The dated AEO record
AI slop is killing brands. Consistency is how you get recommended (2026)
Simon Young and Andy Price on AI slop, lazy prompting, and why answer engines punish inconsistent brands.
What does this video cover?
- 01AI can now do the work of a whole agency overnight. The risk is treating the first draft as the finished piece.
- 02Andy Price briefed ChatGPT, Claude and Grok like staff when rebuilding the Question.marketing identity. The first serious pass took about four hours; refining it still took days.
- 03Answer engines drop inconsistent brands. Conflicting facts across blogs and PDFs, or even capital M versus small m on question.marketing, give models a reason to leave you out.
- 04Fan-out queries pull more sources than any person could read, then cite and recommend. Simon's view for 2027: share of voice is not enough. Brands need to be chosen.
- 05Clean your history before someone else writes it. Old directors, leavers and outdated contacts still attached in public records confuse models.
- 06This week: lock one spelling and one set of numbers, publish what only you know, stop shipping the first AI draft, and ask Claude, ChatGPT, Perplexity and Copilot the question your best customer would ask.
- 07DMG Media figure: 89% drop in desktop click-through rate when an AI Overview is present (25.2% to 2.8%), not a claim of total site traffic lost. Over 900 AI visibility tools is Simon's own ongoing research.
Full transcript
AI slop is killing brands. Consistency is how you get recommended
Simon Young talks with creative director Andy Price about AI slop, lazy prompting, and why answer engines punish inconsistent brands.

The danger of "just good enough"
AI can now do the work of a whole agency overnight. That is not the risk. The risk is people treating the first draft as the finished piece.
Andy spent 34 years as a creative director in UK advertising. He has seen clients try to write their own ads, the Canva years of samey templates, and now a flood of AI posts that local audiences call out in Facebook groups. The pattern is the same: skip the brief, skip the craft, and the work looks synthetic.

Brief the bots like a team, not a vending machine
When Andy rebuilt the Question.marketing identity, he did not take the first Lovable output. He briefed ChatGPT, Claude and Grok like staff. He asked for positioning, rationale, brand guidelines and layout options. The first serious pass took about four hours. Refining it still took days. That is the point: AI collapses the waiting, not the thinking.
You only get out what you put in. "Just good enough" dilutes the brand until nobody, human or machine, knows what you stand for.
Why AI drops inconsistent brands
Simon has been talking to businesses that are suddenly invisible in AI answers. One client already appears in Claude, ChatGPT, Perplexity and Copilot, yet still has conflicting facts across blogs and PDFs. Another example from Simon's own assets: question.marketing with a capital M on one social profile and a small m on another. The bot asked which was correct. When two answers disagree, the safe move for an answer engine is to leave you out.
Andy has banged on about consistency for decades. Clients get bored of their own logo lockups. Customers do not. AI just made the cost of hopping around measurable.

From citation to being chosen
Six months ago you clicked ten sites and made up your own mind. Now a fan-out query pulls more sources than any person could read and returns a citation, then a recommendation. Simon's view for 2027: share of voice is not enough. Brands need to be chosen.
That only happens if you are consistent, trusted, and spoken about well by third parties. Agentic sales will go further. When a conversation looks like a buying moment, the system will only risk introducing advertisers it can defend. Bad or inconsistent reviews become a hard stop.
Clean your history before someone else writes it
A Hong Kong client audit found old directors, leavers still listed, and a former contact under investigation elsewhere still attached in public records. Thirty years of brand history does not disappear because you moved on. You have to publish a clear source of truth and teach the models what is current.
Simon also warns that competitors will start poisoning results with planted inconsistencies. Brands that do not lock their facts will lose by default.
What to do this week
- Pick one spelling, one lockup, one set of numbers, and make every page match.
- Write down what only you know, and put it where models can read it.
- Stop publishing the first AI draft. Brief it, challenge it, then ship.
- Ask Claude, ChatGPT, Perplexity and Copilot the question your best customer would ask. Note who they name, and who they skip.

Watch and listen
Watch the full video on YouTube. Listen to the podcast episode on Will AI Recommend You? on Buzzsprout.
Notes on figures
Simon refers to the Daily Mail losing "80, 89%" of traffic to AI answers. The published DMG Media figure is an 89% drop in desktop click-through rate when an AI Overview is present (25.2% to 2.8%), with a similar mobile drop. Later commentary described an 80 to 90% CTR hit on affected queries, while overall traffic impact can still be low single digits because AI Overviews do not appear on every keyword. See sources in the YouTube description.
The "over 900" AI visibility tools figure is from Simon's own ongoing research for Question.marketing's State of AEO and GEO Software work.
About the speakers
Simon Young runs Question.marketing and hosts Will AI Recommend You? Andy Price is a creative director with 34 years in advertising and marketing, Simon's business partner, and co-author of their earlier brand book.
Where does this fit?
For the bigger picture, read what answer engine optimisation is.
More from the library
- The dated AEO recordSEO Isn't Dead. The Name Is.Published
- The dated AEO recordWhich AI Search Visibility Tools Should You Use? (2026)Published
- The dated AEO recordDoes an AI agent team mean the end of agencies? My 23 Grok Bots (2026)Published
- The dated AEO recordIs the Meta Muse Charm a surveillance Tamagotchi? Should we worry about AI companions? (2026)Published
This entry is part of a dated archive, what we said, when we said it. For what works in AI search today, start with AEO at Question.marketing. For why the dates matter, see the receipts.
Where next?
The library holds every video we have published. The AEO pages explain the method behind them, what it costs and what the evidence actually supports.