelf. It is a recommendation surface with two readers in one interaction — the human, who reads emotionally, and the AI, which reads literally — and the work that buys position in it is different in kind. This essay names the seven tactics most often reallocated to this channel that do nothing, why each fails mechanically, and what each line item should be replaced with instead.
Slingso, What the AI Purchase Channel does NOT reward
Earlier we wrote about why new brands win the AI Purchase Channel. Earlier this week we wrote the Field Guide and the measurement essays. Sitting with brand teams in the past two months, the question that keeps recurring is not “what should we add?” It is “what should we stop doing?” The list of things to stop is more painful than the list of things to start, because the things to stop are line items somebody on the team has spent years getting good at. They are also, in most cases, line items that ate the brand’s marketing budget for the prior decade.
The seven tactics below all work — or worked — in some channel. They do not work in this one. We have run audits, and the brands moving the most into the AI Purchase Channel are the brands that have audited their existing spend and pulled the line items that don’t transmit. The brands that haven’t pulled them are running two budgets in parallel: the one that pays for legacy tactics, and the one that funds the new work. Most brands do not have the margin to run both for long. Pick.
What does the AI Purchase Channel reward, briefly?
The channel rewards specificity (a brand that reads as the precise answer to a precise persona-scenario query), memory (citation in the model’s training and retrieval substrate), and being right (precise, verifiable owned content the AI can lift). We argued the full case in Wednesday’s essay on why new brands win this channel. The list below is the inverse — tactics that do not produce any of those three, in either the assisted mode (where a human is in the conversation) or the delegated mode (where a personal AI agent is acting on the shopper’s behalf).
Why does retargeting not transmit?
Retargeting is a display-ad tactic. It works by following a shopper across the open web after a brand-site visit, surfacing the brand’s ads on third-party properties until the shopper converts. The mechanism is impression frequency on an identified user — the shopper who already knows the brand sees the brand repeatedly until the brand wins consideration share.
The AI Purchase Channel has no equivalent surface, in either mode. In the assisted mode, the shopper is inside an assistant conversation — there is no third-party ad inventory in the conversation for a retargeting pixel to bid into. In the delegated mode, the shopper isn’t on a screen at all — a personal AI agent is doing the work on a server, and the agent doesn’t see ads. A retargeting campaign cannot follow either reader into a recommendation moment and influence the outcome, because the recommendation logic is the AI’s own retrieval over its training and retrieval substrate, not a third-party ad surface.
Retargeting still works in its native channel — display + open-web placements driving back to the brand site. It produces no signal in the AI Purchase Channel. Brands reallocating ten to thirty percent of their digital budget into retargeting because it produces visible CTRs in their dashboards are correctly measuring the wrong channel. The dashboard says the spend is performing. The channel that’s increasingly making the purchase decision sees none of it.
Replace with: spend reallocated to deepening the brand’s owned-content specificity for the three to five persona-scenario combinations that matter most. Persona-specific product page rewrites, schema enrichment, persona-specific third-party reviews and citations.
What about discount-led promotions?
Discount language — “save 20%,” “limited-time offer,” “code FLASH at checkout” — is the surface signature of conversion-rate optimisation in DTC. It is the primary driver of paid social ROAS for most consumer brands. It does not transmit into the AI Purchase Channel.
The mechanism: assistants rarely surface promotional language because the recommendation logic is conditioning on fit, not on price drop. A shopper asking “the best running shoe for a sub-3-hour marathoner” receives a recommendation based on which brand reads as the precise answer to that brief. The assistant may mention price in passing. It is extremely unlikely to surface “and Brand X is currently 20% off.” The promotional layer is below the recommendation layer.
There is one exception worth naming: when the shopper explicitly asks about price comparisons or budget constraints, the assistant will weigh price as one input among many. But it still ranks fit above promotion. A brand with a sharp persona-fit narrative at $52 will beat a generic brand at 20% off ($40) for the persona-conditioned recommendation. The shopper, having received the recommendation, may still shop the discount — but the discount didn’t earn the slot; the fit did.
Replace with: the discount spend continues to make sense in paid social and email. It does not make sense as a way to win the AI Purchase Channel. The substitution is the same: persona-specific owned-content investment, not promotional cadence.
Does domain authority still matter?
Domain authority in the search-era sense — aggregate site-level authority compounded across the brand’s total backlink graph — was a meaningful Google ranking factor for two decades. Brands invested in earning aggregate authority because that authority lifted every page on the domain. The mechanism was the model that Google’s PageRank algorithm and its successors used.
The AI Purchase Channel doesn’t use aggregate domain authority the same way. It uses per-claim authority on the specific claim that’s being made. A brand can have a high aggregate domain authority score and lose a specific recommendation to a brand with lower aggregate authority but stronger per-claim authority on the exact claim the AI is evaluating. A skincare brand with DA 78 can lose the “sensitive skin, 35-year-old, vitamin C serum under $40” recommendation to a brand with DA 42 because the second brand’s product page, schema, and third-party reviews say precisely those things and the first brand’s say something more generic.
Aggregate domain authority is still useful — it correlates with the brand being in the assistant’s retrieval set. It is not sufficient. The work that wins is the work at the claim level: which specific claim is anchored in which specific piece of owned and earned content, with what specific citation.
Replace with: invest backlink and authority work into specific per-claim citations rather than aggregate domain growth. A single verified citation in an authoritative source on a specific claim is worth more in this channel than ten backlinks to the home page.
What about influencer follower counts?
Influencer marketing in social channels rewards reach — follower counts and engagement rates that price out the cost per impression. The mechanism is exposure: the brand’s name flashes in front of N humans, and some non-trivial fraction acts.
The AI Purchase Channel doesn’t read follower counts. It reads the content of the influencer’s published reviews and the verified reach of their work. A creator with 50k followers who has written one detailed, specific review of a product, with structured claims about ingredients, performance, and target persona, can produce more lift in the AI Purchase Channel than a creator with 2M followers who posted a single sponsored image with generic praise.
This inverts the agency-priced influencer mathematics. Agencies have for a decade priced influencer placements by follower tier — top-tier creators are valued at far more than mid-tier creators because their reach is larger. In the AI Purchase Channel, the determinant of value is what was said, not how many people saw it. A mid-tier creator with deep, specific writing on a category is more valuable to a brand trying to win persona-specific recommendation than a mega-creator with broad, vague endorsement.
Replace with: influencer spend reallocated from reach-driven placement to depth-driven placement. A handful of specific, persona-anchored, mid-tier creator partnerships beats one mega-influencer post for AI-recommendation purposes. Brands should ask the creator to publish substantive owned content (a long-form post, a YouTube review, a Reddit AMA) — not just an Instagram story — because the AI retrieves what’s archived and searchable, not what evaporated.
Does generic SEO content still work?
Generic SEO content — high-volume, broadly-targeted blog posts written to chase keyword traffic across a wide category — was a survival tactic in the Google era. Brands published 200 posts about “skincare tips” because the cumulative traffic compounded.
The AI Purchase Channel ranks owned authority on specifics, not on volume of generic posts. A brand with 200 generic blog posts and one persona-specific deep guide will be retrieved on the deep guide and ignored on the 200. The 199 generic posts cost time, money, and authorship attention — and produced no lift in this channel.
There is a related trap: brands moving into the AI Purchase Channel sometimes treat the new channel the same way they treated SEO and start commissioning “AI-optimised” generic content at scale. That work doesn’t transmit either. The channel doesn’t reward volume; it rewards depth on specific persona-scenario combinations. Twenty deep persona guides beat two thousand generic SEO posts.
Replace with: retire the generic content factory. Reinvest the budget into three to five persona-anchored long-form guides per quarter, each addressing a specific persona-scenario combination with structured claims, schema, FAQ markup, and verifiable specifics.
What about hero photography and glossy storefront design?
Hero photography, art-directed product imagery, polished storefront design, and the visual elements that have made DTC brands beautiful to look at — these were the right investment in social channels and on-shelf shopping experiences where the shopper sees the brand as a visual gestalt.
The AI Purchase Channel does not see images the same way. Multimodal assistants read product photography for compositional information (the bottle is glass, the label says X) but they do not weigh the aesthetic polish of the photography in the recommendation. A brand with $200k of agency hero photography and a brand with founder-shot product images on a plain background are read identically by the recommendation logic — provided the descriptive content (product page copy, schema, alt text) is equivalently specific.
There is one place this matters: in the assisted mode, when the shopper has received the recommendation and clicks through to the brand’s site, the photography matters for the human conversion. So hero photography retains a role in the conversion funnel after the AI has done its work. In the delegated mode, the agent doesn’t linger on hero shots at all — it parses the page for structured claims and acts. Either way, photography does not earn the recommendation slot in the first place.
Replace with: keep enough product photography to support post-recommendation conversion in the assisted mode. Stop assuming additional photography investment beyond that point produces marginal channel position. Redirect the saved budget to owned-content depth — which is the only thing the AI reader sees, in either mode.
What about bulk-bought reviews?
Some brands have, over the years, invested in incentivised or volume-purchased review generation to push their aggregate review counts up in the third-party review ecosystem. The mechanism was social proof at scale — a higher review count, even with average ratings, suggested category authority.
The AI Purchase Channel weighs review substance and verified provenance over review volume. Twenty verified, detailed, persona-specific reviews from named verified buyers will produce more recommendation lift than two thousand unverified five-star ratings. The assistants are increasingly able to detect and discount review patterns that suggest manipulation — generic five-star reviews from accounts with no purchase history, sudden volume spikes, identical phrasing across reviews, and so on.
There is a subtler version of the same trap: brands that did not bulk-buy reviews but built their review base through paid acquisition channels often have a review corpus that reads as a population of “anyone who paid for a discount and submitted a rating.” That review corpus is also weakly retrieved by the AI, because the AI reads the reviews and notices they are not persona-specific. A brand’s review corpus is most useful when it reads as the verified voice of the brand’s actual core personas. Volume without persona-specificity is wallpaper.
Replace with: invest in review programmes that surface verified, persona-specific narrative from real buyers. Quality over quantity. Tools that let buyers tag their persona-scenario in the review submission are worth more than tools that just collect star ratings.
Two bonus tactics worth a brief mention
Press logos and “as seen in” trust theatre. “Featured in Vogue / Forbes / Bloomberg” press logos on the homepage were useful when the human shopper was deciding whether to trust the brand. The AI doesn’t weigh logos — it weighs whether the press coverage referenced contains specific claims about the brand for specific persona-scenario combinations. A Forbes article that says “Brand X is one of the year’s interesting beauty launches” is less useful to the recommendation engine than a smaller-outlet piece that says “Brand X’s vitamin C serum is the considered choice for thirty-something professionals with sensitive skin.” The AI retrieves specificity, not prestige.
Email list size as a moat. A large opt-in email list was a defensible asset for the prior decade — it was an owned distribution channel that didn’t depend on platform algorithms. It remains useful for that purpose. It does not, however, produce any signal in the AI Purchase Channel. The AI doesn’t know you have a million subscribers. A brand that’s spent five years optimising for email-list growth and treating that as its primary marketing investment has nothing the channel reads. The list still does its job in its native channel. The asset doesn’t translate.
The line-item audit a brand should run this quarter
Read your last full-quarter digital marketing spend. Identify the line items in each of the seven categories above. Sum them. That number is the budget that is contributing zero to the AI Purchase Channel position you need to be building. Some of it is doing real work in its native channel (retargeting still earns ROAS in display; discounts still drive conversion in paid social). Some of it is not even doing the work the dashboard claims.
Decide what fraction of that budget you can defensibly reallocate over the next ninety days. Reinvest in the work that does compound in the AI Purchase Channel: persona-specific owned-content depth, schema and structured data enrichment, mid-tier creator partnerships that produce substantive owned content, verified-buyer review programmes, per-claim third-party authority.
A brand that runs this audit honestly in Q2 2026 will, by Q4 2026, have a measurement baseline in the AI Purchase Channel and a defensible position on three to five persona-scenario combinations. A brand that doesn’t will spend Q3 wondering why the AI channel “isn’t producing.” We covered that conversation in Wednesday’s essay; the answer is the brand is measuring the wrong channel with the wrong tools, and reallocating from the right ones.
There are no points awarded for working harder in this channel. There are points awarded for working differently. Most brands’ biggest move in the next six months is not adding new line items. It is pulling the ones that don’t transmit and letting the saved budget do the work.
Frequently asked
What is the AI Purchase Channel? The AI Purchase Channel is the new buying channel where the cognitive work of buying — research, comparison, defensibility checks, choice — has moved out of the shopper’s head and into AI. Sometimes the shopper stays in the conversation with an AI assistant (the assisted mode); sometimes they hand the whole job to a personal AI agent — OpenClaw, Hermes, a custom build — and step back (the delegated mode). Either way, the decision happens inside AI. (Read the Field Guide.)
Why doesn’t retargeting work in the AI Purchase Channel? Retargeting is a display-ad tactic that engages impression frequency on identified users across third-party web inventory. The AI Purchase Channel has no equivalent surface in either mode — in the assisted mode the shopper is inside an assistant conversation with no third-party ad slot, and in the delegated mode the shopper isn’t on a screen at all. Retargeting still works in its native channel. It produces no signal in this one.
Are discounts useless in the AI Purchase Channel? Not useless, but downstream of the recommendation. AI assistants rank fit above promotion; the recommendation slot is won by the brand that reads as the precise answer to the shopper’s brief. A discounted but generic brand loses to a full-price specific brand for the persona-conditioned recommendation. Once the shopper has the recommendation, discount messaging still has a role in conversion — but it doesn’t earn the slot in the first place.
Does domain authority still matter for AI assistants? Per-claim authority on the specific claim the AI is evaluating matters more than aggregate domain authority in the search-era sense. A brand with DA 42 and stronger per-claim citations can beat a brand with DA 78 and weaker per-claim citations for the specific persona-scenario recommendation. The brand should invest in per-claim authority — specific citations in authoritative sources on specific claims — rather than aggregate backlink growth.
Should I stop working with mega-influencers? Not stop — re-evaluate the pricing model. Mega-influencer placement is priced by reach, which is what matters in social channels. The AI Purchase Channel reads what the influencer wrote about the brand and how specific that writing was, not how many people saw it. A mid-tier creator with deep persona-specific writing on a category produces more recommendation lift than a mega-creator with a vague sponsored post. The agency-priced reach mathematics inverts in this channel.
Is generic SEO content dead? Not dead in search; ineffective in the AI Purchase Channel. The channel rewards depth on specific persona-scenario combinations, not volume on generic category queries. Twenty deep persona guides outperform two thousand generic SEO posts for AI-recommendation purposes. Brands moving into the channel should retire generic content factories and reinvest in persona-anchored long-form work with structured claims.
Where does the saved budget go? Into the three things the channel actually rewards: persona-specific owned-content depth (product pages, FAQs, guides), schema and structured-data enrichment (so AI agents in the delegated mode can read literally), and per-claim third-party authority (specific verified citations on specific persona-scenario claims). (Read why new brands win this channel.)
The week’s essays read in order:
- Monday: The AI Purchase Channel: A Field Guide — channel definition + first-thirty-days.
- Wednesday: Why new brands win the AI Purchase Channel — specificity, memory, being right.
- Friday (this essay): What the channel doesn’t reward — line items to pull.
The measurement essays:
Slingso is the AI Purchase Channel Manager — the AI team member your brand onboards as the dedicated owner of the channel. Monitor → Analyse → Create → Approve → Measure, continuously.