Agentic commerce arrived carrying a trust problem, and the problem is hiding inside a very encouraging number.
Ask a shopper whether they would let an AI agent buy something for them and the answer is a startling yes. Accenture’s 2026 Consumer Pulse Research, which surveyed 25,590 consumers across 16 countries, found that 74% would trust a personal AI agent more than their best friend to make a purchase on their behalf.
Then watch what happens as the question gets closer to the money.
Ask whether the agent can pick the item inside a budget and brand list, with the human still approving before payment, and 74% drops to 32%. Ask whether the agent can decide at the payment step itself, and it falls to 12%. Ask whether it can initiate and complete the purchase alone, and 9% say yes.
Same shopper. Same agent. Same afternoon. The trust falls off a cliff somewhere between “choose this for me” and “pay for this for me,” and that cliff is the most important design object in agentic commerce right now.
I wrote about the force underneath this in my hub post on gamification and AI: every time AI absorbs a piece of execution, it removes a motivation loop a human used to run, and somebody has to design the replacement on purpose. Commerce is the arena where that removal is happening fastest and where the old playbook is least prepared for it.
Because here is the thing almost every retail team gets backwards. They think agentic commerce means the human leaves the funnel. What the numbers say is that the human moves to a different seat in it, earlier, and stays there with a hand near the wheel.
Speed Run Notes
- Accenture’s 2026 Consumer Pulse (25,590 consumers, 16 countries) found 74% would trust an AI agent over their best friend to buy for them, 32% would let it decide inside set limits with human approval before payment, 12% at the payment step, and 9% fully autonomous. The gradient itself is the finding.
- Trust collapses where loss becomes irreversible. That is Core Drive 8 (Loss & Avoidance) doing exactly what Prospect Theory predicts, and it is why “one-click autonomous checkout” keeps failing consumer research.
- Adobe Analytics, measuring more than a trillion visits to U.S. retail sites, found AI-referred traffic grew 138% year over year in May 2026, converting 54% higher and producing 53% more revenue per visit than non-AI traffic. A year earlier the same channel converted at roughly half the rate.
- The same Adobe work scored how much of a retail page a machine can read, and the best-performing category came in at 63%. Over a third of the content on the strongest sites is invisible to the entity now doing the comparison.
- The payment protocols already encode the psychology. Google’s AP2 signs Intent, Cart, and Payment Mandates separately, which is a pre-commitment artifact and therefore Core Drive 4 (Ownership & Possession) rendered in cryptography.
- Shoppers keep control over anything tied to identity or enjoyment, and delegate anything repetitive. The mandate is the new design surface, and it should be built around the drives that survive delegation.
Table of Contents
- About Yu-kai Chou
- The 74% That Collapses to 12%
- Your Storefront Has a New Reader, and It Cannot See a Third of You
- What the Payment Protocols Already Encode
- Every Loyalty Mechanic You Own Was Built for Eyes
- Where the Human Still Buys
- The Design Surface Moves Upstream
- The Idealised Self Problem
- Four Moves for Agent-Era Motivation Design
- A 30-Day Audit
- FAQs
- What’s Next
- Related Reading
About Yu-kai Chou
Yu-kai Chou is a Human-Systems Architect & Behavioral Designer whose work has impacted over 1.5 Billion Users worldwide through work with LEGO, Microsoft, Porsche, Coca-Cola, Salesforce, and MrBeast.
He has taught and lectured at Harvard, Stanford, Yale, Tesla, Google, BCG, and IDEO.
His books have been referenced by Harvard, Stanford, MIT, Forbes, Wired, US Department of Energy, NIST, NSF, NCBI, and US Department of Education, among many others. Check out Books by Chou here.
The 74% That Collapses to 12%
Start with the shape of the Accenture data, because the shape carries more information than any single percentage.
Delegation in agentic commerce is not one decision. It is a gradient, and consumers sort themselves along it with remarkable consistency.
At the top, 74% will hand over routine work: negotiating a deal, chasing a refund, renewing a subscription, reordering what they always order. Accenture also found 61% want an agent that can shop across several grocery retailers for them, and 71% want one that can plan and book an entire trip across airlines, hotels, and activities.
Those are big appetites. They describe an enormous amount of labor that shoppers are happy to stop doing.
Then the gradient bends. 32% will let the agent choose the item, inside a budget and a brand list, with a human review before money moves. 12% will let the agent decide at the payment step. 9% will let it run the whole thing.
What separates the top of that gradient from the bottom is not complexity. Planning a multi-city trip is vastly more complex than buying a sweater, and shoppers are more willing to delegate the trip.
What separates them is reversibility.
Everything in the 74% band can be undone. A bad negotiation can be re-negotiated. A wrong reorder gets returned. A subscription gets cancelled. The human keeps a hand on the undo button, and the cost of the agent being wrong stays small and recoverable.
The payment step is where undo gets expensive. Money has left. A return is now a process with friction, timelines, and the possibility of eating the loss. The downside stops being an inconvenience and becomes an actual loss.
Behavioral science has a precise account of why that specific threshold produces a cliff rather than a slope. Daniel Kahneman and Amos Tversky’s Prospect Theory established that losses register roughly twice as heavily as equivalent gains. A shopper weighing “my agent might buy the wrong thing” against “my agent saves me twenty minutes” is not running a symmetrical calculation. The possible loss is doing double duty.
In the Octalysis Framework, which I have been developing since 2003 to map why humans do anything at all, that is Core Drive 8: Loss & Avoidance. It is the drive behind every “only 2 left in stock” banner and every abandoned-cart email, and in agentic commerce it flips from being a tool retailers use to being the force working against them.
Read that sentence again if you sell things for a living, because it is a genuine reversal of the last fifteen years of e-commerce practice. Loss aversion used to be your accelerant. In the delegation decision, it is the brake.
Look at what Accenture found consumers want before they will loosen their grip: data safeguards, configurable permissions, instant override, clear recourse, platform reputation, and perceived neutrality.
Every item on that list is a CD8 mitigation. None of them is a reward. Shoppers are not asking to be motivated into delegating more. They are asking to be protected while they do it.
Which means the first job of agent-era commerce design is subtraction: find the perceived losses in the delegation moment and remove them. Points and badges pointed at that moment will do nothing, because you cannot bribe someone out of loss aversion. You can only lower the stakes.
Your Storefront Has a New Reader, and It Cannot See a Third of You
While the delegation debate plays out, something more concrete is already happening to retail traffic.
Adobe Analytics, working from more than a trillion visits to U.S. retail sites, found that AI-referred traffic grew 138% year over year in May 2026. It more than doubled in a year.
The quality of that traffic is the part that should reorganize a roadmap. Those visitors converted 54% higher than non-AI traffic and generated 53% more revenue per visit. They engaged 15% more, spent 53% longer on site, and viewed 23% more pages.
Now hold that next to the same channel a year earlier, when AI-referred visitors converted at roughly half the rate of everyone else. Shopify has reported a similar pattern from its own side, with AI-referred conversion running near 50% above organic search and average order values 14% higher.
A channel went from your worst-converting to your best-converting inside twelve months. Whatever is upstream of that traffic got dramatically better at matching intent to product.
Here is the uncomfortable half. Adobe also measured how much of a retail page a machine can actually read, and scored categories on it. Cosmetics led at 63% readable. Electronics came in at 56%. Sporting goods and apparel sat at 51% each.
Sit with the leader for a second. The best category in the study is 63% legible to the entity doing the comparison. More than a third of the content on the strongest sites is not readable by the reader that increasingly decides whether a product enters the consideration set at all.
For twenty years, conversion optimization has been an exercise in designing for human perception. Hero imagery that creates desire. Social proof that borrows trust. Urgency copy that compresses deliberation. Layout that guides an eye down a page.
An agent has no eye to guide. It reads structure: attributes, specifications, prices, availability, policies, return windows, fulfilment records, verified claims. Accenture’s report lands in the same place, noting that agents compare brands using structured attributes and verified claims, weighing price-to-value ratios and fulfilment history.
So a product page can be beautiful, persuasive, and highly converting for humans while being close to mute for the agent. The photograph that sells the sweater carries no machine-readable information about fibre content, and the return policy that would win the comparison is a paragraph of prose in a collapsed accordion three clicks away.
I have written before about designing documents so the important structure survives contact with a machine reader, in The Minimap Method. The retail version of that problem is now measurable, and most sites are failing it by more than a third.
This is the least glamorous work in agentic commerce and probably the highest-return. Structured product data is not a motivation problem at all. It is a legibility problem, and it gates everything downstream, because a product an agent cannot parse never reaches the human whose motivation you were planning to design for.
What the Payment Protocols Already Encode
The infrastructure for agent purchasing arrived faster than most retail strategy decks noticed, and the way it is built tells you what its architects believe about trust.
In September 2025, Stripe and OpenAI released the Agentic Commerce Protocol as an open standard under Apache 2.0, alongside Instant Checkout in ChatGPT. Etsy sellers came first, with Shopify merchants following.
Google announced its Agent Payments Protocol the same month, with more than sixty launch partners including Mastercard, PayPal, American Express, and Coinbase. The card networks moved on their own tracks too: Mastercard Agent Pay and Visa’s Trusted Agent Protocol both extend card-rail tokenization to verified agents, with Visa issuing an agent identity and a separate consumer consent record.
Now look at how AP2 structures a transaction, because this is the part that should interest anyone who thinks about motivation.
It does not use one signed approval. It uses three separate mandates: an Intent Mandate, a Cart Mandate, and a Payment Mandate, each cryptographically signed, built on W3C Verifiable Credentials.
The engineers who designed that were solving for auditability and dispute resolution. What they built, without using the vocabulary, is a pre-commitment architecture.
An Intent Mandate is a human declaring in advance what they are willing to buy and under what conditions. That is a claim staked before the outcome exists.
In Octalysis terms, that specific psychology has a home, and it is easy to misfile. The open gap between “I have committed” and “it has been completed” feels like scarcity, so people reach for Core Drive 6. It is Core Drive 4: Ownership & Possession. Robert Cialdini’s consistency principle sits here: once a person has declared a position, they work to stay consistent with it, because the declaration is now theirs to defend.
The mandate is the shopper’s. They authored its terms. They own the constraints inside it.
That distinction has a practical consequence, and it is worth stating plainly for anyone building this. Core Drive 6 (Scarcity & Impatience) only enters a pre-commitment flow when the commitment itself is rationed: limited slots, a closing window, a scarce token. Absent that rationing, urgency mechanics bolted onto a mandate flow are decoration on a CD4 structure, and they will read as pressure at precisely the moment the shopper is most alert to being pressured.
The strategic read is this. Every one of these protocols independently converged on keeping a signed human intent at the base of the transaction. Nobody built a system where the agent simply decides and pays, even though that would be simpler to engineer.
They built consent artifacts because they understood, correctly, that the trust cliff is real and that the only way across it is a structure where the human’s authority is legible, provable, and revocable.
Every Loyalty Mechanic You Own Was Built for Eyes
Take an honest inventory of the engagement machinery in a typical retail stack, and ask which pieces still function when the buyer is software.
Points balances. Tier badges. Streaks. Countdown timers. “17 people are viewing this.” Free-shipping thresholds that nudge a basket upward. Spin-to-win popups. Abandoned cart sequences.
I have analyzed a lot of marketing gamification campaigns over the years, and the good ones share a mechanism: they create a feeling in a person who is present, looking, and deciding in the moment.
An agent is none of those things. It does not feel the pull of a nearly-complete progress bar. It does not experience the sting of a timer. It reads a price, a delivery date, a return window, and a set of attributes, and it optimizes against the mandate it was given.
Scarcity messaging is the clearest casualty. A countdown that compresses human deliberation is, to an agent, one more field. Worse, an agent evaluating price-to-value across sellers may read aggressive urgency signalling as noise, or as a reason to check whether the same item sits cheaper somewhere calmer.
FOMO requires a self that can miss out. Software has no such self.
The social layer breaks differently and more interestingly. Core Drive 5: Social Influence & Relatedness has been the engine behind reviews, ratings, influencer proof, and “customers also bought.” Those mechanics were designed to transfer trust between humans through a feeling of consensus.
An agent uses the same data with none of the feeling. Ratings become a numeric input. Review text becomes a corpus to extract claims from. The warmth that made social proof persuasive does not survive the translation, though the information inside it becomes more influential, because the agent actually reads all of it rather than glancing at the first three.
Then there is the loyalty question, which is where this gets expensive. Accenture found that 56% of consumers would tell their agent which brands to consider, which sounds like brand power holding firm.
The next number is the one to bring to a board meeting. Among consumers Accenture classed as behaviorally loyal, 37% said they would allow an agent to switch brands if it found a better fit on grounds like price, availability, or service performance.
More than a third of your loyal customers will pre-authorize their agent to leave you for a better fit. Loyalty built on habit, convenience, and default position is exactly what an agent is built to dissolve, because removing search friction was the entire point of it.
What survives an agent’s optimization is loyalty a human would defend on purpose: brands tied to identity, taste, values, or a relationship that feels personal. Everything else was a switching cost, and agents are switching-cost solvents.
Where the Human Still Buys
The Accenture data does something better than describe a general reluctance. It shows exactly which purchases people refuse to hand over, and the pattern is clean enough to design against.
Recurring services rank highest for delegation. Lifestyle and travel purchases drop sharply as autonomy increases. Consumers hold on to choices linked to identity or personal enjoyment.
The report’s own illustration is perfect: a shopper will happily delegate grocery restocking while still wanting to choose the hotel room, the clothing item, the experience.
Nobody’s sense of self is invested in which brand of paper towels arrives. Choosing a hotel is partly about anticipating who you will be on that trip.
Map that against the Core Drives and the logic is immediate.
Purchases people delegate are ones where the buying carries no motivational payload. There is no accomplishment in reordering detergent, no self-expression, no story. Automating it removes friction and costs the shopper nothing they valued.
Purchases people keep are the ones where the choosing is part of the reward. Browsing hotels is anticipatory pleasure. Picking clothing is identity work. Selecting a gift is an act of relationship, which is CD5 running at full strength, and no shopper wants a machine to have authored the thought behind a present.
Notice what that means for Core Drive 2: Development & Accomplishment in a shopping context. There is a real, if modest, accomplishment loop in finding the good thing: the researched purchase, the deal hunted down, the discovery you get to tell people about. That loop is small, and for a meaningful set of purchases it is the actual product.
Hand it to an agent and the shopper gets the item without the satisfaction of having found it. For detergent, excellent. For the perfect jacket, you have automated away the part they enjoyed.
Which produces a segmentation question far more useful than the demographic ones most retailers are still asking. For every category you sell: is the choosing a chore or a pleasure?
Chore categories should be engineered for delegation. Make your data pristine, your policies machine-legible, your reorder flows agent-native. Compete on being the easiest correct answer.
Pleasure categories should be engineered for the human to stay in the seat, with the agent doing the narrowing rather than the deciding. Compete on making the final choice feel like theirs.
Getting this backwards is expensive in both directions. Automating a pleasure category strips out the experience people were paying for. Insisting on human attention in a chore category means the agent routes around you to a competitor who made it easy.
The Design Surface Moves Upstream
Put the pieces together and the conclusion is fairly stark for anyone whose job title includes the word “conversion.”
The motivational surface of commerce is moving from the storefront to the mandate. From the moment of purchase to the moment of delegation.
When a shopper configures an agent, sets its permissions, defines its budget, names its acceptable brands, and decides how much rope to give it, that configuration moment is now where the human is most present, most deliberate, and most emotionally engaged in the entire transaction.
It is also, in most current products, an ugly settings screen.
Every drive that used to be designed into a product page has an analogue in that moment, and almost nobody is building them yet.
CD4, Ownership. The mandate is the shopper’s most valuable possession in an agentic flow. It encodes their preferences, standards, and history. Make it feel owned: visible, named, versioned, editable, exportable. A shopper who has spent time shaping a mandate has an asset they will defend, which is the closest thing to durable loyalty available in an agent-mediated market. An agent configuration that lives inside your ecosystem and improves with use is a switching cost that a switching-cost solvent cannot dissolve, because the thing being protected belongs to the customer rather than to you.
CD2, Accomplishment. Delegation should produce visible evidence that the shopper’s judgment worked. Not gamified points for using the agent. Real feedback: what your constraints saved you, what your agent rejected and why, where your standards held. This is the accomplishment loop for the delegating human, and it rewards the skill of setting good instructions.
CD8, Loss & Avoidance. This is the drive to design away rather than design in. Instant override, clearly reversible actions, generous and legible return terms, a visible spending ceiling, an audit trail the shopper can read. Every one of these lowers the perceived cost of being wrong, which is the actual gate on the 12%.
CD5, Relatedness. Household and group agents are coming, and Accenture’s grocery finding points straight at them. A mandate that serves a family involves other people’s preferences, and letting a shopper encode care for someone else is a strong motivational thread that pure optimization misses entirely.
On the White Hat and Black Hat axis, everything in that list sits firmly White Hat: ownership, accomplishment, and relatedness, with CD8 handled by removing threat rather than manufacturing it. That is deliberate, and it is a departure from how most conversion machinery has been built.
On the Left Brain and Right Brain axis, the mandate is dominated by Left Brain extrinsic structure. Goals, constraints, ownership, and measurable outcomes. The Right Brain drives that made shopping fun, curiosity and creative expression, stay with the human in the pleasure categories and largely leave the delegated ones.
Black Hat mechanics deserve a specific warning here. The delegation moment is the point of maximum shopper vigilance, which makes it the worst possible place for pressure tactics. Fake urgency in a permissions flow does not accelerate a decision. It tells the shopper that you are the sort of company that applies pressure, in the exact instant they are deciding how much authority to give something that will act on their behalf without supervision.
I have argued for years that Black Hat drives produce fast action and drain people over time. Agentic commerce sharpens that further, because an agent will not feel your pressure and its human principal will resent it.
The Idealised Self Problem
One Accenture finding deserves its own section, because it points at the largest genuinely new opportunity in this whole arena.
63% of consumers want agents that shop for their idealised self: helping them make healthier choices, stay within budget, upgrade more intentionally.
Read that again, because shoppers are asking for something the retail industry has spent decades not offering.
Consumers are volunteering for a system that helps them behave like the person they intend to be rather than the person their impulses produce at 11pm.
Structurally, this is a commitment device, and the behavioral literature on those is strong. A person who pre-commits while calm to constraints that bind them while tempted reliably outperforms a person relying on willpower in the moment.
An agent with a signed Intent Mandate is a remarkably good commitment device. The mandate was authored in a cool state. It executes in whatever state the shopper is in later.
This is CD4 doing its best work: the shopper owns the standard, and the agent defends it on their behalf.
The relevant research on aspiration is worth naming. Self-Determination Theory holds that motivation sustains when a behavior is autonomously chosen rather than imposed. An agent enforcing constraints the shopper wrote is autonomy-supportive by construction. The same agent enforcing constraints a retailer wrote is control, and it will be experienced as control.
That line is the whole ethical design question in this arena, and it will separate the products people keep from the ones they uninstall.
Which brings the cautionary number. Among active generative-AI users, 26% said AI had already led them to buy a more expensive item because it increased their confidence in the decision, and the same proportion said it had increased their basket size.
Confidence-driven upselling is not automatically bad. A shopper who buys the better thing because they finally understood the difference has been well served.
The failure mode is a system that learns confidence is monetizable. An agent optimized on merchant incentives, wearing the language of the shopper’s idealised self, would be the most effective persuasion technology ever built, and it would be trading on borrowed trust.
Accenture’s list of trust conditions included perceived neutrality for a reason. The moment shoppers suspect an agent’s recommendations are for sale, the 12% will fall rather than climb, and it will take years to rebuild.
This is the same tension I keep running into across every arena of this work, from AI and employee motivation to motivation design after GLP-1. AI can execute a motivational design at a scale no human team could reach. That capability is neutral about whose interests the design serves, and the answer to that question is a choice somebody makes on purpose.
Four Moves for Agent-Era Motivation Design
1. Make yourself legible before you make yourself persuasive.
If the best retail category is 63% machine-readable, structured data is where the return is largest. Product attributes, specifications, availability, return windows, and policy terms belong in structured, parseable form rather than in prose an agent has to guess at. Persuasion cannot operate on a product that never entered the comparison.
2. Sort your catalogue by chore or pleasure, then design differently for each.
Chore categories compete on being the easiest correct answer for an agent. Pleasure categories compete on keeping the human in the deciding seat with the agent narrowing rather than choosing. One taxonomy, two entirely different design programs, and the cost of confusing them is losing either the transaction or the experience.
3. Design the mandate as a product surface.
Whatever configuration screen a shopper uses to set an agent’s authority over your category is now your highest-intent human touchpoint. Treat it that way. Ownership signals, real feedback on how their constraints performed, override that works instantly, and an audit trail written for a person to read.
4. Replace scarcity with reversibility.
The mechanic that moves the delegation needle is not urgency. It is lowering the cost of being wrong. Generous returns stated in machine-readable terms, visible spending ceilings, staged autonomy that expands as trust is earned. Every removed loss moves a shopper further along the gradient, and unlike a countdown timer, it works on the agent’s evaluation and the human’s anxiety at the same time.
A 30-Day Audit
Week 1: measure your legibility. Take your twenty highest-revenue products and check what a machine can actually extract: attributes, price, availability, return terms, fulfilment data. Anything living only in an image, a PDF, or a collapsed accordion counts as invisible. Score the percentage and write it down, because that number is your ceiling in agent-mediated demand.
Week 2: split the catalogue. Sort every category into chore or pleasure using one test: is the choosing itself part of what the customer enjoys? Get the disagreements on the table, since the contested categories are usually where the money is.
Week 3: audit your mechanics against a non-human buyer. List every engagement mechanic you run and mark which ones require a human present and feeling something. Those are your agent-blind spend. Do not switch them off, because humans still buy in your pleasure categories. Stop counting them as coverage for the delegated ones.
Week 4: find the losses in your delegation moment. Walk the path a customer takes to authorize an agent for your category. Name every point where being wrong costs them something: unclear returns, no override, invisible spend, no audit trail. Fix the two worst. That is the CD8 subtraction work, and it moves more delegation than any reward you could attach.
Then repeat the legibility measurement in ninety days. It is the one number in this arena that responds quickly to effort, and it gates everything else.
FAQs
What is agentic commerce?
Agentic commerce is commerce where an AI agent acts on a shopper’s behalf, doing the comparison, negotiation, and in some cases the purchase itself, working from permissions and preferences the human set in advance. It is distinct from AI-assisted shopping, where a human is still making each decision with AI help. Open standards including the Agentic Commerce Protocol and Google’s Agent Payments Protocol now provide the transaction infrastructure.
Do loyalty points still work when an AI agent is buying?
Mostly no, in the delegated categories. Points, tiers, streaks, and badges create a feeling in a human who is present and deciding. An agent reads a points balance as one numeric input among many and weighs it against price and fulfilment. Loyalty that survives is loyalty a human would actively instruct their agent to honour, which means brand attachment tied to identity, taste, or values rather than accumulated switching costs.
Why do shoppers trust agents to choose but not to pay?
Because payment is where the mistake stops being reversible. Accenture found 74% would trust an agent to buy on their behalf in principle, 32% would let it decide within set limits before payment, and 12% at the payment step. Prospect Theory explains the drop: losses weigh about twice as much as equivalent gains, so the possibility of an unwanted purchase outweighs the time saved. In Octalysis terms that is Core Drive 8, Loss & Avoidance, and it is answered by reversibility rather than by rewards.
What should a retailer do first?
Measure how much of your top products’ information a machine can read. Adobe found the leading retail category sits at 63% machine-readable, so most sites have a large and fixable gap. Structured product data gates everything else, because an agent that cannot parse your product never surfaces it to the human whose motivation you were planning to design for.
Is designing for AI agents the same as SEO?
They overlap on machine-readable structure and diverge on purpose. Search optimization aims to win a click from a human who then evaluates the page. Agent optimization aims to win an evaluation from software that reads structured attributes, policies, and fulfilment records, and that may never render your page at all. A site can rank well and still be poorly readable by an agent, which is why Adobe measures the two separately.
Which Core Drives matter most in agentic commerce?
Core Drive 8 (Loss & Avoidance) gates delegation, and it is handled by removing perceived losses rather than adding rewards. Core Drive 4 (Ownership & Possession) governs the mandate, which is a pre-commitment artifact the shopper authors and defends. Core Drive 2 (Development & Accomplishment) applies to the delegating human, who should see evidence their instructions worked. Core Drive 5 (Social Influence & Relatedness) transfers to household and gift contexts. Core Drive 6 (Scarcity & Impatience) mostly stops functioning, since an agent does not feel urgency.
What’s Next
The five arenas in the gamification and AI hub all describe the same removal: AI takes over a piece of execution, a motivation loop goes quiet, and someone has to design the replacement deliberately.
Commerce is the arena where that shows up as revenue fastest, and where the industry’s existing toolkit is least suited to the change. Fifteen years of mechanics aimed at human perception are becoming irrelevant in exactly the categories shoppers are most eager to delegate.
The work ahead is a rebuild rather than an optimization. Make your catalogue legible to a machine. Decide which categories the human should keep. Design the delegation moment with the care that used to go into a product page.
If you want the full framework this analysis runs on, the 8 Core Drives, White Hat and Black Hat motivation, and the game techniques that implement them, it is all laid out in my books. Browse the books here.
And if you take one thing into next week: measure how much of your best product page a machine can actually read. The answer is usually worse than anyone on the team expects, and it is the cheapest problem on this list to fix.
Related Reading
- The Octalysis Framework for Gamification & Behavioral Design — the complete framework behind every Core Drive referenced above.
- What is Gamification? Definition, Examples & Framework — the starting point if the Core Drive vocabulary is new.
- How Core Drives Interact: Cocktail Effects and Combinations — why CD4 and CD8 together behave differently than either alone.
- Left Brain vs Right Brain Core Drives — the extrinsic and intrinsic split that decides which mechanics survive delegation.
- The Best Gamified Productivity Apps — how mandate-style commitment mechanics work in shipping consumer products.
- Negativity Bias: Why Bad Is Stronger Than Good — further reading on the asymmetry underneath the trust cliff.