Content refresh workflow

Select refresh candidates, diagnose the opportunity, document changes, and validate the outcome.

Derived analysis Partially automatable Intermediate

A content refresh should solve a documented user or business problem, not merely change a date or add keywords. Search Console can identify pages with existing Google visibility and show how exposed queries changed; it cannot prove content is outdated, reveal the entire topic market, or guarantee that an edit caused later growth.

Stage 1: Build a candidate pool

Use enough history to distinguish sustained movement from daily noise. Compare equivalent periods and account for seasonality.

Good candidate signals include:

sustained click and impression decline across relevant queries;
stable impressions with weaker CTR within a comparable position/context;
meaningful impressions for relevant subtopics the page answers incompletely;
an important page losing to a clearly better internal page or switching URLs;
changed products, laws, statistics, interfaces, or recommendations;
strong historic business value with recoverable relevance;
an update required for accuracy, trust, accessibility, or user experience regardless of SEO.

Exclude or separately diagnose:

accidental noindex, canonical, robots, status-code, or rendering problems;
a sitewide migration or ranking incident;
market-wide demand decline;
temporary news/event spikes returning to baseline;
pages with too little data for a reliable conclusion;
content whose purpose is obsolete and should be retired rather than padded.

Stage 2: Score candidates

Use a documented model, for example:

priority = business value × recoverable demand × confidence ÷ effort

Evaluate:

absolute lost clicks and impressions;
exposed relevant-query breadth;
current position and result context;
conversions, revenue, subscriptions, or strategic importance outside GSC;
accuracy and trust risk;
availability of original expertise or new evidence;
technical/editorial effort;
whether another page should own the intent.

Do not rank by percentage decline alone. A large percentage on a tiny baseline is rarely the best first project.

Stage 3: Capture the baseline

For each selected canonical page, save:

exact URL and Google-selected canonical;
comparison dates and data finality;
clicks, impressions, CTR, and average position;
material query groups and individual exposed queries;
country, device, search type, and appearance segments;
other pages shown for the target query groups;
analytics engagement and business outcomes;
current title, description, headings, structured data, internal links, and visible dates;
screenshot or export of the current Search result when useful.

Search Console attributes most performance to Google's canonical. Confirm it before joining page history or editing a duplicate that receives no credit.

Stage 4: Diagnose the page-query relationship

Demand declined

If impressions fall while position is broadly stable, check seasonality and Google Trends. An editorial rewrite cannot restore vanished market interest by itself.

Visibility declined

If relevant impressions and position fall, review intent satisfaction, competing results, originality, accuracy, internal discovery, and site-wide context. Do not assume word count is the issue.

CTR declined

If impressions remain but clicks fall, compare similar position, device, country, query, and appearance groups. Inspect whether the title and snippet accurately communicate the page's distinctive value. Result-layout changes can reduce CTR without a page defect.

The wrong page appears

Filter a query and inspect Pages over time. Decide whether the current page, another page, a merge, or a new purpose-built page best satisfies the task. Multiple pages receiving impressions is not automatically harmful cannibalization.

Stage 5: Choose the intervention

Decision	Use when
Refresh in place	Same purpose and URL remain correct; accuracy, depth, evidence, or presentation need improvement
Consolidate and redirect	Several pages duplicate one intent and one destination can genuinely replace them
Split content	One page serves materially different tasks that deserve distinct destinations
Retire with 404/410	Content is obsolete with no useful replacement or archival purpose
Leave unchanged	Evidence is weak, movement is expected, or the page already serves users well

Redirect only to a true substitute. Redirecting unrelated retired content to a homepage or broad category can create a soft 404 and disappoint users.

Stage 6: Make a people-first update

Google's guidance emphasizes original, substantial, trustworthy content created primarily for people. During the refresh:

correct factual errors and expired recommendations;
add first-hand evidence, original analysis, examples, or expert review;
answer the core user task completely without padding;
remove obsolete, redundant, or unsupported material;
make authorship, sourcing, and update methodology clear where expected;
align the title and main heading with the page's actual purpose;
improve useful internal links and navigation;
update structured data to match visible content;
preserve the URL when its identity and purpose remain stable.

Do not add every visible query verbatim. Query rows are evidence of association and are incomplete because of privacy and top-row limits.

Stage 7: Handle dates honestly

Update the visible “last updated” date and dateModified only when the page changed substantially enough that readers benefit from knowing. Keep visible and structured dates consistent, include an accurate time zone where supplied, and do not use a future date or the date of an event described by the page.

Changing a date alone does not make content fresh or guarantee Google will display that date.

Stage 8: QA and release

Before publishing:

review facts, citations, spelling, and accessibility;
confirm 200, indexability, canonical, title, and robots directives;
test structured data where relevant;
verify internal links and redirects affected by consolidation;
confirm sitemap lastmod reflects a meaningful update if used;
add a concise Search Console annotation and a durable external change record;
record the release cohort and unchanged comparison cohort.

Custom annotations are shared across a property, capped at 200, hidden in comparison mode, and deleted after 500 days. They are context, not an experiment log.

Stage 9: Measure the outcome

Allow time for recrawl, reprocessing, demand exposure, and business conversion. Use several checkpoints rather than one arbitrary day:

early: live correctness and recrawl/index evidence;
intermediate: impressions and relevant-query/page mix;
outcome: clicks, qualified engagement, and business value over a complete comparable period.

Compare:

before versus after using equivalent complete dates;
year-over-year where seasonal;
updated pages versus a similar unchanged cohort;
targeted query groups versus unrelated query groups;
device/country/appearance segments where the hypothesis predicted a difference.

Google notes that competitors, user sentiment, news, and other changes can occur simultaneously. Describe results as consistent with the update unless the design supports a stronger causal claim.

Refresh brief template

Canonical URL:
User task and business role:
Evidence period:
Observed problem:
Technical causes excluded:
Queries/pages involved:
Intervention and why:
Facts/evidence to update:
Sections to add, revise, or remove:
Internal links and structured data:
Release date and annotation:
Leading and outcome measures:
Comparison cohort:
Decision date: