Trendzeist says the next Black Friday peak starts on 1 October 2027, and that a page meant to rank for it should be live by 20 August 2027.
I ran this in early November 2026. Any store writing its first Black Friday page in November is, in practice, writing for 2027. The demand is here (November is the top month on a five-year average), but a page published two or three weeks before the sale has very little time to get crawled, indexed and tested in the results before the traffic is gone.
That's the whole case for a trend-driven calendar. The data on when people search has been public for twenty years. Most calendars still get planned from memory and whatever showed up on LinkedIn.
Three clocks every calendar has to respect
A topic is worth publishing when three things line up, and each has its own Trendzeist tool:
| Clock | Question | Tool |
|---|---|---|
| Season | When does demand peak every year? | seasonality |
| Momentum | Is it growing right now, or coasting on an old peak? | interest_over_time |
| Channel | Is it growing on web search, or somewhere else? | compare_channels |
Then a fourth tool, what_changed, turns the three into a weekly habit.
Clock 1: seasonality
seasonality pulls five years of weekly data, averages each calendar month, and counts a month as a peak only when it lands in a year's top two in at least three of those years. A one-off spike can't fake a season that way. Four keywords from my clients' world:
| Keyword | Seasonal? | Strongest months | Weakest months | Last 12 months |
|---|---|---|---|---|
| black friday | Yes, 20.87x | November (100), October (14) | April, May | Up 17% |
| seo audit | No | August (100), May (90), September (89) | December (45), January (47) | Up 113% |
| website redesign | No | June (100), August (94), May (91) | December (41), January (45) | Up 231% |
| woocommerce | No | Flat, every month between 79 and 100 | December (79) | Up 26% |
Black Friday is the extreme: 20.87 times more interest in its peak months than in the rest of the year, with a publish-by date returned for you.
The middle two rows are the useful ones. "seo audit" and "website redesign" don't pass the seasonality test, because their top month moves around from year to year. But both run at roughly twice their December level through the summer. Treat it as a tendency: a site redesign offer launched in mid-December goes out when interest is at its lowest, and a page published in April has time to rank before the May to August bulge.
WooCommerce interest is flat across the year. There's no right month, so publish on momentum alone.
Clock 2: momentum, and why the window decides the verdict
The same keyword, two windows:
| Window | Verdict for "claude code" |
|---|---|
| 5 years, last 12 months vs the 12 before | Up 656%, "new" topic |
| 12 months, last quarter vs the quarter before | Down 49%, falling |
Both are true. Claude Code barely existed two years ago, so the long view is an explosion. Inside the last year, interest peaked in late March and slid back. And seasonality refuses to guess at a season for it: only two full years of data, three required.
Choose the window that matches the decision. For "what do I write this month", use the 12-month window and its direction_now, which measures the recent slope. For "is this topic worth building a cluster around for the next two years", use five years. A seasonal topic that's also falling year over year gets a refresh, not a new post. The tool's own guidance says the same.
Clock 3: channel
Web search isn't the only place demand lives. compare_channels splits a topic across five Google surfaces and compares growth, since each channel has its own 0 to 100 scale and the absolute levels can't be compared:
| Surface | "n8n", last 3 months | "claude code", last 3 months |
|---|---|---|
| Web search | -49% | -49% |
| YouTube | -14% (stable) | -8% (stable) |
| News | +202% | +213% |
| Images | -51% | -20% (stable) |
Same story for both tools. Web search interest halved, news interest tripled, and video held steady. For a site like mine, which lives on tutorials, that means fresh how-to posts are chasing a shrinking audience. The opportunity is timely pieces when something ships or breaks, and keeping the evergreen guides current for the people who still search.
The weekly review
The three clocks tell you what's true today. what_changed tells you what moved since last week. It saves a snapshot of the related queries every time a Trendzeist tool fetches fresh data, then compares the current list with the snapshot from at least since_days ago:
what_changed(seeds=["claude code", "n8n"], since_days=7)
On a first run there's no history to compare, so it records a baseline and lists the current breakouts and rising queries as first_seen, marked "no history yet". When I ran it again the next morning, no snapshot was a week old, so it compared with the one from the night before and said so: "compared with one from 0.7 days ago (7 requested)". The comparison starts to mean something in the second week, when each item comes back with an action:
| Change | Action |
|---|---|
| A breakout that wasn't in last week's snapshot | Investigate now |
| A known query promoted to breakout | Investigate now |
| A newly rising query | Consider |
| A long-standing top query that briefly rose and settled | Refresh the existing post |
| A query that cooled or left the list | Review: check its curve first |
The last row has a warning built into the tool: leaving Google's related-query list isn't proof that demand fell. The list is relative and bounded, so a query can drop off while its demand holds. Check interest_over_time and only retire a page when direction_now says falling.
Once the archive holds a few months of snapshots, recurring_topics adds a line to the review: topics that broke out again and again, under different seeds. Those deserve one durable page refreshed at each spike instead of a new post every time. My archive covers nine days so far, so it had nothing to report yet.
Trendzeist also ships a weekly_review prompt that runs this as a meeting, with four outcomes: publish now, plan for a date, refresh, or retire. I add Search Console to it, so every item comes with "do I already have a page for this?":
Run my weekly content review for these seeds: claude code, n8n,
woocommerce, llms.txt.
Use Trendzeist what_changed (7 days). For every new or promoted item,
check interest_over_time (12 months) and seasonality. Run
recurring_topics for the same seeds. Then check Search Console for an
existing page and its position.
Sort everything into PUBLISH NOW, PLAN FOR A DATE, REFRESH or RETIRE.
Every PLAN item needs a publish-by date. Every RETIRE item needs a
falling direction_now as evidence, or it stays.
The calendar this produced
Built from the 3 November 2026 data across this series, for my own site:
| Item | Action | When | Why |
|---|---|---|---|
| "Is llms.txt worth it?" answer page | PUBLISH NOW | This month | Gap with steady news coverage and no Wikipedia definition (AEO research) |
| WordPress MCP setup guide | PUBLISH NOW | This month | Sustained demand, zero impressions, sits on two of my services (visibility audit) |
| WooCommerce checkout troubleshooting | REFRESH | Now | Black Friday is when a broken checkout costs the most. Page four, flat year-round demand |
| Black Friday checkout readiness checklist | PLAN FOR A DATE | Live by 20 August 2027 | 20.87x seasonal peak, too late for this year |
| SEO audit explainer | PLAN FOR A DATE | Live by April 2027 | Up 113% over 12 months, summer bulge |
| Claude Code setup guides | REFRESH | Quarterly | Web demand down 49%, my share up. Keep current, don't add more |
| More OpenClaw coverage | RETIRE from the plan | Now | Interest went from 100 to 2. The existing page stays as it is |
Seven rows, and only two of them are new posts. I'd be suspicious of a data-driven calendar that came back mostly new posts: this kind of review mostly tells you what to fix and when, and keeps you from writing posts nobody searches for anymore.
Where this goes wrong
- Five years of weekly data still reflects five specific years. A season that started recently, or a market that changed, shows up late.
- The six weeks in
publish_byare a default. A new site in a competitive niche needs more time, and an established page on a quiet query needs less. - Seasons are local. Pass
geofor the market you sell to:geo="BR"orgeo="US-CA"change both the peak and the trough, andTRENDZEIST_HL=pt-BRreturns Portuguese queries. - None of this is volume. A perfect seasonal peak on a topic nobody searches is still nobody.
Run it on your calendar
Run seasonality on the five topics your business depends on, write the publish-by dates into the calendar, and run the weekly review prompt every Monday. Ten minutes a week, most of it Trendzeist waiting politely between requests to Google.
If you'd rather have it running without you, the Trend-to-Publish Content Pipeline is the version I build for clients: an n8n or Claude workflow that runs this review on a schedule, scores the opportunities against Search Console, drafts in your voice, and lands posts in WordPress behind a human approval step. It's the same design as the n8n SEO pipeline I wrote up, with the trend data added. To see whether it fits your site, book a free 30-minute Intro Call.
This is the last post in the series. The first one covers the setup and the full opportunity analysis.
What does your content calendar say you're publishing next August, and how did you decide?
