The prevailing wisdom has been that AI search is a visibility problem. Get cited, stay in the answer, keep an eye on your ranking. Simple enough to plan a quarter around.
This week the story moved from visibility to arithmetic. New data shows automated systems, AI agents, crawlers, bots, now generate 57.4% of all web requests, against 42.6% from actual humans. That is not a future prediction. That is where the web already sits, as of Q2 2026. Your website’s most frequent visitor is a machine, and it has been for a while without anyone announcing it.
Everything else this week sits downstream of that shift. Google removed advertisers’ ability to opt out of Local Inventory Ads on Shopping campaigns, effective from 31 August, whether you wanted the change or not. A Semrush analysis found only 15.2% of ChatGPT topic categories have a clear brand owner, leaving more than half still completely unclaimed. And Google’s own John Mueller warned that a routine bot-verification screen served to Googlebot can quietly get indexed instead of your actual page, invisible to anyone checking the site by hand.
The thesis for the week is this: control is shifting away from marketers in small, procedural ways, an API setting here, a crawler-blocking decision there, and the brands paying attention to the plumbing are the ones who will still have a say in how they’re found. Here is what mattered, why it mattered, and what to do with it on Monday.
Board Analysis, The Digital Maze’s weekly marketing briefing, goes deeper on all of this every Friday. Subscribe here.
Search & Intelligence: Backlinks Aren’t Dead. They’re Just Not The Whole Story Anymore.
Search Engine Land’s look at how AI search is reshaping link building strategy lands on a conclusion that should reassure anyone who has spent years building a digital PR programme: backlinks still matter. Nearly 70% of marketers say backlinks have a “big impact” on search rankings and 30% say they have a “moderate impact,” and the top result in Google’s SERPs still carries nearly 4x more backlinks than results in positions two through 10. Almost three-quarters of SEO leaders believe backlinks positively influence the likelihood of content appearing in AI search too, so the old signal hasn’t disappeared.
What has changed is where the citations are actually coming from. Reddit alone now appears as a source in 3% of all AI citations, a genuinely large share for a single platform that most digital PR programmes still treat as an afterthought. That is a signal that AI citation behaviour is diverging from classic ranking factors even while broadly agreeing that authority still matters. A separate Search Engine Land guide published the same week reinforces the wider point: AI search results are increasingly personalised per user rather than ranked identically for everyone, built on entities, context and multimodal signals rather than a single universal position. “The biggest change in digital discovery isn’t simply that AI generates answers. It’s that those answers are personalised for individual users,” the piece argues, and notes that people are twice as likely to click through to a source flagged as one of their “Preferred Sources.”
For marketing leaders, the read is that link building doesn’t need reinventing, but it does need redirecting. The channels that earn citations in an AI-search world (Reddit threads, entity-rich content, Preferred Source relationships) aren’t necessarily the same ones that topped a traditional backlink audit two years ago.
Why this matters:
- Nearly 70% of marketers still rate backlinks as having a “big impact” on rankings, so digital PR budgets don’t need cutting.
- Reddit alone accounts for 3% of all AI citations, meaning community platforms deserve a real line in the outreach plan.
- Being flagged as a user’s “Preferred Source” makes them twice as likely to click through, a new metric worth tracking alongside rank.
Read more: How AI search is reshaping link building strategy, Search Engine Land, 20 July 2026
Also worth your time this week:
Paid Media & Performance: Google Just Took Away Your Opt-Out On Local Inventory Ads
From 31 August, Google will turn on Local Inventory Ads by default for every Shopping campaign, and advertisers will lose the ability to disable it through the Google Ads API entirely. Search Engine Land’s coverage confirms the mechanics: “Advertisers using Shopping campaigns will no longer be able to disable Local Inventory Ads through the Google Ads API,” and API versions v25.1 and above will actively error if a developer tries to set the local setting to false. Anyone managing Shopping campaigns programmatically, whether in-house or through an agency, needs to move to the Inventory filter to retain any granular control at all, and that update needs to land before the end of August, not after.
The rest of the pillar this week was less about forced change and more about better questions. A Search Engine Land piece on B2B PPC argued that lead-volume-only reporting is actively misleading advertisers with long sales cycles: “the longer the sales cycle, the less lead volume tells you,” and pointedly, “a campaign that generates 100 low-quality leads can look better in a dashboard than one that generates 15 highly qualified prospects.” The fix is straightforward but under-adopted: feed offline conversion data, qualified lead, sales-qualified lead, opportunity created, back into Google Ads so automation optimises for pipeline rather than raw form fills. As the piece puts it, “sales feedback shouldn’t be treated as anecdotal. It’s performance data.”
Google also promoted its “Missed Growth Opportunity” tool from a standalone Labs feature into the main Recommendations tab, giving advertisers a clearer, more prominent view of budget and bid constraints. As Search Engine Land notes, “Google is giving advertisers a more prominent way to evaluate the cost of budget and bid constraints,” a small UI change that makes an existing budget-efficiency conversation harder to ignore.
Why this matters:
- Local Inventory Ads become mandatory for Shopping campaigns from 31 August, so API integrations need updating now, not in September.
- Lead-volume reporting alone misleads long-sales-cycle B2B advertisers, so offline conversion data should be feeding back into Google Ads.
- Google’s promoted Missed Growth tool makes the cost of tight budgets and low bids visible directly in the Recommendations tab.
Read more: Google confirms Local Inventory Ads will be enabled by default, Search Engine Land, 20 July 2026
Also worth your time this week:
Web Development & Design: A Bot-Check Screen Could Be Quietly Replacing Your Page In Google’s Index
Search Engine Journal’s report on a warning from Google’s John Mueller describes a failure mode most technical SEO teams haven’t thought to check for. Bot-verification or “are you a bot” screens served by CDNs and security layers can be indexed by Google instead of the site’s real content, because these interstitial pages often look identical across many different domains. Google’s systems can treat them as duplicates and pick a different site’s identical verification screen as the canonical version, effectively pushing your real page out of the index entirely. Mueller’s framing is precise about why this is dangerous: “the issue is that it’s receiving the wrong content,” not a broken page or a 404 that would show up in a normal crawl error report.
The genuinely awkward part is diagnosis. A human visiting the site sees the real content, because the bot-check screen only triggers for automated requests, which means the standard “just check it in a browser” instinct will tell you everything is fine when it isn’t. Mueller’s advice is to check Search Console’s URL inspection and page indexing tools directly rather than trusting a manual visit, since that is the only place the discrepancy actually shows up.
A second, related Mueller clarification this week covers Search Console’s “Validate Fix” button, an easy feature to misuse. Mueller explained that clicking it prompts Google to check a sample of affected URLs and, if that sample passes, queues the remaining known-affected pages for faster recrawl: “it’s not so much that we wait and see if this is actually working better, but we’ll try to recrawl that a little bit faster.” The important caveat is that it should only be used once every instance of an issue is genuinely fixed, not a single test URL, and is unnecessary for routine issues that resolve through normal recrawl cycles anyway. Elsewhere, a Search Engine Land piece reframed schema markup’s real value as machine-readable infrastructure for AI systems rather than a shortcut to rich snippets: “JSON-LD provides explicit information about entities and relationships that can help machines understand your site. Visibility is a byproduct of being understood.”
Why this matters:
- Identical bot-verification screens across sites can get indexed as duplicates, pushing your real page out of Google’s index entirely.
- The problem is invisible on a normal browser visit, so it needs checking directly in Search Console’s URL inspection tool.
- “Validate Fix” should only be used once every instance of an issue is genuinely resolved, not to test a single fixed URL.
Read more: “Are You A Bot” Screens Can Get Your Pages Dropped By Google, Search Engine Journal, 19 July 2026
Also worth your time this week:
Brand & Connection: AI Search Is Quietly Favouring The Local Plumber Over The National Brand
The Drum’s piece on why national brands are losing AI search where it matters most, drawing on analysis from BrightLocal CEO Myles Anderson, makes a genuinely counterintuitive point: brand recognition, the thing national advertisers have spent decades and budgets building, isn’t what wins AI search recommendations for local intent. LLMs prioritise deep, current, geographically specific expertise over broad brand scale, meaning a well-reviewed independent local business can out-recommend a household name in the moment it actually matters, when someone asks an AI assistant where to go. As Anderson puts it, the models are “looking for deep expertise, authority and sort of genuineness around a certain topic,” not recognisable logos.
There’s a trust gap sitting underneath this shift that marketing leaders should note carefully. 58% of consumers have tried AI tools for local recommendations, and 31% do so monthly, so the behaviour is already mainstream. But only 18% trust AI enough to act without verification, and 82% still check Google afterwards. That means AI recommendation is increasingly the first touch, not the final decision, which changes what “winning” an AI citation is actually worth: it buys you a place in a shortlist that still gets independently verified, not a guaranteed conversion.
Two other Drum pieces this week reinforced the same underlying idea from different angles: cultural specificity and coherent identity earn trust in ways broad reach can’t buy. Esports organisation G2’s limited-edition streetwear collaboration with the One Piece manga, which has sold more than 600 million copies worldwide, sold out on the strength of genuine fan participation rather than licensing scale. As the piece frames it, “reach is rented… participation… is owned,” and “the commercial value isn’t the logo on the hoodie. It’s the trust you earn.” Separately, an opinion piece on Britain’s own “brand” problem used sobering polling to make the same point at a national level: “economic optimism in Great Britain is at its lowest level since Ipsos started measuring it in 1978. Nearly three-quarters of voters believe things are getting worse. Only 8% believe they are getting better,” concluding that “when a country’s identity becomes coherent, brands can borrow against it. When it fractures, there is nothing left to borrow.” The lesson scales down to any brand: coherence has to exist before you can trade on it.
Why this matters:
- AI search favours deep local expertise over brand scale, so a well-reviewed local competitor can out-recommend a national brand.
- 58% of consumers already use AI for local recommendations, but 82% still verify independently on Google afterwards.
- Cultural specificity beats generic scale for trust, whether that’s a streetwear collaboration or a national identity.
Read more: National brands are losing AI search where it matters most, The Drum, 16 July 2026
Also worth your time this week:
AI & Automation: 85% Of ChatGPT Topics Still Have No Clear Owner
A Semrush analysis of 1,094 categories across more than 50,000 brands, covered by Search Engine Land, found that only 15.2% of ChatGPT topic categories currently have a clear “owner” brand. A further 31.2% have an emerging leader, and 53.7% remain completely unsettled. Branded search volume turned out to be the strongest predictor of ownership, ahead of traditional SEO signals, prompting Growth Memo founder Kevin Indig to note that “traditional SEO metrics aren’t enough to explain who owns a topic. While they play their role, there’s more to it.” For marketing leaders, the practical read is that most of the AI answer landscape in your category is still genuinely unclaimed, which is a rare kind of opportunity in a channel that usually arrives already dominated by incumbents.
Google, meanwhile, kept expanding what AI Mode can actually do inside Search itself. New integrations let US users connect Instacart, Canva and YouTube Music directly, taking actions such as adding items to a cart, generating design templates, or building a playlist without leaving the AI-generated answer. Google says “users can securely connect supported services and interact with them from AI Mode,” with more partner integrations planned. A separate rollout added a live “Top Stories” carousel inside AI Overviews for developing topics on US mobile, surfacing fresh perspectives and preferred sources: “now, for some searches when you have a question about a developing topic, you’ll start seeing a prominent carousel, which will also highlight your Preferred Sources.”
Read together, both stories point the same way. Google is building AI Mode into a genuine destination, not just an answer box, while brand ownership of AI-generated answers in most categories is still up for grabs. The window to establish topic ownership, before it consolidates the way traditional search rankings did, is open right now.
Why this matters:
- Just 15.2% of 1,094 analysed ChatGPT categories have a clear brand owner, leaving most topics winnable.
- Branded search volume predicts AI topic ownership better than traditional SEO metrics do.
- Google AI Mode now supports direct actions through Instacart, Canva and YouTube Music, extending how much a single AI answer can do without a click-through.
Read more: ChatGPT topic ownership is rare, and SEO alone doesn’t explain it, Search Engine Land, 21 July 2026
Also worth your time this week:
Case Study of the Week: 190% More Search Visibility From Fixing The Fundamentals
This week’s theme, that control over visibility comes down to unglamorous fundamentals more than headline-grabbing AI moves, is exactly what played out for KOREC Group, a UK and Ireland supplier of geospatial and surveying equipment that came to The Digital Maze with a website held back by redirect loops, a weak backlink profile, slow page speed and thin, duplicated content.
We fixed the technical foundations first, then strengthened the link profile through content marketing and digital PR, and built a content hub strategy with sharper keyword and category targeting to improve both relevance and user experience.
- +190% increase in search visibility
- +19% increase in organic traffic (Q1 2023 vs Q1 2022)
- +48% increase in goal completions
- -38.5% reduction in bounce rate
Read the full KOREC Group case study.
Bringing It Together
Every story this week is really about the same shift: control over how your brand is found is moving into places most marketing teams aren’t yet looking. It’s in the traffic split between bots and humans that your analytics dashboard doesn’t separate out. It’s in an API setting Google can simply switch off on your behalf. It’s in a CDN security screen that Googlebot might be reading instead of your actual homepage. None of these are content problems. They’re operational ones.
At the same time, the ChatGPT topic-ownership data is the most optimistic story of the week hiding in plain sight. Most categories are still unclaimed, and branded search volume, something well within a marketing team’s control, is the strongest predictor of who wins them. The brands that treat this as a genuine land grab now will be much harder to dislodge once the remaining 53.7% of categories settle.
The practical takeaway isn’t to panic about bots or API changes. It’s to widen what “site health” means on your team’s checklist, past uptime and page speed, to include the things a human visit will never reveal.
Three Takeaways for the Week Ahead
- Stop reading your analytics at face value. With automated traffic now at 57.4% of all web requests, ask your analytics or dev team to separate bot and agent traffic from human sessions before you trust a dashboard number again.
- Get your bot-verification screens checked in Search Console, not just in a browser. A CDN security check served to Googlebot instead of your real page can quietly get your content dropped from the index, and it won’t show up on a normal visit.
- Go after the 53.7% of AI topic categories still unclaimed. Branded search volume, not SEO effort alone, is what’s winning ChatGPT topic ownership right now, and most categories haven’t settled yet.
Want help putting any of this into action?
We work with ambitious brands across SEO, PPC, web development, and content strategy. If this week’s news has raised questions about your technical foundations, your paid media budget, or where your brand stands in AI search, we would like to talk.
Get in touch with The Digital Maze.
Board Analysis is The Digital Maze’s weekly marketing briefing, published every Wednesday. It covers search, paid media, web development, brand, and AI for marketing managers, directors, and leaders who need to stay ahead without wading through every trade publication themselves.
