Dear SEOs: In a year of tighter cash flow, ‘it depends’ isn't good enough. Know how to forecast your targets.
Dear SEOs: In a year of tighter cash flow, ‘it depends’ isn't good enough. Know how to forecast your targets.
The first AI performance report from a platform has arrived in BWT. I've used it to:
🟣 See what people as asking via branded grounding queries
🟣 Review competitor name queries to understand AI systems positioning
🟣 See what's not included; pages expected to perform but don't
The contrarian view often has an unofficially reserved spot, albeit placed lower down.
Be the one to challenge convention. To say something new. To dissent.
Content that is not interchangeable is visibility worthy.
At a certain point, Google is no longer looking for corroboration. It is looking for contrast.
If you want to earn visibility on an established topic, stop conforming.
Most searches don't have a single right answer. Google aims to surface a mix of views, both in classic search and AI responses.
SEOs rarely recognise this. Instead we churn out copycat content. Prompting AI to analyse SERP results & mirror the output.
Consensus scaled across hundreds of pages. Today, Google displays up to 10 versions. With AI, it will need only 1 per viewpoint. The rest aren't wrong. They're redundant.
Most SEOs are committing a marketing cardinal sin. The sycophancy of agreement.
AI systems and social algorithms reward sameness.
Search surfaces are, ironically, one of the last bastions that show different sides. Google will often support the overarching sentiment, then quietly add a but.
Sktech of woman looking at GA4 AI chatbot traffic
Track AI chatbot traffic with a GA4 exploration. Add the dimension Session source/medium & the metric Sessions.
Filter by regex
.*meta.ai.*|.*perplexity.*|.*claude.*|.*mistral.*|.*gemini.*|.*chatgpt.*|.*copilot.*|.*manus.*|.*grok.*|.*deepseek.*
Enjoy analysing!
Its was a real estate website, so the articles section was small in comparison to the whole domain. But yes, the vast majority of the culled articles drove few clicks.
Before publishing, run the Anonymous Test. Strip away header, footer, sidebar and other template elements. If no one could tell who wrote it, rework it.
This approach creates less content. It also removes excuses for publishing forgettable work.
The goal isn’t to be the top ranked page explaining a concept. It’s to be the brand with a perspective on it.
Winning content closes the distance between information (the what) and insight (the so what). Brand equity is built where standard advice fails.
When something feels thin or falls flat, the problem is rarely the writing. The fault is the focus. You pointed out the obvious rather than the interesting.
Stop optimising for the search bar and start optimising for the share buttons.
These articles were anti-advertisements. They didn't persuade, intrigue, or differentiate. They taught the market that the brand had nothing interesting to say.
It doesn't matter whether it's written by a human or AI. If it's commodity content, it doesn't deserve to be published.
Woman carrying an elephant with caption "your SEO content is an anti-advertisement"
I deleted over 60% of articles from a real estate website. The CMO was nervous for 3 weeks. Then clicks began to climb.
The blog was full of "perfect" SEO content. EEAT compliant. Ranking. Yet, mostly ignored.
Few clicks. No shares. No deeper site engagement. Because it read like ChatGPT wrote it.
table of SEO predications
My SEO ins and outs for 2026. What are your predictions?
The December 2025 Core Update is here, but the GSC data is not.
We already can't fully rely on GSC for accuracy. Now timeliness has gone as well.
These are not 'nice to have' features. They are the product.
Respect that Google. Deploy the data.
These are mostly cosmetic level issues. But they are also costly ones.
Not only because they block agent conversions, which are set to steadily grow. But also because they deter humans.
So trigger an AI agent for every conversion task on your site and remove the friction.
ChatGPT agent conversions sit at only 17%. Most abandoned because of poor website usability.
The top 5 conversion blockers are:
🟣 Requiring registration to early
🟣 Hard to fill forms
🟣 Poor filter functionality
🟣 Usability of on-site search boxes
🟣 Pop ups covering the active window
AI hasn't change that ratio. It simply punishes anyone who still gets it wrong.
Your website will always be your digital home. But conversions happen at the shop.
And with AI intermediating the journey, the website is no longer the default store front.
Being the destination matters. But distribution matters more. It's irrelevant how polished the site is, if you ignore the battle for brand salience, which happens off-site.
Effective marketing strategies focus around 60% of effort on long term brand building and 40% on short term sales activation.
Your brand should spend most of its time away from home, working on its digital availability.
🟣 Networking on social
🟣 Building friendships with digital PR
🟣 Sharing insights on Discover & News aggregators
🟣 Being recommended on search surfaces
This is where brand growth happens.
How will AI search change the role of websites? At the core, it wont.
A website is your brand's digital home. It stores your base facts and relationships. It empowers conversations and conversions. This role is consistent.
So before you rush off to manufacture false consensus, think beyond immediate results to consider if you can deliver on the brand promises you’re making.
Manufacturing social proof isn’t new. But the potential backlash of that negative sentiment being included in generative AI is. And these complaints have permanence.
The real person who feels disappointed and deceived. Who will take to Reddit or other review platforms to rebuke the recommendation. A criticism that will then be picked up by LLMs.
When SEOs pollute trusted spaces to manipulate LLMs, they forget the human. The real person who reads the thread, buys the product and uses it expecting a higher level of quality because of the recommendation.
So you successfully manipulated Reddit. Now what?
I saw a LinkedIn “growth hack” inserting product recommendations into Reddit to gain AI citations.
It wasn’t the post that annoyed me. It was the response. Thousands applauding the “amazing insights”. Without questioning the consequences.
What if your best marketing actions come from complaints?
These expose the moments where expectations and brand promise diverge.
The cause is either execution drift, only fixed by owning it… Or positioning misalignment... e.g. you can't be cheap & top quality.
www.zoomsphere.com/blog/what-if...
It is no coincidence agent bounce rates mirror human ones.
AI agents are exposing what users have endured for years.
Your UX sucks.
Treat every agent bounce as feedback. Fix the friction.
This isn’t about pleasing machines. It’s about finally meeting human expectations.
Agent behaviour isn’t broken. Your site is.
In ChatGPT Agent Mode, 63% of first clicks bounce.
The most common causes are:
🟣 Bot challenges
🟣 301 redirect to an irrelevant location
🟣 4xx or 5xx status codes
🟣 Registration walls
🟣 Slow load speed
🟣 Landing pages don't satisfy the search intent