Our 2027 bets, logged like incidents

Four falsifiable calls on where agentic commerce and catalog operations land by the end of 2027. Dated, sourced, due for grading.

Prediction is the safest genre in tech. You publish in January, collect the applause, and nobody ever comes back to check.

We run a log. Log entries carry dates, and dates have consequences. So here are four bets on 2027, written down where you can grade us, each with its failure condition attached. If we miss, the log will say so.

The board as it stands

Two things happened this year while everyone was still arguing about chatbots.

In January, Shopify and Google published the Universal Commerce Protocol, an open standard for AI-driven transactions. Merchants now sell through Agentic Storefronts inside ChatGPT, Microsoft Copilot, and Google's AI Mode, and Shopify Catalog broadcasts product data to those channels automatically.

The storefront is no longer the only front door.

Then Adobe measured who was walking through the new one. In March 2025, traffic referred by AI assistants converted 38% worse than traffic that arrived the old ways. By March 2026 it converted 42% better. An eighty-point swing inside a year.

One disclosure before the wagers. What Shopify has shipped is fact, and cited below. What happens next is our read, nobody's roadmap.

Bet one: the premium outlives the novelty

The loud number first: AI-referred traffic to US retail sites grew 393% year over year in Q1 2026. Loud numbers fade. That is what growth rates off a small base do, and anyone selling you a strategy on the growth rate alone is selling momentum, not mechanics.

Our bet sits on the quiet number instead. An agent arrives pre-qualified: it already read the specs, compared the alternatives, and checked the return policy before the click ever happened. By May, the conversion premium was still widening. We bet it survives 2027 even as the growth rate cools.

Fail condition: AI-referred conversion drops back below human traffic by December 2027.

Bet two: your product page gets a report card

Adobe began scoring retail sites for AI visibility this year, and product detail pages came in as the worst page type on the site, at 66%. The pages that answer an agent's boring questions, returns and FAQ, scored above 80. Merchants spent a decade polishing the page humans skim and neglected the one machines actually read.

We bet AI visibility does to product data what Core Web Vitals did to page speed: turns an invisible engineering concern into a scored, budgeted line item. By the end of 2027 we expect PDP machine-readability to be a standard audit deliverable across the Shopify ecosystem, ours and everyone else's.

Fail condition: nobody is scoring it by then (we would be wrong, and a little relieved).

Bet three: the one-trick apps go free or go home

Sidekick is included with every Shopify plan now, and writing product descriptions sits on its official list of chores. That is a platform doing what platforms do: absorbing the generic layer, one feature at a time.

So by the end of 2027, we bet the standalone AI description generator app is either free or gone. What survives above the platform is the accountable layer, systems that run one specific catalog with verification and rollback, at a scale where "just review each output" stops being a sentence a human can finish. It took us 642,000+ SKUs to hold this opinion firmly.

The honest half of this bet: if Sidekick can hold your whole catalog in one conversation, use Sidekick. Don't hire us.

Fail condition: paid single-feature description apps are still a healthy category.

Bet four: being listed stops being the moat

Shopify's new Agentic plan lets merchants who aren't on Shopify at all push their products into Shopify Catalog. Read that as a distribution story: the catalog is becoming a commons.

When every product sits in the same feed, presence stops differentiating and data quality starts. Shopify's own merchant guidance already says it plainly: structured fields and factual specifications, not marketing adjectives. An agent doesn't feel a vibe. It parses fields.

We bet the crowded categories of 2027 get decided in the attribute tables, and the losers never learn why.

Fail condition: stores with thin, unstructured product data still winning head-to-head in agent-surfaced results.

How we'll grade this

The scoreboard, pinned for December 2027:

• Bet one dies if the conversion premium disappears.
• Bet two dies if nobody scores PDP machine-readability.
• Bet three dies if one-trick apps keep charging.
• Bet four dies if presence alone still wins.

We'll publish the grades in this same log, misses included. Especially the misses; those are the entries worth reading.

Predictions are cheap. Dated ones aren't.

If you'd rather not wait until 2027 to learn whether agents can read your catalog, OKART's Infrastructure Stress-Test checks it now, read-only, in a few days.

Sources

1. The agentic commerce platform: Shopify connects any merchant to every AI conversation: Universal Commerce Protocol co-developed with Google, Agentic Storefronts, the Agentic plan, and Copilot Checkout, announced January 2026.
2. Agentic Commerce: Benefits & How To Get Started: Shopify Catalog broadcasting product data across AI channels, and Shopify's merchant guidance on structured fields over marketing copy.
3. Shopify Sidekick: official capability list, including product description writing, and inclusion with every Shopify plan.
4. AI traffic to US retailers rose 393% in Q1, and it's boosting their revenue too: Adobe data, Q1 2026 traffic growth, and the March 2025 to March 2026 conversion reversal.
5. Adobe: AI-referred traffic to retail sites doubles in a year: May 2026 data showing the AI conversion premium still widening.
6. Adobe releases Q1 AI traffic report and first-ever retailer AI visibility data: Adobe's AI visibility scorecard, product detail pages scoring 66% as the worst page type, text-heavy pages (returns, contact, FAQ) above 80%.