Another month, another pile of updates designed to sound helpful while quietly shifting the goalposts.
But look a little closer at August, and the through-line is unmistakable: the platforms are taking away your controls, and the AI systems are changing faster than any dashboard can show you.
Here's the August digital marketing industry briefing — with the fluff cut out.
Google is sabotaging your tracking software
Google is rolling out
/goto
passthrough URLs in search results. Instead of a click going straight to the destination, it routes through
google.com/goto?url=...
first — and those links can't be decoded.
Google's official line? Protecting against "evolving forms of abuse". Which in practice means AI companies and third-party tools can no longer scrape search results the way they used to
Why you should care: keyword tracking across platforms is now unstable. The resources needed to follow those redirects could push up prices for the AI and SEO tracking software you rely on. Watch your tools closely. If something looks off, this is probably why.
The spam update no one can stop talking about
A short, "nothing notable" spam update rolled out from the 18th to the 21st — the third announced one of 2026. No new policies. Sounds boring, right? Except SEOs are still reporting volatility after it finished.
Here's the translation: the noise isn't over because the update rolled out. If your rankings have been swinging, check you're compliant with existing spam policies, then buckle up — this one keeps giving.
Claude just made AI slop a lot easier to spot
Anthropic signed the EU AI Act's transparency code in July and jumped into action by early August. As of the 2nd, Claude now embeds machine-readable watermarks and provenance metadata into AI-generated content.
The kicker? Anthropic is working with third parties to help them detect these watermarks. That means Google and other engines will likely soon be able to see where content was generated or passed through AI.
Plenty of people use Claude responsibly to refine and edit. But the door has opened on enabling platforms to spot pure AI slop at scale. Big win for transparency — and a big slap for anyone publishing unedited AI-generated filler as if it were authored.
AI Mode is quietly polluting your Search Console data
This one's sneaky. Follow-up queries typed into AI Mode — things like "yes," "no," "tell me more" — are being tracked in Google Search Console as brand new queries. And there's no baked-in filter to remove them.
There's a manual fix. Head into the Performance report, click +Add Filter > Query > Custom (regex), and experiment.
- To find long-tail conversational queries (10+ words):
^(?:\S+\s+){9,}\S+$ - To filter out one-word follow-ups: ^(?!yes$|yeah$|yep$|yup$|no$|nope$|okay$|ok$|sure$|thanks$|thank you$|please$|great$|perfect$|exactly$|right$|correct$).+$
Careful with the second one though — filter out all single-word replies, and you'll also delete data you actually want. Tune it to your own language.
Digital PR: Reach is quietly rewriting the rules of syndication
After Reach said last month it would prioritise quality over quantity, PRs braced for less syndication. Turns out the worry was misplaced — Reach is still publishing across titles. But here's the twist: single stories are going out across multiple publications without a canonical tag, with slight wording tweaks to dodge duplicate-content flags.
Historically, that canonical tag was what marked a piece as syndicated. Without it, an article can look like a standalone piece to search engines. PRs get the visibility benefit of one-coverage-many-places, without it being marked as such — but the wording shifts suggest Reach is prioritising localisation and audience tailoring, not straight copy-paste.
The lesson: report these results honestly and transparently to clients, whichever way you choose to label the coverage. This new distribution model may well become the norm. Know what you're delivering.
The heatwave story has gotten darker
Another August, another heatwave — and another battery of reactive pitches. But the tone has shifted. The gap for reactives is moving from light-hearted "stay cool / sleep / exercise" tips toward serious angles: rising anxiety in younger generations, social inequality and health risks.
The message is blunt, and it applies to every reactive story, not just the weather: just because you can pitch it, doesn't mean you should. Pushing a flippant product plug during a genuine climate and health crisis reads as tone-deaf. Consider the wider narrative before you send.
Google's 'preferred sources' — the rare lever that runs through your audience
Google has added a "preferred source" button, so users can pick a newsbrand they trust and have it prioritised in Top Stories, AI Overviews, and AI Mode. And crucially — this is one of the very few AI search visibility levers driven by your readers, not the algorithm.
It won't get clicked on its own. Pair it with an explicit ask in newsletters, on social, in post-purchase emails. And start treating preferred-source count as a reportable metric, not a vanity stat.
PPC: Google is about to throw open your exact match guardrails
Brace for this one. From 1 September 2026, Google is auto-upgrading eligible Search campaigns to 'AI Max for Search' if they use Automatically Created Assets or campaign-level broad match. That switch unhelpfully enables Search Term Matching (query expansion) and Text Customisation (AI ad copy) by default.
Read that again: exact and phrase match keywords will no longer act as strict guardrails. Google's AI can trigger ads on queries far beyond your explicit keyword list, and AI-written headline copy could bypass everything you crafted by hand.
Act now. Either turn off ACA and campaign-level broad match to opt out entirely, or build hard negative keyword lists to constrain Google's AI before your budget leaks on irrelevant queries.
AI Search: hold on, because the ground is moving daily
Three stories in AI Search this month prove how unstable this terrain is.
- Reddit lost 86% of its ChatGPT citation share in under a week — from 3.83% down to 0.52%, with no confirmed cause for the second drop. Any AI strategy resting on one platform is fragile by design. Spread coverage across several credible sources; report AI visibility as a trend across engines, not one monthly figure.
- Google swapped the model under AI Mode with no formal announcement. Gemini 3.7 Flash arrived on the 14th — no core update, no status dashboard, revealed via a Google employee on X. If AI answers shift, it may well be a model change, not a ranking change. Annotate it on your charts exactly as you'd annotate a core update — and rerun your prompt set after every model swap before you draw conclusions.
- Generative UI is hitting AI Overviews. Google is building custom layouts, calculators, and interactive tools on the fly — live globally in English via AI Mode already. Any page ranking mainly because of an embedded tool now competes directly with the results page. Defend clicks with a tool Google can't rebuild in a box — saved results, logged-in data, proprietary inputs — and judge tool pages on mentions and assisted conversions, not sessions.
The paid social squeeze
Meta removed the option to exclude ad placements and platforms — pushing everyone toward expanded placements so its algorithm decides where ads run. Manual placement strategy is largely obsolete; control is shifting to creative and audience inputs. If you were steering spend away from brand-unsafe slots by hand, that lever is gone.
On the flip side, Meta added dedicated exclusion-only audiences — permanent, standalone lists (opted-out customers, employees, restricted groups) that can never be accidentally flipped into an inclusion audience. A genuine win for compliance and account hygiene, especially in regulated industries.
Snapchat has joined the fight against AI slop: only human-made content is now eligible for Spotlight recommendations. TikTok is pushing its content off-phone into billboards and cinema via new out-of-home partners. The direction of travel is unmistakable — platforms want authentic, human content, and they're rewarding it.
The August bottom line
Google is taking away your knobs, hiding its own changes, and letting its AI decide more of your targeting — while deluging your data with noise. That's a lot of "you don't need to worry about it" wrapped around a lot of genuinely worrying shifts.
Don't get swept along. Audit your AI Max campaigns before 1 September. Filter the junk out of your Search Console. Spread your AI visibility across engines instead of betting on one. And treat every silent model swap like the core update no one told you about.