Here’s what stood out this month, how it’s reshaping the SEO and GEO landscape, and some thoughts, opinions and guidance to help you navigate.
Google’s June 2025 Core Update officially began rolling out on June 30. The rollout is set to take three weeks, concluding around July 20. So far, the volatility appears consistent with prior core updates.
My take: Google has offered very little detail, which is generally a good thing with Core Updates. It means it's a standard, broad update where the goal is to reward “quality content”, not to impose penalties or punish sites. It's also not introducing new signals. This may seem academic because sites with pages which lack substance, originality, or EEAT (experience, expertise, authority and trust) may still feel the pinch – and that pinch won't be numbed by the knowledge that you haven't technically been punished; others have been rewarded. The impact is the same either way. However, it does (or at least should) skew the approach to finding a route forward, from knee-jerk, remedial panic to a more proactive and holistic thought process around quality improvement.
Google Search Console is now reporting impressions, clicks, and position data for AI Mode users (which is only fully live in the US and India right now, but will be hitting the UK and the rest of the world very soon, after some regulatory niggles are ironed out, perhaps as soon as Q3 2025). Unfortunately, AI Mode reporting data is lumped in with overall search performance data. There’s no filter yet to split out AI Mode traffic or stats from traditional search data, so at best you can assess the overall impact of AI Mode on your organic visibility as a whole, but you still can't create precise reports or extract insights and action plans just for AI Mode.
My take: The best thing you can do is keep an eye on impressions. Recent studies suggest impression trends remain largely unaffected by AI, which makes sense when you think about it. Roughly the same number of people are searching in Google (i.e. the erosion of Google's market share from AI experiences is negligible in the grand scheme of things) so the scope of opportunity for impressions is the same. Clicks have become the moving goalpost – they have been gradually dropping since the introduction of AI Overviews (May 2024 in the US, then August 2024 in the UK), sometimes levelling-off, sometimes increasing momentarily as Google tests new features, but definitely losing fidelity as a reliable benchmark/baseline metric for period-over-period analysis; and now AI Mode is further reduce clicks and adding more complication to forecasting.
From an SEO analyst’s perspective, impressions remain the cleanest, most stable metric for benchmarking and gauging whether or not your SEO footprint is expanding, holding steady, or declining in the new AI-infused Google Search era. It's still not perfect because AI surfaces encourage people to make more searches, potentially leading to more impressions (with a lower CTR)... but impressions are at least at the top of the reporting funnel and therefore require the least amount of 'reading between the lines', correlating most closely with the combined reality downstream of search interest and visibility.
Google and Bing both dialled up testing of richer AI integrations this month:
My take: Collectively, these changes signal a further shift towards folding AI deeper into search behaviours across image, video, audio... not just text/LLM input. The endgame looks increasingly multimodal. Thinking harder about your brand assets now – quality, formats, optimisation – could pay dividends in the next phase of search's evolution.
Instagram has confirmed plans to make its content indexable in search engines in July 2025. Details remain thin, but it’s a safe bet that standard social SEO principles will apply when optimising for Google visibility: keyword usage in profiles, captions, hashtags, and engagement signals (in Instagram and Google) will all likely shape what ranks well in search. This was certainly true of Twitter, for example, and it's how Google looks at it's own platform, YouTube, when surfacing content in Google Search.
My take: Given the platform’s influence and user base, this could become a notable organic channel in the months ahead. Time will tell just how heavily these SERP Features are promoted and engaged with. This could prove to be nothing more than an a great bit of PR for Instagram.
Slipping on the tin-foil hat for a moment... the last time Google opened the floodgates to a UGC platform like this, it was due to a highly criticised deal (words like 'backroom' and 'shady' have been bandied around) brokered with Reddit. Reddit was promoted in search and Google got to train its AI on all of Reddit's user-generated content. If AI search is becoming more multimodal, perhaps this 'opening up' of Instagram in Google is the consumer-facing end of the deal ushering in the next wave of AI training data acquisition by Google.
In a significant shift, Cloudflare announced that all new accounts will now block AI crawlers by default. Anyone wanting AI bots to scrape their content must explicitly opt in using a new settings toggle.
Given Cloudflare’s scale as the largest web application firewall, this move could influence negotiations between site owners (as a collective) and AI companies seeking training data. It’s certainly the clearest sign yet that the web is fragmenting into camps: those who want to be part of AI’s knowledge base, and those who don’t. It could be empowering, but only if enough websites opt-out, otherwise there simply won't be enough leverage.
My take: From a visibility standpoint, it’s simple: if you block AI bots, you’re opting out of being surfaced in AI-generated answers. Whether that’s a win or a loss depends on your business model and content strategy. Our job is to make brands more visible in search, including generative AI search, so of course we think it would be a bad idea to block AI crawlers. Whether you choose to take a more principled approach, though, is entirely up to you!
Interestingly, somebody asked Google's spokesperson, John Mueller, if they should use the LLMs.txt file (a proposed new standard for controlling AI crawlers, similar to robots.txt but providing context alongside crawl priorities and rules) and his answer was "Why bother? No LLMs respect LLMs.txt anyway!". Cloudflare appears to be taking the proverbial bull by the horns and wrestling AI crawl control into the hands of the average site owner, giving them leverage and choice. On that level alone, you have to hand it to them for being proactive and principled; and for having the courage to give some of their Silicon Valley neighbours a bloody nose. Of course it's also a great PR stunt, so there is that.
If you have any thoughts or questions, or would like to discuss how we can help you to optimise in light of these changes, please reach out!