The dumping ground
Ideas, questions, links and half-finished thoughts. Dont let the good ones slide.
No notes match your search.
Information gain > content volume
Audit tools are excellent at finding things.
they are shit at deciding about what matters first.
Check the business impact
Before fixing the red score: what is it costing the business?
If the answer is “probably nothing”, keep moving.
Depth or just length
Need to look at whether Google rewards true topic depth, or whether we have simply decided that long pages look more serious.
Connected pages
Maybe ranking pages are not better.
Maybe they are just better connected to the rest of the site.
A lot of “SEO strategy” is deciding which obvious thing to do first.
The red score makes the problem feel more important than it is.
Contradiction · August 3, 2026
The red score makes the problem feel more important than it is.
Meanwhile the thing actually costing the business money sits quietly in the background because nobody has built a dashboard for it.
Missing meta description: red.
Wrong people landing on the page: nothing.
Should compare what the audit says is urgent with what the business is actually losing money from.
Suspect they are rarely the same list.
FAQ habit
Why does every content brief ask for FAQs whether anyone has asked those questions or not?
Pull actual questions from sales calls.
Stop inventing customer concerns in a keyword tool.
Customer voice is more important than anything marketing can write
Search volume tells you how often something is searched.
It does not tell you whether the person searching is worth speaking to.
A page can rank well for the right thing but still attract the wrong people.
Traffic is not the result. It is just people arriving.
Most reports are written to prove the work happened.
Very few explain why the work mattered.
information gain is just experience made visible.
Not novelty for the sake of it. Evidence that someone has actually done the work.
AI certainty
The more certain someone sounds about AI search, the less data they normally seem to have.
Everyone is building for LLM visibility without agreeing what visibility actually means.
Mentioned accurately
Maybe the opportunity is not being mentioned more often by AI.
It is being described more accurately.
AI referral test
Check whether AI answers are sending useful visits or just producing screenshots for LinkedIn.
Accepted answers
AI can reproduce the accepted answer.
It struggles more with the awkward answer that only appears after doing the work.
it just fakes it
Collect the mistakes
Collect examples where AI confidently gets the business wrong.
Could be more useful than another visibility tracker.
Schema theatre
Structured data can help a machine understand the page.
It cannot make the page worth understanding.
Internal linking is content design disguised as SEO.
Maybe crawl depth is really a measure of how much the site believes in a page?
if you think a page is important, does the architecture, weak linking kills a pages importance
Link count test
Compare internal link counts with enquiries.
Then compare them with rankings.
Suspect those are three different stories.
The page is not the unit
SEO tools make us look at URLs individually because URLs are easy to score.
The page may not be the unit.
The site might be.
Be careful with “authority”.
It has become one of those words people use when they mean “seems quite good”.
Patterns around expertise
Does Google understand expertise, or patterns that often appear around expertise?
Different question.
Quick wins later
Revisit the last ten “quick wins” six months later.
Did anything happen?
Publish failed work
There is probably more value in publishing what did not work than manufacturing another clean success story.
Case study structure
What did we expect?
What happened?
What changed because of it?
Probably enough structure for most case studies.
Case studies backwards
Most case studies are written backwards from the successful outcome.
That is why they feel fake.
Tell the difficult bit
Do not remove the bit where nobody knew what to do next.
Clients ask for redesigns
Clients rarely ask for clarity on a problem.
They ask for a redesign because they think design is the issue, but is it?
Some of the shttiest designs convert better. WTF
“Make it pop” shows client doesn’t know how to brief or give feedback.
Client language
Collect the phrases clients use when they know something is wrong but cannot explain it.
Those phrases are probably better research than the polished brief.
Translate later
Do not translate customer language into marketing language too early.
Professional copy
Most copy becomes weaker when it is made to sound more professional.
Professional often means nobody would naturally say it.
Proof before claim
Proof is usually buried further down the page than the claim.
Wrong order.
Maybe the best CTA is sometimes just the next useful question.
Review every CTA on the site.
How many are asking for too much, too soon?
Conversion optimisation should reduce uncertainty. It should not create panic.
Countdown timers
Observation · July 9, 2026
Countdown timers
Countdown timers are the website equivalent of someone standing too close.
They do not make the offer more useful.
They just make leaving it feel more urgent.
If the decision needs a clock ticking beside it, the page probably has not done enough to make the decision feel obvious.
Restraint converts
There is a post in why restraint converts.
Need examples, not opinions.
Homepage restraint
Maybe homepage design is mostly deciding what not to explain.
Useful pages
Stop calling every useful page a landing page.
Some pages are allowed to help people without immediately demanding a form submission.
WordPress is not slow. Shitty builds are slow.
Old decisions
A plugin stack is often just old decisions refusing to leave.
Audit how many plugins are solving problems created by other plugins.
Does custom code solve it?
Custom code
Custom code should remove complexity, not create a more impressive version of it.
The boring solution
Write about the boring solution being the better solution.
Need a real example where the clever version made sense but wasnt the best solution.
Configuration counts
Developers like rebuilding things because configuration does not feel like real work.
It still counts.
Technical debt is a decision you avoided when it was cheap, returning later when it is expensive
A website can cling onto the middle with shite code and bad tech seo for years.
Documentation later
Documentation is mostly useful six months after everyone thought it was unnecessary.
Inherited decisions
Make a list of inherited decisions nobody can explain.
Then see which ones everyone is afraid to change.
A theme, two builders, an optimisation plugin and a pile of snippets is still an architecture.
Just a really shite one.
Cache is not a strategy
Do not use caching to avoid finding out why the page is doing so much work.
Core Web Vitals vs rankings.
Still no clean relationship.
Check 50 sites again before writing this crap.
real data, real info, real gain
The interesting bit lives somewhere between the two.
A site with 200 missing alt tags can score worse than one with an accidental noindex.
Technically logical. Practically ridiculous.
Analytics confidence
A neat dashboard can make weak tracking look authoritative.
all fur coat and no knickers
Consent mode gap
How much of the trend is real, and how much is the measurement changing underneath it?
Events nobody uses
List every analytics event.
Who has used it to make a decision in the last six months?
Activity into achievement
There is something in how dashboards turn activity into achievement.
Average hides the page
Site-wide averages are good at hiding the one page that is actually broken.
Clicked more ranked lower
Clicked more. Ranked lower.
Why?
Research in inboxes
The best content research is usually sitting in customer emails nobody has thought to read.
Depth versus coverage
There is something here about depth vs coverage.
Most content is wide and shallow. Very little is narrow and useful.
There is a difference between being original and disagreeing with everyone for attention.
Avoid it.
Change my mind publicly
Could review old posts and mark the bits I no longer agree with.
A personal site should show movement, not just certainty.
Failed beside finished
Show the failed version beside the finished one.
The gap between them is probably the useful part.
Mess is the feature
The mess is the feature.
The best thing about a personal site might be that it does not know what it wants to become yet.
Personality > SEO + Personal Brand > Pitch
Organic evolution?
Portfolio polish
Portfolios remove the personality.
The more polished they become, the more forgettable they are.
Write about being wrong
People only show the wins, which makes all the work look easier than it was.
Research before conclusion
I enjoy the research more than the conclusion.
Might be why half of this never becomes a post.
Probably bollocks
Could make a note kind called “Probably bollocks”.
Might become the most useful one.
Come back to this
Come back to this.
No idea what “this” was.