Content strategy for the LLM era. Six principles that replace the old playbook.

The content playbook that won 2018 to 2023 SEO ("comprehensive ultimate guides, target a primary keyword, structure for featured snippets") doesn't survive contact with LLM answer engines. Some of its assumptions are now actively harmful. Some are still true. Knowing which is which is now the job.

Six principles to replace the old defaults, with what they mean for how you operate.

#Principle 1. The page is the unit, not the keyword

Old model: pick a keyword. Build the page that ranks for it. Optimize for that keyword across the page.

New model: pick a question a buyer would actually ask. Build the page that answers exactly that question. Optimize for the question, not the keyword.

Operational implication: more pages, each more focused. URL slugs that look like questions ("how-to-block-bytespider") or specific answers ("perplexity-api-citations-explained") not abstract topics ("ai-crawlers"). Internal linking organized around question hierarchies, not keyword clusters.

#Principle 2. The lead is load-bearing

The first 100 words of a page now matter more than the next 1000. LLM answer engines read the lead carefully, decide whether the page is worth quoting, and lift directly from it when it is.

Operational implication: stop writing introductions that "set the context." Write the answer in the first paragraph. Then expand. The SEO opener of 2018 ("In today's competitive marketplace...") is worse than no opener at all.

#Principle 3. One original data point beats a thousand hot takes

The signal AI engines weight most reliably as "this site knows something firsthand" is original data. A small benchmark you ran. A survey of 200 customers. A dataset you assembled. Even modest original-data posts tend to get cited far more often than pure synthesis posts.

Operational implication: budget time for original research. Not big research, just small honest measurements. "We audited X sites and found Y" beats "Here's a survey of the literature on Z" every time.

#Principle 4. Refresh is the new publish

The "publish and forget" model is now mathematically losing. Pages without a recent dateModified lose citation rate fast in fast-moving topics.

Operational implication: 20% of your content team's time should be refresh, not new posts. Top 20 pages on a quarterly cadence, top 5 more frequently. Drive-by date changes without real edits is detected and penalized; the refresh must be real.

#Principle 5. The byline is part of the page

Anonymous content underperforms attributed content. The engines build a model of authors over time, and bylines tied to verifiable people accumulate weight.

Operational implication: every post has a single named author. The author has a public-facing profile somewhere reputable (LinkedIn, their own site, an academic page). The CMS supports Person schema. "Posted by Marketing Team" is the wrong answer.

#Principle 6. Distribution stays in the playbook

Despite all of the above, distribution and credibility-building outside your own domain still matter as much as ever. Maybe more.

LLMs cite the same surfaces humans cite. If your product is discussed favourably on Reddit, in industry newsletters, on review sites, in conference talks, those mentions feed back into citation likelihood. Inversely, if your only online surface is your own website, you're concentrated risk.

Operational implication: keep doing the off-site work. Get on podcasts. Run experiments and write them up so others link to them. Maintain a presence on the platforms your buyers use (LinkedIn at minimum, ideally X / GitHub / the niche communities of your vertical).

#What this changes operationally

A content team built around the old playbook usually has:

• Three to five SEO writers producing 8 to 15 posts a month
• A keyword-cluster pipeline managed by an SEO strategist
• Distribution as an afterthought (newsletter promotion if anything)
• Refresh treated as nice-to-have

The team needed for the new playbook looks more like:

• One to three editorial writers producing 4 to 6 substantive posts a month, with bylines
• One research / data person who runs small studies that become publishable findings
• Refresh as 20% of writer time, scheduled
• Distribution as a named function (PR-lite, podcast outreach, community management)

This is a budget reallocation, not a budget increase. The cost of "ultimate guides" goes down. The cost of "original measurements plus distribution" goes up. Net cost is similar. The output is better content that AI engines actually want to cite.

#What hasn't changed

Some things are still true.

• Technical SEO foundations still matter. A slow site, a broken canonical, a missing sitemap will hurt you in both Google and AI
• Audience-fit still matters. A great post on a topic nobody searches for is still a great post nobody reads
• Editorial quality still matters. Bad writing isn't redeemed by good structured data

You're not throwing out the old playbook entirely. You're updating its priorities for an audience that includes machines that summarize for humans.