Planning is the fastest-growing thing people do in Google AI Mode. In Google's 2026 report "How people are using AI Mode in the U.S.", planning-style "Do" queries grew about 80% faster than AI Mode queries overall: the fastest-growing behaviour Google named, ahead of "which" decision queries (+40%) and brainstorming (+30%). For any brand whose customers plan trips, workouts, projects, meals, or events, that makes the "Do" intent a direct channel. This playbook turns it into a content plan.
It draws on Google's documented usage data and applies the same method as the rest of the corpus. It is the "Do" companion to the Decide/shopping and Learn playbooks. All figures are Google's own platform data, dated to mid-2026 and not independently audited, so read the direction (planning is a large, fast-growing use of AI search) as firmer than any single number.
Why is the "Do" intent worth optimising for?
Because it is where AI Mode is growing fastest and where the questions are unusually answerable. Google groups AI Mode usage into five intents. Explore, Decide, Learn, Create, and Do: and reports how each grew against AI Mode queries overall:
| Behaviour | Intent | Growth vs AI Mode queries overall |
|---|---|---|
| Planning | Do | +80% faster |
| "Which" decision queries | Decide | +40% faster |
| Brainstorming | Explore | +30% faster |
| Image creation | Create | More than tripled since the start of the year |
Source: Google, AI Mode U.S. Insights report (mid-2026). Google describes "Do" as logistics, schedules, to-do lists, and exercise plans, and has surfaced planning inside AI Mode's Canvas planning experience. Two behaviours make planning content a natural fit: AI Mode queries are about triple the length of a classic search, and follow-ups grew more than 40% per month: the shape of someone working a task through step by step.
A planner rarely asks one question. They ask what the steps are, in what order, how long it takes, and what they need, so the content that wins the "Do" intent hands over a complete, ordered plan, not a paragraph of encouragement.
What does the planning question tree look like?
AI engines fan a single task into many sub-questions and assemble the answer from different sources. Cover the whole branch set, each as a self-contained chunk a model can lift:
| Branch | Example question | What to publish |
|---|---|---|
| The plan | "How do I plan a 3-day Kyoto trip?" | An answer-first, numbered plan |
| Order | "What order should I do this in?" | Steps in explicit sequence |
| Timing | "How long does each step take?" | Realistic, ranged durations |
| Prerequisites | "What do I need before I start?" | A what-you-need checklist |
| Logistics | "When and where do I do each part?" | Schedule, location, booking notes |
| Constraints | "What if I only have a weekend / small budget?" | Stated assumptions and variants |
Covering several branches well can get you cited multiple times in one answer. The query fan-out multiplier at work (the mechanics are in how query fan-out works). The defining feature of planning content is that it must be actionable: a model should be able to lift a step, a duration, or a checklist item and drop it straight into the plan it is building for the user.
How should you structure a planning page?
Apply the extraction rules that get any chunk selected, tuned for tasks:
- Answer-first, then detail. Open with the plan in brief. "A 3-day Kyoto trip breaks into arrival, temples, and day-trip", then expand each step. Give the model the shape before the specifics.
- Real steps as an ordered list. Sequences belong in numbered steps, not prose, and they map cleanly to HowTo structured data.
- Durations and quantities in tables. Time, cost, distance, and counts are facts. Put them where a model can lift them, not buried in narrative.
- State assumptions up front. "This assumes two days and a mid-range budget." Scope clarity is one of the strongest signals a chunk gets selected.
- One task per chunk, so a section answers its question without the rest of the page.
This is the same discipline as getting cited by AI and optimising for AI Overviews; planning content simply leads with order, timing, and checklists.
Why does specificity win planning queries especially?
Because a plan is only useful if its details are real. Vague planning content ("allow plenty of time", "pack accordingly") gives a model nothing to lift; specific, corroborated detail does. Make the actionable facts concrete and verifiable:
- Name real durations, distances, and costs: "the temple circuit takes about 4 hours on foot". Dated where they drift.
- Give checklists, not adjectives: an explicit what-to-pack or what-to-prepare list beats "be prepared."
- Show the source of any hard number (opening hours, prices, travel times) so it is safe to quote, and keep it current.
Over-claiming or inventing specifics is worse than omitting them: a plan that sends someone wrong is the fastest way to be described as unreliable. The opposite of the goal.
How do you keep planning content cited over time?
Planning facts drift. Prices, hours, schedules, tooling, and live AI retrieval favours recently-reviewed pages, so put planning pages on a refresh cadence and make updates substantive (corrected times, new steps), never a bumped date. The mechanics are in the content-freshness citation cliff. Evergreen task structures ("how to plan a marathon build-up") decay slowly; time-bound plans ("2026 holiday shopping timeline") decay fast and need attention.
Where should you start?
A focused first pass:
- Pick one task your customers genuinely plan around your product or category.
- Map its planning question tree (the table above) and write a self-contained chunk for each branch. Plan, order, timing, prerequisites, logistics, constraints.
- Lead each with the answer, put durations and checklists in lists and tables, and add HowTo structured data.
- State your assumptions and keep the hard numbers current and sourced.
- Track whether AI engines cite and recommend you for the task. Across engines, over time.
That last step closes the loop. Knowing whether AI Mode and the other engines actually surface, cite, and recommend your brand for the tasks your customers plan. Tracked daily rather than spot-checked. Is exactly what Buffy Intel measures: presence, citations, and share of voice across every major engine.