Somebody asked me last week why their blog post sits at position 14 while a thinner piece of content holds position 3. Fair question. And the honest answer is rarely "your writing is worse." It's usually that the ranking page answers a question Google trusts it to answer, faster and more clearly, than yours does.
Content optimization is that gap-closing work. Not keyword stuffing, not rewriting everything from scratch — just making what you already published easier for a search engine to understand and easier for a person to trust.
I want to show you how I approach it now, because the rules shifted in the last couple of years. Some of what worked in 2020 actively hurts you today.
Key Takeaways
- Search intent beats keyword density. Every time.
- Being cited by AI answer engines is the new front page — and it rewards different writing than blue-link rankings did.
- An SEO score of 92 tells you almost nothing about whether you'll rank. I'll explain why below.
- Updating existing content usually returns more than publishing new posts.
- The 80/20 rule applies hard here: a handful of pages carry most of your organic traffic.
What content optimization actually means (and what people get wrong)
The phrase content optimization gets used for two very different jobs, and confusing them wastes months.
Job one is technical: formatting a page so a crawler parses it correctly. Title tags, headings, internal links, image alt text, structured data. Job two is editorial: making the page genuinely serve the person who searched.
Most people obsess over job one because it feels measurable. You run a tool, it gives you a score, you fix the red items, the number goes up. Satisfying. Almost meaningless on its own.
I made this mistake early. I had a page scoring 98 in a popular audit tool and sitting on page four. I had another page scoring 71 that ranked in the top five for a term with real commercial value. The tool was measuring compliance with a checklist it invented. It was not measuring whether anyone wanted to read the thing.
The checklist trap
Here's the thing most optimization advice skips: a crawler checks whether your page can be understood. It doesn't check whether your page is worth understanding. Those are separate problems, and only the second one moves rankings in competitive spaces.
So when you optimize content, ask two questions in this order:
- Does this page fully answer the query that brought someone here, including the sub-questions they'd ask next?
- Can a machine read the structure without guessing at it?
Get the first one wrong and no amount of technical polish saves you.
Optimization tips that actually move rankings
The tactics below are ordered by how much return I've seen from each, not by how easy they are.
1. Match the intent, not the keyword
Type your target phrase into a search box and look at what comes back. If the top results are all comparison tables and yours is a 2,000-word essay, you're fighting the format, and the format usually wins.
I once spent two weeks writing a detailed tutorial for a phrase where every ranking page was a simple definition. Nobody wanted the tutorial. Rewriting it as a short definition with one example took it from invisible to position 4 in about six weeks. Same keyword. Different shape.
2. Structure for direct answers
This is the biggest shift since AI answer engines entered the picture. When someone asks an assistant a question, the assistant pulls a short, self-contained passage and cites a source. If your answer is buried in paragraph nine of a rambling intro, you won't be the source.
- Put the answer in the first sentence of the relevant section, not the last
- Keep answer paragraphs to two or three sentences
- Use a heading that reads like the question a real person would type
- Add a short summary block near the top for the very impatient
The catch? This structure also makes your page better for humans. That's not a coincidence, and it's the main reason I'm not worried about AI killing search traffic.
3. Use internal links as context, not decoration
Every internal link is a sentence about what your page is about. If a page about email deliverability links to your pricing page with the anchor "click here," you've told a crawler nothing. Link with descriptive anchors to your genuinely related pages, and link back from those pages too.
On one site I ran, cleaning up internal anchors — nothing else, just the anchor text on about forty existing links — moved three pages up an average of five positions. Boring work. Real effect.
4. Refresh before you publish new
New posts feel productive. Updating old ones usually pays more.
Why? An existing page has history, inbound links, and a crawl pattern. You're improving something the engine already knows. A brand new page starts from nothing and competes for attention against everything else you publish that month.
My rough split now is one new piece for every three refreshes. That ratio would have sounded absurd to me a few years ago.
5. Make your data citable
Vague claims are invisible to answer engines. If you have a number from your own work — a conversion rate you improved, a time you saved, a cohort you measured — state it plainly with the context around it. That's the material assistants quote, because it's specific and attributable to a real source.
| Optimization task | Effort | Typical return | Best for |
|---|---|---|---|
| Intent rematch on existing page | Half a day | High | Pages stuck on page 2–3 |
| Answer-first restructuring | 2–3 hours | High | Any informational page |
| Internal anchor cleanup | Low | Medium | Sites with 50+ pages |
| Schema and metadata fixes | Low | Low–medium | New or under-crawled sites |
| Full rewrite from scratch | Days | Variable | Pages with zero traction after a year |
What is the 80/20 rule in SEO?
Roughly 20% of your pages generate about 80% of your organic traffic. That's the shape of almost every site I've looked at, and it holds whether you have forty pages or four thousand.
The practical consequence is uncomfortable: most of your content library probably does very little. Not because it's bad, but because it targets queries nobody searches or duplicates something you already covered.
What I do with this: pull your top performing pages quarterly, look at what they have in common, and either improve or fold the rest. On a client project last year we consolidated eleven thin posts into three thorough ones. Combined traffic from those topics went up, not down. Fewer pages, more strength per page.
Is SEO dead now with AI?
No. But the job description changed.
What died is the version where you wrote a page stuffed with a keyword, collected links, and won. What replaced it is a version where you have to be quotable. Answer engines need sources. They need clear, specific, well-structured passages from pages they can verify. If you write those, you get cited — and a citation in an AI answer sends you traffic you'd have had to fight much harder for otherwise.
Here's what I've watched change in my own numbers. Pages structured as direct answers to specific questions picked up referral traffic from assistants steadily through the past year. Pages built around keyword density did not. Same topics, different architecture.
What does concern me is the temptation to write for the machine instead of the reader. That backfires. Assistants are trained on human-written text, and pages that read like they were assembled by a bot get nothing.
What are the top 5 SEO strategies?
Five things, in order of leverage:
- Pick queries you can actually win. A term where the top results are all major brands is not a plan, it's a wish.
- Answer completely. Cover the question and the three follow-ups it implies.
- Structure for extraction. Clear headings, short answer blocks, no buried conclusions.
- Build topical depth. One page rarely ranks alone; five related pages reinforce each other.
- Maintain, don't just publish. Set a review date for every page you own.
Notice what's absent. No link-building scheme, no technical trick. Those still matter, but they amplify good content. They don't rescue content that answers nothing.
Is an SEO score of 92 good?
It means your page passes most of that tool's checks. It does not mean you'll rank.
Scores are useful as a floor, not a ceiling. If a tool flags a missing title tag or a broken heading hierarchy, fix it — that's real. But once you're in the high eighties, the remaining points are usually cosmetic, and chasing them has near-zero effect on visibility.
I've seen pages scoring in the low seventies outrank pages scoring in the high nineties on the same query. The difference was intent match and depth of coverage, neither of which shows up in a score.
Use the number to catch structural mistakes. Then close the tool and go read your page as if you were the person who searched. That reading tells you more than any score will.
What I do differently now
A few habits that came out of things going wrong.
I write the answer first and the introduction last. Sounds trivial. It stopped me from burying the point under three paragraphs of scene-setting, which was my worst habit for years.
I keep a running list of questions readers actually send me. Best source of headings I've found. Far better than guessing at keyword variations, because the phrasing is real.
And I stopped treating every page as finished. Every page gets a review date the day it goes live. Half of them never change. The other half turn out to be the ones that move.
A couple of loose ends
How often should I refresh content?
Review anything that gets impressions but few clicks every three to six months. If a page holds position 1–3 and keeps converting, leave it alone. Movement isn't the same as progress.
Do I need a tool to do this?
You need one for crawling and tracking. You don't need one to know whether your page answers the question. That part you can judge by reading it out loud.
The thing I keep coming back to: none of this is separate from writing well. The pages that rank now are the pages that answer a specific question clearly enough that both a person and a machine can point to the exact sentence that solved it. Everything else — the scores, the audits, the checklists — is scaffolding around that one job. Get the answer right and the structure is easy. Get it wrong and no tool will save you.
