What I'd change in Storyscaping if I rewrote it in 2027
I reread my own book this summer. Not out of vanity — out of suspicion. Darren McColl and I published Storyscaping in 2014, back when "mobile-first" was a bold strategic posture and nobody had ever typed a question into a box and gotten a paragraph back instead of ten blue links. Twelve years is long enough for a book to become either a foundation or an artifact. I wanted to know which one I'd written.
Most of it holds. One assumption is broken. It isn't the one people expect.
What holds
The spine survived. Brands that only tell stories lose to brands that build systems people can live inside. The Organizing Idea — one proposition that aligns what you say with what you make — still separates coherent companies from expensive noise. Matching luggage is still lazy. The hero is still the customer, never the brand; every year I meet another CEO who wants to be Luke Skywalker, and the job is still to be the ruby slippers.
And the first kiss chapter holds hardest of all. You can read a beautiful story about someone's first kiss. You can watch a film about it with perfect lighting and a swelling score. Neither one touches the memory of your own. We remember what happened to us. That's why the strongest thing a brand can do is not tell you its story but become part of yours — through a shared experience. Story making, not storytelling.
Nothing about AI weakens that. If anything, it's the only claim in the book that gets more valuable as the cost of telling goes to zero.
Where the book breaks
Here's the assumption I'd tear out: that when the shared experience happens, the brand and the human are alone in the room.
In 2014 that was safe. Every model, every diagram, every touchpoint inventory in that book assumes a person moving through an experience space with their own eyes and their own hands. Someone walks in. Something happens. A memory forms. We never once contemplated that the customer would send a proxy.
They do now. Adobe's research puts 73% of consumers using AI as their primary source of product research. Shopify saw orders originating from AI-driven search grow roughly fifteenfold between January 2025 and January 2026. Adobe clocked AI-referred traffic to US retail sites up 805% year over year on Black Friday 2025. McKinsey models five trillion dollars of agentic commerce volume by 2030; Bain puts 15–25% of all e-commerce through agent channels in the same window. Meanwhile Pew's behavioral study of nearly 69,000 real Google searches found that when an AI summary appears, people click through to an actual website 8% of the time instead of 15%.
Read those two facts next to each other, because that's the whole problem in miniature. Influence is climbing. Presence is collapsing.
A machine now stands between the mentor and the hero, and it is very good at its job. It reads the reviews. It compares the specs. It fills the cart. Every one of those was a moment where a brand could have been present when memory formed — and a mediated experience is not a shared experience. You cannot have a first kiss by proxy.
That's the chapter I owe you. The book was written for a world where the story system's only reader was human.
Nobody delegates something they enjoy
Before anyone panics: the agent doesn't absorb everything. It absorbs what nobody loved. People hand off the parts of the experience that were never the experience — the twelve-tab comparison, the reorder, the form, the hold music, the specs. They do not hand off the parts that make them feel something. Nobody asks an agent to taste the wine, walk the store, or feel the door of the car close. Nobody outsources the moment.
So use it as a diagnostic. Map every interaction you have with a customer and sort it into two columns: what they would gladly delegate, and what they'd never let anyone do for them. The first column is cost. Automate it, and thank the machine. The second column is your moat, and it's where the entire budget should go.
Most companies have it backwards. They've spent a decade making the first column beautiful.
The pillar I'd add
The book has four pillars to an Organizing Idea. It needs a fifth: designing for a reader who isn't human.
Not as an SEO chapter. As a craft chapter. Your brand now has two audiences with completely different appetites. The human wants meaning, texture, and a reason to feel something. The machine wants structure, evidence, availability, and unambiguous facts. A story system that only satisfies the first never reaches the second — and in a growing share of categories, the second decides whether the first ever hears from you.
The strange part is that we described the architecture in 2014 and didn't know what we were looking at. We used the King James Bible as our example of a story system: 1,500 pages nobody reads front to back, structured instead into books, chapters, and verses so any fragment can be pulled out, quoted, and still make sense on its own. That's not a book. That's a retrieval system. Machine-legible narrative, four hundred years early.
That's the standard now. Can a fragment of your brand be pulled out of context, quoted by something that has never met you, and still be true, useful, and unmistakably yours?
For most companies the answer is no, and the reason is boring: the product data is a mess. Mirakl's work puts 42% of purchase abandonment on insufficient product information. In agent-mediated commerce, your product data is creative work. It's the copy. Treat it like a database problem and you'll lose to someone who treats it like a brand problem.
The Organizing Idea stops being a creative tool
This is the change I care about most.
In 2014 the Organizing Idea was an alignment device. You wrote it, everybody nodded, and it kept the work from drifting. Useful, but soft — a compass in a room full of humans who could use judgment.
In 2027 a machine writes your product descriptions, answers your customers, drafts your emails, and improvises brand behavior in situations you never briefed. It will do that thousands of times an hour and it has no taste. Which means the Organizing Idea has quietly changed jobs. It isn't a compass anymore. It's a constitution.
When generating is free, the scarce and valuable thing is the thing that says no.
So I'd rewrite that chapter with a hard output: don't just write what the brand stands for. Write what the brand will never say, never claim, never offer, never discount, never joke about. Ten refusals, in plain language, specific enough that a system can enforce them. Then hand that document to whoever owns your model layer — because right now, in most companies, nobody has handed them anything, and the brand is being improvised by a stranger.
Two more moves
Chapter six goes to the front. "Walk the Walk" — the argument that brands must behave authentically, not just speak authentically — was the moral center of the book buried in the middle. It's now the most operationally urgent chapter, because every answer a machine gives about you is brand behavior. Not messaging. Behavior. Which means the gap between what you claim and what you do is no longer discovered by a journalist in six months. It's surfaced in a paragraph, in three seconds, to someone who asked a neutral question. And I'd tear up the measurement chapter. We spent pages moving readers from media mix modeling to marketing mix optimization, and I stand by the direction. But when an agent mediates the purchase, your behavioral data stream doesn't start at the first touch — it starts around add-to-cart. Everything upstream happened somewhere you can't see, in a conversation you weren't invited to. The whole edifice of attribution I was arguing to improve is standing on a floor that's being removed. That's not a chapter to update. That's a chapter to rewrite from a blank page, and honestly, I don't yet know the answer. Neither does anyone selling you one.
Monday morning
1. Ask the four assistants your customers actually use the ten questions that precede a purchase in your category. Read the answers as brand behavior, not as a ranking report. Most executives have never done this once. 2. Write the ten refusals. Give them to whoever owns your model layer this week. 3. Sort every customer interaction into gladly delegated and never delegated. Move budget toward the second column. 4. Move product data out of IT's backlog and into the marketing conversation, with a named owner. 5. Instrument the agent handoff — the moment a machine acts for a customer — before you try to fix attribution.
What I wouldn't change
The title.
Storyscaping was never about stories. It was about building worlds people can enter and come out of with something that happened to them. That's still the work. The customer is still the hero.
They just hired staff. And a mentor whose gift can't be understood by the hero's staff never gets into the story at all.
