Content Decay Analyst
Finds the pages that used to work, and why they stopped
- 8hard rules
- Things it must never do, enforced in every run
- 5handover rules
- Conditions where it stops and gives the work to a person
- 8output fields
- A structured result, not a paragraph you have to re-read
- ≤6steps per run
- Bounded work: at most 5 tool calls, enforced by the runtime
Read-only by design. This agent has no path to sending, spending, publishing or committing anything. Where that limit is the product, removing it would remove the reason to trust the output.
Overview
Content teams are measured on publishing, so they publish. Almost nobody is measured on the library, and a library decays: a page that ranked and converted two years ago slowly stops doing either. The decline is gradual enough that no alert fires, and the site total stays flat because new pages replace what old ones lose. The site looks stable while the library rots underneath it.
That is the expensive part, because updating a page that already has authority costs a fraction of what a new page costs to rank, and it works faster.
The distinction that makes this actionable is why a page declined. A page whose whole topic lost interest is not worth updating — the demand went away. A page holding position while the topic grows is losing share to something better. A page that fell after a site change did not decay, it broke. A page still ranking and no longer converting has a page problem, not a traffic problem. Four causes, four responses, and a traffic chart shows the same downward line for all of them.
There is also usually a pile of pages competing for the same query, written years apart by people who did not know the others existed.
This agent reports each cause with the evidence and ranks by recoverable value. It publishes and edits nothing.
Benefits
Traffic recovered cheaply
The pages already have the authority a new page has to earn.
Effort spent where it pays
Rather than on a topic whose demand has gone.
Cannibalisation resolved
Which usually lifts several pages at once.
What it does
Separates lost demand from lost share
A topic nobody searches for any more is not worth updating. A page losing to something better is. The chart looks identical.
Catches the break, not the decay
A page that fell the week of a site change did not decay. It broke, and that is a different team's fix.
Finds pages that rank and no longer convert
A page problem wearing a traffic problem's clothes.
Finds pages competing with each other
Written years apart by people who did not know the others existed, splitting one page's authority.
Ranks by recoverable value
Updating a page with authority costs a fraction of ranking a new one, and works faster.
Publishes nothing
It never edits, rewrites or publishes a page. Generated updates are how a library gets worse.
What this changes
What this changes about your week
- It takes over
- Finding which pages stopped working and why.
- Done today by
- Nobody. Teams are measured on publishing, not on the library.
- How often
- Never — the total stays flat while the library rots underneath.
- Typical manual effort
- Each decline needs its cause separated: lost demand, lost share, a break, or a conversion problem.
Why Content in particular. Updating a page that already has authority costs a fraction of ranking a new one. But a traffic chart shows the same downward line for four causes, and only two are worth acting on.
Runs unattended
Started by you or by an event, and it finishes on its own. Nothing waits for someone to be at a desk.
The same standard every time
The two-hundredth item is held to the bar the first one was. Consistency is the part people cannot sustain.
It cannot act on its own
Content Decay Analyst has no path to sending, spending or committing. That limit is why its output is safe to act on.
How it runs
This agent runs server-side through the PROMIVO runtime. Each run is logged step by step and every tool call is permission-checked before it executes.
- 1InputValidated against the agent input schema
- 2Modelclaude-sonnet-5 with automatic fallback
- 3Toolsextract_structured_data
- 4PermissionsRuns READ actions automatically; anything higher waits for approval
- 5OutputStructured result matching the output schema
- 6LoggingTokens, duration and every step recorded
Demo
Demo dataIllustrative sample output, abridged.
Input
{
"pages": [],
"period": "24 months",
"currency": "USD"
}Output
{
"escalate": true,
"declining": [
{
"url": "/guides/invoice-financing",
"cause": "lost-share",
"evidence": "Search demand for the topic is up 14% over the period. The page fell from position 3 to 11. The audience is still there and going elsewhere.",
"whatItNeeds": "An update. The page already has the authority a new page would spend six months earning.",
"worthUpdating": true,
"declineFromPeak": 71,
"recoverableEstimate": 41000
},
{
"url": "/blog/2023-tax-changes",
"cause": "lost-demand",
"evidence": "Search demand for the topic is down 89%. The page held its ranking throughout.",
"whatItNeeds": "Nothing. The page did not fail; the question stopped being asked.",
"worthUpdating": false,
"declineFromPeak": 84,
"recoverableEstimate": 0
},
{
"url": "/pricing-calculator",
"cause": "technical-break",
"evidence": "Traffic fell 96% in the week of the March migration and never recovered. This is not decay and it is not a content problem.",
"whatItNeeds": "An engineering fix, urgently. It has been broken for five months.",
"worthUpdating": true,
"declineFromPeak": 96,
"recoverableEstimate": 88000
},
{
"url": "/solutions/logistics",
"cause": "conversion-decline",
"evidence": "Ranking and traffic are stable. Conversion fell 58% after the form was changed in May.",
"whatItNeeds": "The page, not the traffic. No content work would have found this.",
"worthUpdating": true,
"declineFromPeak": 0,
"recoverableEstimate": 26000
}
],
"improving": [
"6 pages have gained without being touched. Worth knowing what they have in common before the next round."
],
"disclaimer": "An analysis of your own analytics and search data. Nothing has been written, edited, published or unpublished, no ranking or traffic outcome is guaranteed, recoverable values are estimates from each page's own history rather than forecasts, and pages where demand data is missing are reported as unseparable rather than classified.",
"cannotSeparate": [
"Demand data is missing for 9 declining pages, so lost demand and lost share cannot be told apart for them."
],
"competingPages": [
"Four pages target the same query, written between 2022 and 2025 by people who did not know the others existed. They are splitting authority that would rank better as one."
],
"escalationReason": "A top-decile page has been broken since the March migration with 88,000 of estimated recoverable value, and four pages compete for one revenue-driving query.",
"notWorthUpdating": [
"11 pages have lost demand rather than share. Updating them recovers nothing, and saying so is the point."
]
}Requirements
Integrations
No integrations required.
Before you start
- Page-level traffic, ranking and conversion history
- Search demand per topic over the same period, where available
- Publish and update dates per page
- Dates of site changes or migrations
- Requires the All-Access plan or higher
Use cases
Quarterly content review
What decayed, why, and what is worth recovering.
Flat traffic with steady publishing
The classic sign that new pages are replacing lost ones.
After a site migration
Which declines are breaks rather than decay.
Planning a content quarter
Updates ranked against new pages on cost to recover.
Pricing
All-Access
Billed monthly through your PROMIVO subscription. Cancel at any time.
Runs consume your plan allowance for agent executions and tokens. See plan limits
FAQ
Will it rewrite the pages?
No. It never edits or publishes anything. Generated replacement content is how a library gets worse — the pages worth recovering are worth a person's attention.
How does it know demand fell rather than ranking?
By comparing the page's own trend against the topic's search demand over the same period. Where demand data is not supplied it says the two cannot be separated rather than guessing.
Does it do keyword research?
No. It works from your own analytics and search data, on pages you already have. It is about the library, not about finding new topics.
Reviews
Reviews open once customers have run this agent.
Need this agent to work differently?
Tell us what to change and our team will scope a customised version for your business.
Customize this agent