You have eight weeks before peak season, one approver, and no new page scope in the final two weeks. I would rather enter peak with a handful of accurate buyer pages than a publishing queue full of promises operations cannot keep.
I used to tell clients to publish right up to the deadline. I was wrong. Last-minute publishing leaves discovery and indexing until the moment you need sales. What the deck calls “seasonal agility,” I call paying for pages that may not sell in time.
This plan stops expanding before demand peaks. The first six weeks earn you a small set of pages you can actually fulfill. The final two verify the facts, indexing, recommendations, and buying paths. The freeze protects the work that can still make money.
1. Before work starts: define what the season must sell
Cost: two weeks of new-page production. What it buys: time to verify what matters.
Say you run a shop that lives off December, or a services firm that books spring in February. The outcome is not rankings or mentions. It is booked jobs you can staff, orders you can ship, or rooms you can fill at prices you will honor.
I start with four decisions:
- What you will sell.
- Where you will sell it.
- What you will charge.
- When you must stop accepting orders or bookings.
That list decides every page you keep. Everything else is cut, even if it ranked last year.
Losing two weeks of production feels like falling behind. But thin seasonal duplicates published in week seven can split link equity, confuse buyers, and arrive too late to help. Reliability is the better purchase.
A freeze on expansion is not a freeze on fixing what is wrong. Prices, dates, availability, and broken checkout still get corrected. New scope does not start. Put that distinction in the brief before anyone opens a content calendar.
2. Week 1: confirm capacity and record the baseline
Cost: one hard conversation with operations, plus a repeatable baseline. What it buys: a scope you can fulfill and a way to judge progress.
Confirm availability, service areas, capacity, prices, deadlines, and cutoff dates. For installations, establish the last bookable date. For goods, establish the last shippable date and what happens when stock runs out.
Cut anything you cannot fulfill, even if it drove traffic last season. A page that promises what operations cannot deliver does not create demand. It creates refunds.
Then record a baseline before anyone touches the site:
- Pick 30 to 50 buyer prompts covering brand, category, comparison, and local or use-case questions.
- Run the same set across ChatGPT, Perplexity, Gemini, and Google AI Overviews, using fresh chats or search sessions as appropriate.
- Score each answer: 2 for a direct recommendation, 1 for citation-only, 0 for no mention.
- Run each prompt 3 to 5 times per engine and record the variation.
Keep the non-brand questions. Category and comparison prompts are the honest test; brand-only prompts flatter you. Repeat the runs because one AI answer is not a measurement.
That sheet is your baseline, not a causal control group. Without it, every later improvement is a story told from memory.
3. Week 2: commission implementation, not a recommendations deck
Cost: website access and an approver who responds within 48 hours. What it buys: work that reaches the site.
This is where seasonal SEO usually dies. The agency sends a 40-page PDF, the dev queue ignores it for three weeks, and peak arrives with nothing changed. “We can’t implement your recommendations” is a leading reason clients fire SEO agencies. Advice is not the scarce resource here. Implementation is.
The provider’s scope must include:
- Content implementation.
- Technical fixes.
- Weekly reporting on shipped work and outcomes.
If they will not touch the site, they are not your seasonal provider.
Your side is equally plain: provide access, establish a factual approval path, and name one decision-maker who answers within 48 hours. One approver is the constraint working for you. No committee. No “run it past everyone for two weeks.”
If you are considering groas, confirm in writing that the seasonal scope covers implementation across Google and AI search, not advice alone. The fit is execution with an action log and a named human strategist accountable for direction. A recommendations-only provider leaves the deadline risk with you.
Cut audits without access, keyword lists without pages, and deliverables that require your team to do the hard part. Buy shipped work, not homework.
4. From week 2: require one useful report every Friday
Cost: one page of reporting and a weekly blocker review. What it buys: fewer surprises before the freeze.
I set the report format before the first fix ships. Same format every Friday. Blockers before wins.
The page contains:
- Unresolved blockers: what is waiting, who owns it, and the decision date.
- Work shipped: pages fixed or published, with URLs and a change log.
- Google visibility: indexed pages and Search Console clicks, not impressions alone.
- AI answers: mentions and citations from the fixed prompt set across ChatGPT, Gemini, Perplexity, and Google AI Overviews.
- Enquiries and sales: calls, forms, bookings, and orders.
If traffic leads the report and enquiries hide on page four, the provider is optimizing toward the wrong outcome. For the prompt method, use the 30-prompt test for brand mentions and citations so the counts stay comparable week to week.
Separate observation from attribution. “Citations rose from 4 to 7 on the fixed set in the week we corrected prices” is observed. “Our edits caused the rise” is a claim. With answers varying run to run, one good week proves little.
I also want the limits stated plainly: no guaranteed rankings, citations, or seasonal revenue. groas’s action log makes the discussion more useful: what shipped, why it shipped, and what happened afterward. That beats a slide saying visibility is up and hoping nobody asks which edit did it.
Cut reporting theater. Keep the numbers, the changes, and the decisions needed next.
5. Weeks 3–4: repair buyer pages before adding URLs
Cost: two weeks of repair-first discipline. What it buys: pages that can be found and can take an enquiry.
A page that cannot be crawled cannot rank. A page that ranks but cannot take an enquiry cannot sell. I work through the site in that order:
- Check access. Look for robots restrictions, password walls, and staging noindex directives left over from the last redesign.
- Check indexing. Review Search Console for crawl errors and inspect the seasonal URLs that should drive calls or orders.
- Repair internal links. Link buyer pages from the homepage, main category, and footer where those links make sense.
- Test conversion paths on mobile. Submit forms, check phone numbers, follow booking links, and walk through checkout.
I have watched accounts spend week three debating headlines while the main booking form returned a 404 on iPhones. Fix the till before you paint the shop.
What gets cut is the reflex to create a new seasonal URL for every underperforming page. Overlapping pages can split link equity and risk keyword cannibalization. Update the existing page when it already answers the buyer’s question.
For recurring peaks, keep one permanent URL without a year in the slug and refresh it each season. Do not launch /christmas-2026/ just to sit beside /christmas-2025/.
A new page earns its place only when a distinct buyer question needs a distinct answer: delivery cutoffs versus installation lead times, for example. Publish those missing pages in weeks three and four. The three-to-six-week publishing lead time is a planning allowance, not a promise that Google will rank them by launch.
6. Weeks 5–6: finish the evidence buyers need to choose
Cost: the final content work before the freeze. What it buys: fewer unanswered purchase questions.
Google rankings get you seen. Shortlist evidence gives a buyer a reason to choose you. I check every buyer page for four things:
- Suitability: who it is for and who should skip it.
- Limitations: lead times, service boundaries, and stock limits.
- Honest comparisons: when a cheaper option fits better.
- Purchase requirements: measurements, access, ID, deposits, or booking windows needed to fulfill the order.
Most seasonal pages list features and prices, then stop. The buyer has to guess the rest or ask ChatGPT. Put the answer on the page first, in plain sentences, so an AI answer has something correct to quote.
These weeks finish the pages already in scope. They do not reopen the URL list.
Next, reconcile the facts across the site, Google Business Profile, delivery or booking pages, and the two or three third-party listings buyers actually check. If the site says order by December 18 and the profile says December 20, the buyer gets two answers to a question with one deadline.
Re-run the fixed prompt set and read for accuracy before counting mentions. AI-cited pages do not simply mirror Google’s top 10, and a wrong price quoted by Gemini can hurt more than no mention.
Fix the source page, not the prompt. At the end of week six, page and section scope closes. Corrections stay open.
7. Weeks 7–8: freeze expansion and verify the buying path
Cost: no new pages, sections, templates, or routine system changes. What it buys: two weeks focused on accuracy and a working purchase path.
Treat this like a retail code freeze: stop routine releases that could break the site or connected systems during the busiest period. Finish integrations, checkout changes, and new apps before week seven.
Theme, checkout, and tracking changes stay locked unless an urgent repair is necessary. Prices, dates, availability, and offer settings can still be corrected. The same named approver signs off on urgent repairs; the freeze does not create a second committee.
My rule is simple: correct the wrong date. Do not launch a new template.
Use the remaining time in this order:
- Reconfirm seasonal facts with operations. Capacity and availability may have moved since week one. Check every live promise against what the business can still deliver.
- Inspect indexing in Search Console. Read the result honestly. Google does not guarantee crawling, indexing, or serving a page even when it follows Search Essentials. Indexing makes a page eligible to appear; it does not guarantee visibility.
- Re-run the fixed prompt set. Check recommendations, quoted facts, and citations. Log variability instead of chasing one flattering answer. Several weekly cycles moving in the same direction are more useful than a single good run.
- Test the buying path end to end. Check the enquiry form, booking link, checkout, phone number, and analytics. Repair what is broken without expanding scope.
No provider can promise indexing or an AI citation by your launch date. Do not multiply pages because a citation has not appeared. Last-minute URLs are a poor substitute for verifying the pages already live.
Keep the new idea in the backlog. Keep the working checkout in production.
8. After the peak: lift the freeze in order
Cost: a review of every seasonal URL within a week. What it buys: a cleaner starting point for the next season.
When the freeze lifts, do not celebrate by publishing. Give every seasonal URL one of four decisions:
- Keep: retain the permanent page that earned links and enquiries.
- Update: return seasonal prices, dates, and availability to their evergreen state.
- Redirect: consolidate yearly duplicates into the permanent page.
- Retire: remove one-off pages that never sold and no longer have a job.
The business keeps four responsibilities no provider can take away: availability, pricing, claims, and approvals. If operations cannot staff it, ship it, or honor the price, no SEO ranking or AI mention fixes it.
For the next phase, compare this sprint with $1,500 a month, one approval call, six months, which puts enquiry tracking before content for the rest of the year. The freeze ends. The discipline does not.
Lift it in order: first reopen measurement changes, then broader site fixes, then new pages. Use what the season taught you to decide which expansion deserves approval.
Eight weeks buys a small set of pages you can fulfill. Two frozen weeks verify them. After the peak, their performance decides what stays. I would rather enter peak with six correct pages than sixty hopeful ones. Correct pages take money. Hopeful ones take meetings.

