Ask most people what content decay looks like and you get a picture of a slow fade: traffic drifts down, the post slips off page one, and eventually somebody “refreshes” it — which usually means re-sprinkling the keyword and bumping the year in the title.
That is not the thing that kills a post. Here is this week’s wisdom: content decay is not your post getting old. It is your post getting wrong — and carrying on ranking while it does.
The receipt: two sentences I published in June that were false by September
My agent refreshed one of my own guides yesterday — a piece on MCP tools published 29 June and live for 78 days without a single complaint. Two sentences in it, both perfectly accurate the day they were written:
- “Can I use MCP tools with Claude.ai on the web? Not yet for custom servers.”
- “For most solopreneurs, 8–12 servers is a comfortable operational range.”
Both are now wrong, and not by a little.
Connectors you add at claude.ai/customize/connectors are pulled straight into Claude Code — provided your active auth is a claude.ai subscription login. (They quietly do not load when ANTHROPIC_API_KEY, an auth token or a third-party host such as Bedrock is supplying the credential — which is exactly the headless shape this container runs in.) And the 8–12 ceiling? Engineered away. Tool search is on by default now: tool names load at session start, and the full definitions get fetched on demand. The context cost that made 8–12 a sensible number is no longer the constraint it was.
Nothing announced any of this. No 404, no error, no rank collapse, no monitoring alert anywhere. The page kept ranking and kept being read. It simply stopped being true.
⚡ GET THE AI EDGE
Weekly AI tips that actually save you time and money. No fluff, no hype — just what works.
One upstream date, 32 of my own pages
The change has a date stamped on it. The MCP spec moved to revision 2026-07-28, and with it version negotiation moved off the one-time initialize handshake and onto every individual request. That hands you a brutal little staleness test: any guide still describing version agreement as a handshake predates July 2026.
So I ran that date against my own library this morning:
- 303 published posts on this site.
- 55 of them mention MCP.
- 32 of those 55 went live before 2026-07-28.
Thirty-two pages that could not possibly have accounted for a change that had not happened yet. None broken. None flagged. All of them still handing readers an answer that used to be right.
The check that takes five minutes
- Date your sources, not your posts. Open the changelog, spec or release notes for every tool you have written about and find the last date it changed. That date, not your traffic graph, is your audit filter.
- Search your library for the claim, not the keyword. Rank trackers tell you position; they cannot tell you a sentence is false. Grep your own archive for the assertion itself — “not supported”, “you can’t”, a specific number — because that is where staleness hides. And read the whole result set, not page one of it.
- Correct in public; never quietly delete. Update in place, keep the URL, keep the link equity, and say plainly what changed. Every competitor still ranking on the old answer is wrong in precisely the same way and has not noticed yet.
The wisdom
Your rankings tell you how you are doing against your competitors. They tell you nothing whatsoever about whether you are right. Those are two separate questions, and only one of them comes with an alarm attached.
Ranking number one with a wrong answer is the worst outcome available to you — maximum reach on your least accurate work. So audit by vendor date, not by traffic dip.
Want an agent running that pass across your library every day instead of you? Book an automation strategy session and we will map it out.
📥 FREE: THE AI PLAYBOOK
The exact tools and workflows I use to run a one-person agency. 25 years of marketing experience distilled into an actionable guide. Yours free.
AI & Digital Marketing Strategist
With over 25 years in digital marketing, e-commerce, sales, and international business, Jon Jones is a seasoned entrepreneur dedicated to empowering small businesses through innovative AI solutions. His journey began in investment banking on Wall Street, guiding companies through capital acquisition and complex markets, and has expanded to include expertise in web development, brand design, product development, and courses and training.
Fluent in Mandarin and with a rich foundation in both finance and tech, Jon launched JonJones.ai to provide accessible, transformative tools for small businesses to streamline operations, generate leads, and build sustainable growth. His mastery of AI, automation, and digital strategy has enabled numerous businesses to bridge the gap between traditional practices and modern technology, ensuring they thrive in today’s digital landscape.
