October saw AI go after everyone’s content. Meta wants publishers’ posts to help train its models, Reach PLC has signed up to Google’s news AI pilot, and AI Overviews now sit on more than 8 in 10 US brand searches. In PPC, Google Ads has given us better control over where budgets go, Meta has told advertisers to stop fiddling with their campaigns, and Google has launched Universal Cart just in time for peak season. This is what you need to know from October to be seen in the busiest weeks of the year.
In this round up:
SEO
1. Google continues to ‘poison’ search results, leading to inaccuracies in rank tracking tools
It’s getting harder for rank tracking tools to provide accurate data, and the latest comments from tools have revealed that Google is quite literally poisoning the well. Major rank tracking platforms have been reporting that Google appears to be deliberately showing altered, or ‘poisoned’, search results to automated scraping tools.
According to Ahrefs’ own statement, between 5% and 20% of tracked search results have been impacted – with results not matching what human users see. Whilst they, and other tools including Semrush and Accuranker, scramble to correct their data and clean up historical metrics, it’s becoming clearer that Google is escalating measures against data collection and AI scraping.
This causes a headache for website owners, with reporting and results monitoring displaying false position drops and spikes that don’t reflect reality. If you’re seeing any weird results, make sure to interrogate them and manually check results in real time before taking any corrective action.
Sources: Search Engine Roundtable, Ahrefs
2. Google wraps up its latest spam update rollout
Google’s search status dashboard confirmed the completion of its September spam update on 8th October. Rolling out over nearly a fortnight after kicking off on 24th September, it marked the longest spam update rollout of the year. While these rollouts typically take aim at low-effort manipulation and scaled automated content, Google kept things broad this time around, giving no specific policy focus or targets.
Any website owners who believe they have been affected by the spam update should compare Search Console data from before, during, and after the update – or send a signal to their trusted SEO experts to dig into the results.
Source: Search Engine Journal
3. Critical malware was found in Rank Math’s latest SEO PRO update
Wordfence issued an urgent advisory after malicious code was found in copies of the latest Rank Math SEO PRO update (version 3.0.123). This was quickly resolved, with Rank Math pulling back the infected version, but updating to the now clean version of the plugin might not automatically resolve any active breaches.
They advise that any website owners who installed the infected software during the following period restore their websites from pre-incident backups, or get their hosting provider to perform a manual cleanup:
- Start: October 15, 2026 at 20:02 UTC
- End: October 16, 2026 at 07:12 UTC
If you downloaded but didn’t install or activate the affected version, you should be safe.
Source: Wordfence
4. Google clamps down on fake authors
Over the past month, Google rolled out updates to its Search Central documentation, including a warning against creating fake authors and team members in an attempt to boost credibility. Website owners have been explicitly told not to use:
- AI-generated headshots
- Fabricated credentials
- Falsified author bios
For years, publishers have used fake personas as a shortcut to game E-E-A-T signals. Google has now made it clear this is deceptive behaviour, and is a direct indicator of a low-quality page. Get caught doing this and your whole domain’s visibility is likely to suffer – there are no shortcuts to brand credibility.
Sources: Search Engine Roundtable
Digital PR
1. Publishers urged to delete Facebook and Instagram accounts over T&Cs change
Meta has updated its terms so that, by using its services, publishers agree to Meta scraping their content to “develop, improve, research, train and maintain” its products, AI included. Publishers who don’t agree have until 30th October to stop using all of Meta’s platforms, including Facebook Pages, Instagram business accounts and WhatsApp Business accounts.
James Rosewell, director of campaign group Movement for an Open Web, called it a “shameless land grab” and the worst kind of tech platform bullying. The group is urging publishers to update their website terms to prohibit all scraping, and to close their Meta corporate and business accounts together by the same deadline.
For PRs, it’s a squeeze either way. Publishers who stay are effectively agreeing to the change, so anything they publish (including the stories we provide) could end up training Meta’s AI. Publishers who leave lose reach and ad revenue, and might even change the stories they cover in certain niches, as social audience sentiment has a big pull on what journalists think is relevant.
It’s too early to say what publishers will decide, or whether Meta will back down, but we’ll be watching. It could change how some publications approach promoting their content and selling advertising.
Source: Press Gazette
2. Reach PLC and Google announce News AI partnership
Staying on the topic of AI scraping news content, Reach PlC has announced an AI partnership with Google, giving it enhanced content rights for Reach news stories across the UK and Ireland. It’s part of Google’s news AI pilot programme, where Google pays publishers for use of their content in “experimental AI features”. Nobody is saying what those features are yet, but expect more publisher content in GEO search snippets and in Gemini’s citations.
Reach is one of the UK’s biggest news publishers, so this will be felt across a big chunk of the media landscape. Our stories now need to work harder than just earning clicks and social engagement. Where it makes sense, build in something LLMs can use, like a clear answer to a specific question, so the content is easier to cite.
It’s early days, but PRs should put their SEO hats on and watch how Gemini uses Reach’s content: which types of stories get cited, how often, and how much.
Source: Reach Plc
3. AI answer engines bypass local news outlets, going straight to official sources
Research from AI search visibility specialists Searchable found that ChatGPT cites national newspapers almost as much as local ones (18.8% vs 21.2%), even for queries relating to local matters. The tests covered where to find local news, crime, planning, housing, roads, council decisions and local events. Only one in seven searches pointed to a local provider, while one in three went to an official council, police force or other official body website. Of the top 20 news sources ChatGPT cited for local queries, just one was a local news title (Wales Online).
It comes as Ofcom’s 2026 News Report found that one in five people now use AI apps as a news source, rising to 37% of 25-34-year-olds.
ChatGPT breaks each news question into several searches behind the scenes, then builds its answer from the pages that come up. National news and official body sites are clearly the ones it prioritises and trusts for local answers.
Don’t write off local press just yet. National news may be ChatGPT’s favourite, but local titles still hold their own in its citations, so both matter for campaign reach and visibility. A regional angle also earns you hyper-relevance with your audience and a wider net of publications to pitch.
Source: Press Gazette
PPC
1. Google Ads introduces centralised dashboard for tracking policy issues
Google Ads has launched a Policy Summary dashboard (Admin > Policy > Summary), a single place to find, review and manage policy issues across a whole account. It splits problems into account-level violations and individual ad or asset group restrictions, lets you address or dismiss them in one go, and can sort violations by cost, conversions and ROAS. The legacy Policy Manager (Tools > Troubleshooting > Policy Manager) isn’t going anywhere yet.
Agencies can fix flags on top revenue-driving ads first instead of wading through hundreds of low-impact alerts, catch disapprovals on big-spending campaigns sooner, and spot the silent asset disapprovals that tend to hit Performance Max asset groups. It also saves clicking into every campaign when auditing client accounts.
We’re not convinced it’ll change much though. It’s tucked away in the admin section, which we rarely see day to day, and big issues tend to land one at a time rather than in a pile. If you want to keep an eye on problems, we’d recommend scripts that monitor the account and email you when something goes wrong.
Source: Search Engine Land
2. Google confirms AI-generated review summaries in Search ads
Google has confirmed it’s adding AI-generated summaries of positive customer reviews to eligible text Search ads. It pulls from verified customer feedback and turns the good stuff into short commentary within the ad. It’s live in English in the US, UK, Australia and India, across all devices and ad positions on Google.com. Sensitive verticals like pharmaceuticals and healthcare are excluded.
Social proof right at the point of intent means more credibility and more space in the ad, and annotations like this have historically lifted click-through rates. The catch is that negative or outdated reviews now shape what Google’s AI shouts about, so PPC teams will need to work closer with clients’ reputation management teams. Copywriters also lose some control over the story buyers see, as Google writes these automatically.
We think this is a big deal and could improve performance considerably. Our one worry is that it favours bigger businesses, as smaller ones may struggle to collect enough verified reviews (those that come via a follow-up email after a purchase) to qualify. Overall, a great change.
Source: Search Engine Land
3. Google Ads expands Smart Bidding Exploration to Performance Max and Shopping
Smart Bidding Exploration, which used to be Search only, now covers Performance Max and standard Shopping campaigns. It bids on relevant queries the algorithm previously thought were too unlikely to convert, to pick up incremental conversions. To qualify, campaigns need Target ROAS bidding and no daily budget constraints. Google says it’s an auction expansion layer rather than a new way of targeting.
It should help mature tROAS campaigns that have hit a ceiling and stopped finding new converting query themes, and give retailers extra reach at key shopping moments. The trade-offs: budgets need to stay unconstrained, so keep a close eye on spend, and clients should expect some wobbles in ROAS while the system finds its feet in unfamiliar auctions.
We’ve noticed that mature pMax campaigns tend to settle on a handful of search terms once they learn those terms convert, brand terms included. This is a welcome change, as we can use pMax to scale accounts with extended reach and cut any search terms that don’t look viable.
Source: Search Engine Land
4. Google Ads extends Demand Gen cost adjustments from 3 weeks to 60 days
Google has stretched its automatic cost adjustment programme for Demand Gen campaigns from 3 weeks to 60 days. If conversion performance falls well short of targets during the ramp-up, Google gives you cost credits or adjustments. It only applies to eligible accounts running Demand Gen with Target CPA or Target ROAS, and Google applies it automatically.
Demand Gen sits on visually rich, upper-funnel surfaces (YouTube Shorts, Discover, Gmail), so a 60-day safety net makes testing brand-new creative less risky and experimental budgets easier to sell to nervous clients. High-ticket or longer-sales-cycle products also get more time to settle, and ROAS targets get some protection during early scaling.
Even so, Demand Gen is rarely used at Dark Horse and has some stigma in the industry. Because it concentrates on upper-funnel traffic, it can really drag down ROAS. Anything that tames its hit-and-miss nature is welcome, but this may not be enough.
Source: Search Engine Land
5. Google Ads introduces channel prioritisation and budget allocation controls
Google has added controls across Search, Shopping, Performance Max and Demand Gen that let you rank channels as “High” or “Low” priority and cap the share of budget going to lower-priority inventory like Search Partners or the Google Display Network. High-priority networks get funded first, and the rest only get what’s left. By February 2027, the old spend allocation setups for Search campaigns opting into Display will be deprecated in favour of this, so existing network configurations will need moving over.
You can keep Search Partners and Display on for cheap incremental reach without them eating core search budget, pMax buyers get a lever they’ve long asked for to steer spend towards core Google channels, and high-intent search queries stay fully funded when auctions get competitive.
This is the best change Google has made in quite some time. We find Search Partners and Display hit and miss, and this lets us keep testing them instead of cutting them altogether, which plenty of managers do before the channels get a chance. Now we can set them to low priority and watch the results.
Source: Search Engine Land
Paid Social
1. Meta says advertisers should let campaigns gather data
Meta says advertisers react to ups and downs too quickly, leaving its system too little time to learn who to show ads to and producing confusing results. Its rule of thumb is about fifty sales or sign-ups per ad group each week. Below that, its own analysis found each purchase cost around 28% more. Its advice is to set clear goals, give campaigns enough time and data, judge them against what you set out to achieve, and run proper tests. Changing things without enough data, it says, is just guesswork.
Check your budget isn’t spread too thinly. Lots of small ad groups may each get too few results to learn from, which could be pushing your costs up. Agree ground rules before busy periods like Black Friday as well: how long a campaign runs, and how much data it needs, before anyone touches it, with clients and managers signed up. It stops panicked edits after a bad day or two, which can restart learning and make things worse.
Take the “leave it alone” advice with a pinch of salt though. It suits Meta, because it keeps your money flowing through its system. Set clear limits on what you’ll pay per result and on total spend, and use tests to check your ads are really driving extra sales.
Source: Social Media Today
2. Meta adds individual performance data for ad creative
Meta’s Ads Manager has widened its “Breakdown” feature, which shows how each part of an ad is performing. You can now see how each image or video does on its own, and how each version of copy or headline does. There are gaps: carousel and collection formats aren’t included, and some custom conversion goals, like donations and subscriptions, aren’t covered. The article notes it fits a wider pattern of Meta wanting advertisers to lean on its AI to decide who sees their ads, while removing some manual targeting options.
You can see which parts of an ad are doing the work instead of judging it as one lump. That makes it easier to drop weak images or copy, back the strong ones, and give design and writing teams feedback based on real numbers.
With fewer manual targeting controls, creative is now your main lever, so the quality and variety of what you give Meta matters more. Test genuinely different ideas and angles, and work closely with whoever makes the creative.
Source: Social Media Today
3. Meta introduces new AI-powered tools at Advertising Week
Meta’s AI business assistant is getting a lot more capable. It will remember what you tried in past quarters, your busy seasons and your goals, and you can upload your strategy and brand voice guidelines so its suggestions fit your business. There are more AI video tools too: you can turn existing images into video ads, and the Ads Creative Studio, which uses performance data to show what’s working and suggest what to try, is opening up to more brands.
Day-to-day campaign work could get quicker and more conversational, with drafting ads, tweaking targeting or adjusting budgets done by asking the assistant. The more it knows about your business the better it gets, so write down your brand voice, goals and what has worked before in a form you can share with it.
Don’t trust it blindly though. Many of these features are new or still in testing, and what they produce can be hit and miss. Check everything it creates or changes before it goes live, including wording, images, video, targeting and budget.
Source: Social Media Today
AI Search
1. AI Overviews now appear on more than 8 in 10 brand searches
DemandSphere data shows AI Overviews on 26% of the branded queries it tracks on 1st September, rising to more than 80% by the end of the month. Google hasn’t announced or explained the change. SE Ranking checked 1,000 US brand searches in October and found an AI Overview on 82.5% of them, against 40.4% in Germany, so the rollout isn’t the same everywhere. Of the brand pages cited in those AI Overviews, 62% don’t rank in the organic top 10 for the same search, with product and About pages cited far more often than they rank.
Commercial searches have followed. Semrush found AI Overviews on 88% of 400 commercial keywords (reviews, pricing, alternatives, brand vs brand and best X for Y) in August, up from 46% in September 2025.
Branded search used to be the most predictable part of the results page. Now Google summarises your brand above your own listing, drawing on your site alongside reviews, YouTube, competitors and other third-party pages. Steady branded rankings no longer mean steady branded clicks, so a drop in brand CTR since late September might be the AI Overview, not a ranking problem. Search your brand name and main commercial terms while signed out, and note what the AI Overview says, which pages it cites and whether anything is wrong or out of date.
We’d treat About, product, pricing and support pages as brand pages, because Google is quoting them. Track AI Overview presence and wording on branded terms alongside rankings, and check each market separately, as UK results may not match the US.
Source: SE Ranking, Semrush, Semrush
2. New research suggests ChatGPT sends a tiny share of Google’s traffic
Marie Haynes analysed 12 months of GA4 data from 100 websites and 145 million sessions. ChatGPT sent just 0.62% of the sessions Google organic did, and all AI assistants together made up 0.73% of search and AI traffic. ChatGPT visitors converted at 13.8% against 3.02% for Google organic, but Google still drove 96.5% of the conversions in the channels she looked at.
Separately, Kevin Indig compared AI visibility with later branded search and Direct traffic across 632 tests and couldn’t find a reliable link between them.
On referral clicks alone, ChatGPT is a small channel for most businesses, and it’ll look that way in GA4 for some time. But clicks were never the main way AI search affects a brand. The biggest AI search surface is Google itself, where AI Overviews now sit on most brand and commercial searches (see above). So GEO belongs inside your SEO, digital PR and content plans, not in competition with them for budget.
We’d measure AI search as an influence channel: report AI visibility and how accurately you’re described alongside clicks and conversions, and ask new customers where they first heard of you.
Source: Marie Haynes, Growth Memo
3. Google adds more links to AI Mode and AI Overviews, but still no click data
On 15th October Google announced more inline links in AI Overviews and AI Mode, placed higher up in answers more often. It’s testing a section at the top of some answers showing the one or two official pages you need to complete a task, with a line on why each link is useful. News carousels for timely topics are rolling out to more countries and languages and will highlight a searcher’s Preferred Sources, whose use Google says has grown more than 7x since the start of the year. Related article links now get image thumbnails, and AI Mode is testing clickable creator names on video results. Separately, some AI Overview citation cards have started showing AI-written summaries instead of text from the cited page.
That new task section is prime real estate for any brand that owns a task, like booking, claiming, applying, cancelling or tracking an order. If the right page for a task isn’t clear, Google might surface a third-party page or an old URL on your own site instead. And with Google rewriting how some pages are described in citations, the opening lines of key pages need to say plainly what the page is for.
List the tasks customers come to you to complete and give each one a single clear, crawlable, current page. Search Console’s AI performance reports still show no click data, so judge the effect on landing page clicks and conversions.
Source: Google, Search Engine Roundtable, Search Engine Roundtable
4. GPT-6 makes ChatGPT answers interactive, and a bug hid its traffic for a week
OpenAI started rolling out GPT-6 in ChatGPT on 7th October, reaching Free and Go users from 8th October, and puts ChatGPT at more than 1.2 billion weekly users. It comes with Intelligent UI, which builds charts, buttons, forms and small working tools such as calculators into answers.
From around 8th October, many ChatGPT links lost their utm_source=chatgpt.com tag and arrived with no referrer, so analytics tools logged those visits as Direct. Adobe’s Cedric Huesler saw ChatGPT visits drop by about 31,000 a day and Direct rise by about 26,000 a day on the same 500 pages. He says OpenAI confirmed it was a bug, and a fix began rolling out on 15th October.
Expect October reports to show a dip in ChatGPT traffic and a matching rise in Direct between roughly 8th and 15th October. That’s a tracking gap, not lost visibility. It also shows how much AI referral reporting relies on a tag OpenAI controls and can change without notice, and most ChatGPT clicks still carry no referrer. Intelligent UI is ChatGPT’s answer to the generative UI Google added to AI Overviews in August: simple tools, calculators and comparisons can now be answered inside the chat, so pages built around them need a reason to click.
Annotate 8th to 15th October in GA4 and client reports, and compare ChatGPT and Direct on the same landing pages before calling any change in AI traffic. Pair GA4 with AI visibility tracking so one missing tag doesn’t skew what you report.
Source: Search Engine Roundtable, OpenAI
5. Universal Cart brings checkout into AI Mode ahead of peak season
Google launched Universal Cart in the US on 15th October across Search (AI Mode included) and the Gemini app. Shoppers can save products from different retailers in one cart, get alerts when prices drop or items return to stock, see relevant promo codes, and check out with Google Pay or move everything to the retailer’s own cart. Checkout runs on the Universal Commerce Protocol (UCP) and the retailer stays seller of record. Launch merchants include Nike, Sephora, Target, Ulta Beauty, Walmart, Wayfair and many Shopify stores. Adobe expects AI traffic to US retail sites to rise 130% year on year this holiday season, with the biggest jump on Thanksgiving (159%).
The cart runs on Google’s Shopping Graph, so accurate feed data (price, stock, variants and promotions) decides whether your products get saved, tracked and bought. Shoppers can also buy without ever visiting your site, so there are fewer chances to cross-sell or capture first-party data, and product data and reviews have to do the selling. It’s US only for now and there’s no UK date, but it shows where Google Shopping is heading, and this holiday season is its first big test.
Use Q4 to get Merchant Center feeds and promotions accurate, and ask your ecommerce platform about UCP readiness (many Shopify stores are already launch merchants). Track AI-referred traffic and orders separately over Black Friday so you can see what AI shopping actually drives.
Source: Google, Business Wire
Rankings slipping, drowned out in the AI answers or outbid heading into busy season? Get in touch and we’ll help you sort it.