GEO pathway · foundation 05

Nothing tells the model it stopped being true.

A fact doesn't expire on its own in a generative engine's retrieved context. It sits there, dated or undated, until something newer and more authoritative displaces it — and most changes in the world never publish that signal.

Learning boundary. This lesson teaches an operating model. It is not a guarantee of ranking, citation, or inclusion in any specific engine's output.
Learning objective

Separate four currency conditions.

By the end of this lesson, you should be able to inspect an entity's published facts for staleness risk, assign evidence-based readiness scores, identify which facts decay fastest, and explain why an unchanged page is not evidence that nothing has changed.

01 · Explicit dating

Does the fact carry a visible timestamp?

Date every claim that can change — a title, a metric, a status — so an engine has a basis for treating it as time-bound rather than permanent.

02 · Change detection

Is there a process that notices when a fact goes stale?

Set an owner and a review cadence for time-sensitive claims — without one, a fact only gets updated when someone happens to notice, which for most entities is rarely.

03 · Decay-prone claim triage

Which facts are actually at risk?

Separate volatile claims — role, funding stage, headcount — from stable ones like founding year or degree earned; treating everything as equally perishable wastes effort where it isn't needed and misses it where it is.

04 · Republish signal

Does updating a fact actually surface as a change?

Update the modification date, resubmit to sitemaps, and refresh cached versions when a fact changes — an edit an engine's retrieval pipeline never re-crawls is functionally the same as no edit at all.

Practice lab · evidence before confidence

Score an entity's currency risk.

A founder's bio page states their title with no date attached. The title changed eight months ago when they became CEO of a new venture; the page still lists their prior role. No one on the team is assigned to review the bio, and the last edit to the underlying file — including the change eight months ago — never updated the page's dateModified field or triggered a sitemap resubmission.

  1. Score each dimension from 0: absent, 1: documented, or 2: tested.
  2. Name every claim that requires evidence before awarding a point.
  3. Identify the critical gaps causing the stale fact to persist.
  4. Write the smallest safe next experiment.

Answer rationale

Explicit dating is 0 — the title has no timestamp, so an engine has no signal it might be time-bound rather than current. Change detection is 0: no owner, no cadence, and the actual title change eight months ago is direct evidence the gap is real, not theoretical. Decay-prone triage is effectively 0 in practice — a title is one of the most volatile claims on a founder bio, and it was treated with the same lack of attention as a stable fact like a founding year would receive. Republish signal is 0: even the one real edit that happened never updated dateModified or resubmitted to a sitemap, so even a correct edit made today would carry no freshness signal forward. The appropriate next step is fixing the current title and setting dateModified and a sitemap resubmission as a standing part of any future edit — not just correcting the one stale fact in isolation.

STOP → DATE

Continue the system