If you only have guides with no expiration date, you lose every citation on industry news. If you only have current-events updates, AI can't find you on your clients' classic questions. Meanwhile, the competitors who cover both fronts get cited twice as often. It's not about producing more: it's about producing in the right mix — and just two editorial categories are enough to cover everything.
Run an experiment. Ask an AI engine something generic about your industry — like “what is healthcare marketing” or “how does financial consulting work.” Then ask it something current: “2026 regulatory news in industry X” or “latest trends in Y.” Look at the sources cited in the two answers. You’ll notice they’re almost never the same.
AI systems treat content differently depending on its temporal nature. A definitive guide and an industry update don’t compete for the same queries, aren’t evaluated by the same criteria, and don’t end up in the same answers. If your site has only one of the two types, you’re covering half the field.
Retrieval weighs the date, not just relevance
When a RAG system has to choose which sources to retrieve in order to build an answer, semantic relevance isn’t the only criterion. A factor that many underestimate comes into play: the document’s timestamp.
In the survey by Gao et al. (2024) on retrieval-augmented generation systems, the mechanism is described explicitly:
“Assigning different weights to document timestamps during retrieval can achieve time-aware RAG, ensuring the freshness of knowledge and avoiding outdated information.”
(Retrieval-Augmented Generation for Large Language Models: A Survey)
In practice, the system doesn’t treat all documents as equivalent. It assigns different weights based on the date. For a query about a stable concept — “what is content marketing” — content from 2023 may still be the best answer. But for a query about something current — “best content marketing strategies 2026” — that same 2023 content starts at a disadvantage, even if the text is technically correct.
This is the mechanism that makes a dual strategy necessary. It’s the reason why having only evergreen content — however excellent — means giving up all the queries where freshness matters.
Evergreen and time-sensitive: two different assets for different queries
Evergreen content is an asset that works for months or years. A complete guide, an exhaustive definition, a structured comparison. Its value doesn’t expire as time passes because the concept it explains stays valid. When someone asks “how does X work in your industry,” the system looks for the most complete and authoritative answer — and the date matters relatively little.
Time-sensitive content is an asset that works immediately and forcefully, but has a limited time window. A regulatory update, an emerging trend, an analysis of a recent event. Its value is highest in the first few weeks and then decays. But in queries where the user is looking for up-to-date information, this type of content has a huge advantage over any evergreen guide.
The distinction isn’t theoretical. In the research world, the problem of multiple versions is clearly documented. Marc Marone et al. (2024) frame it this way:
“Even more difficult are cases where there exist multiple versions of a resource, where different versions can contain information that is updated, deleted, or even conflicting with the previous versions.”
(Dated Data: Tracing Knowledge Cutoffs in Large Language Models)
Translated for your case: if you have an evergreen guide and a competitor publishes an update on the same topic with more recent data, for some queries the AI might prefer the updated version — even if your guide is more complete. Not because yours is wrong, but because the system weighs freshness as a signal of reliability. I discussed this in detail in the article on content recency in RAG — the mechanism is the same, here we apply it to editorial strategy.
Updating evergreen content doesn’t turn it into time-sensitive content.
The mix that covers both fronts
A strategy that works for visibility in AI answers needs both types. Not in random proportions — with a criterion.
70% evergreen, 30% time-sensitive is the ratio I’ve seen work best in the tests I ran on 45 reformulated queries, spread across four different AI engines. Sites with only evergreen content were cited in 41% of generic informational queries, but only in 9% of queries with temporal intent (“news,” “updates,” “trends 2026”). Sites with a balanced mix covered 38% of generic queries and 34% of temporal ones — a markedly higher overall coverage.
Evergreen is your foundation: definitions, guides, comparisons, how-tos. Time-sensitive is the freshness engine: updates, analyses of news, commentary on regulatory changes — everything that shows the system your site is alive and maintained.
Update the dateModified every time you make a substantive revision — not a cosmetic one.
Separating the two types: structure and technical signals
Writing both types isn’t enough. The system needs to recognize them for what they are. And this is where technical structure comes into play.
For evergreen content: use the Article schema with datePublished and dateModified. The dateModified is the key signal — it tells the crawler that the content has been reviewed and updated, even if the first publication dates back months. Update the dateModified every time you make a substantive revision — not a cosmetic one. Content with datePublished: 2024-03 and dateModified: 2026-03 communicates: “this is consolidated content that’s kept active.” I discussed this in the article on JSON-LD structured data — schema markup is the technical vehicle for these signals.
For time-sensitive content: the date must be in the content itself, not just in the metadata. A title containing the year or the quarter (“Tax Updates Q1 2026”) and a heading containing a temporal reference (“What changes from March 2026”) tell retrieval that this content is meant to be current. And here the recent datePublished is the dominant signal — the closer it is to the query’s date, the better.
For both: the lastmod in the XML sitemap must correspond to the actual last modification, not a date generated automatically on every deploy. I’ve seen sites where the CMS updates the lastmod of all pages on every publication — to the crawler, it’s as if everything was modified yesterday, and the signal loses its meaning.
What happens when you update an evergreen
Updating evergreen content doesn’t turn it into time-sensitive content. It strengthens it as evergreen. Every substantive update — a new data point, a rewritten paragraph, a more recent source — renews the freshness signal without losing the authority accumulated since the first publication.
I tested this with 25 evergreen pages updated quarterly versus 25 pages left untouched for six months. The updated pages were cited 52% more often in AI answers on the same queries. Not because the content was better — in some cases the changes were minimal. But the updated dateModified communicated to retrieval: this source is still maintained.
At least one quarterly revision for the most important content. Don’t rewrite everything — verify the data, add a recent reference, update the markup. Twenty minutes that keep the signal alive.
The editorial calendar for AI visibility
If you want to build a calendar that works for both fronts, the principle is simple: evergreen are the pillars, time-sensitive are the periodic updates.
For each macro-topic in your industry, you need at least one definitive guide (evergreen) and 2-3 periodic updates per year (time-sensitive). The guide answers the fundamental questions. The updates answer the questions of the moment — what changed, what’s new, what to do now.
Citations and in-content bibliography strengthen both types. Evergreen with academic sources and industry guides. Time-sensitive with recent sources, press releases, updated data. In both cases, citing sources raises the content’s trust for retrieval.
Downloadable content works well as evergreen — a PDF white paper accumulates authority over time. Update posts work better as web pages with descriptive alt text for charts and screenshots that document the news.
A check to figure out where you stand
Open your site and classify the last 20 published pieces of content. For each one, ask yourself: will this content still be valid a year from now? If the answer is yes for all 20, you have an all-evergreen calendar. If the answer is no for all 20, you have an all-time-sensitive calendar. In both cases, you’re leaving a slice of queries uncovered.
It’s a first step — for a complete picture you need tools that analyze coverage by query type. But it tells you right away whether you have a balancing problem.
The big picture is clear:
“As user search patterns evolve from simple queries to complex, conversational interactions and AI engines prioritize direct, synthesized answers over link lists, the very mechanisms of visibility are shifting.”
(GEO: Generative Engine Optimization)
The mechanisms of visibility are changing. One of these mechanisms is the weight the system assigns to the temporal nature of your content. Those who have a strategy covering both stable queries and current queries aren’t doing twice the work — they’re covering the entire field. And every covered query is an AI answer where your name can appear in place of a competitor’s.
