The content marketing ROI report Google never wanted to build

What Google actually turned on, and why it was never a gift#

A fox does not thank the farmer for a fence it was forced to build; it simply notes the fresh gap and starts using it before the frost sets in. That is roughly the correct posture for content marketers this week, because on 31 August 2026 Google finished rolling out two new Search Console features to every single site on the internet: a dedicated Generative AI performance report, and a toggle that lets a property opt out of Google's generative AI search features altogether, as Search Engine Land reported the same day.

The report itself is almost defiantly dull: no clicks, no queries, simply impressions of a page's links inside an ai overview or ai mode answer, sliced by page, country, date and device, exactly as Google's own help documentation confirms. Discover gets a twin report of its own, per a companion support page, because Discover's generative feed is a different beast entirely.

Beside the report sits a switch. Under Settings, then Search generative AI, a property can choose to be excluded from ai overview, ai mode and Discover's generative features in one single move, and Google's help page is blunt about the trade: opt out and a site forfeits every impression those three surfaces would ever have sent it, though the setting 'isn't used as a ranking or inclusion signal' anywhere else in Search, which is exactly the kind of caveat a content marketing roi conversation needs written down before anyone gets excited.

None of this was Google's idea, whatever the announcement's cheerful framing suggests. On the very same day, Google's own blog post admitted to 'engaging with regulators like the UK's Competition and Markets Authority', and that one clause is doing a great deal of quiet work.

The order behind it, published that same afternoon, told Google to hand publishers 'clear and detailed metrics on user engagement with their search content in search generative ai features', in the dry language of the CMA's own conduct requirement. Sarah Cardell, the regulator's chief executive, called the wider package a 'world-first requirement' in the CMA's announcement, and for once nobody in marketing had to translate a legal ruling into content marketing kpis: the regulator basically wrote the brief itself.

The content marketing roi gap this report was built to close#

Ask any in-house content lead what keeps them up past midnight and the honest answer is rarely a keyword ranking, it is content marketing roi: the maddening task of proving a blog post earns its keep once the click stops being the one unit everyone in the building agrees on. Content marketing metrics built for a world of ten blue links quietly stopped describing the world content teams actually work in sometime around 2024, and nobody replaced them.

Seer Interactive's research and development team tracked the damage properly: 2.43 billion organic impressions, 5.47 million queries, 53 brands, fourteen months of data running from January 2025 to February 2026, published in their April 2026 report. Organic click-through on queries carrying an ai overview fell from 3.19 per cent to a floor of 1.31 per cent by December 2025, a collapse that would sink any content marketing roi model still standing on the old assumption that impressions and clicks move together.

The floor was not permanent. By February 2026 that same click-through rate had climbed back to 2.36 per cent, per the same Seer Interactive data, while queries carrying no ai overview at all rose steadily from 2.75 to 3.82 per cent across the identical window. Citation still pays: Seer measured 120 per cent more organic clicks per impression for a page cited inside the summary than an uncited page on the same query, though even a citation still trails a plain organic result by 38 per cent.

That chart is precisely the evidence content marketing roi reporting has lacked for two years: a real trend, named brands, a stated method, rather than one vendor's cherry-picked screenshot pinned to a slide. It is also exactly the shape of measurement the new Search Console tab still cannot give you, because Google ships the impression half of the sum and quietly keeps the click half to itself.

So call the new report what it plainly is: an honest, half-finished gift. It will not hand any content team a finished content marketing roi model on its own, but it hands every content team one number Search Console never offered before, which page is actually being read inside an ai overview, and that is not nothing, as folkfox has been arguing since the first wave of AI Overview changes landed.

Inside the report: what a google ai overview actually counts#

Open Search Console's performance report today and a fourth tab now sits beside the three that were already familiar: Generative AI. Filter it and four dimensions appear, per Google's own documentation: which pages showed up, which countries saw them, which dates, and which devices, all counted purely as impressions inside an ai overview or ai mode answer. A google ai overview citation and an ai mode citation currently share one tab; Discover's generative feed earns a separate report of its own, per a twin help page.

What the report withholds matters rather more than what it gives. No clicks, no queries, no honest way yet to build content marketing kpis that tie a google ai overview appearance directly to a conversion. Google's Search Central announcement frames the report as a first step, and its wider guidance on ai features is equally clear there is no special trick: standard search fundamentals still decide who gets picked.

A content marketing roi conversation that starts from zero verified impressions data is worse than one that starts from impressions alone, so treat the new tab as a floor to build from rather than a ceiling worth complaining about. Google's own documentation is careful to warn that the report can appear 'even without sufficient impression data to display', a hedge worth reading twice before assuming a quiet dashboard means a broken one.

None of this replaces proper generative engine work, it simply gives that work a scoreboard for the first time, and folkfox's own SEO and GEO practice has spent the last year building pages that survive being quoted rather than merely being ranked, and the new tab is the first place that effort finally shows up in the content marketing metrics you already track, as a number instead of a hunch.

The opt-out switch, and the catch nobody is advertising#

Reaction on the day was fast, and mostly wary. One AEO consultant put the stakes plainly on X within hours of the worldwide rollout:

That last line is the one worth sitting with. The toggle is not a ranking signal, Google says so directly, and there is no evidence it lies about that particular claim. What it decides is narrower and, for a lot of sites, larger: whether a page can ever be the one an ai overview quotes, full stop.

A weak content marketing roi number is not, on its own, a reason to flip the switch. Opting out does not fix a page that fails to earn a citation, it simply guarantees the page will never earn one, which is a stranger kind of failure to explain to a chief marketing officer at the next quarterly review.

There is a second, smaller cost too, one folkfox flagged in August, when an earlier opt-out mechanism carried Top Stories placement out of the door along with everything else. The new toggle is broader, and the same caution applies: read what you are actually switching off before you switch it off.

How a content team should actually use this, starting Monday#

None of the above matters if the new tab sits unopened next to the fifty other browser tabs already fighting for a Monday morning. Here is the routine folkfox runs with clients the week a data source like this appears, condensed into four honest phases.

The point of the routine is not to produce a single, tidy content marketing roi number by Friday, because that number does not exist yet and anyone selling it to you is guessing. The point is to stop guessing in the dark and start guessing with a torch, which is a smaller claim and a much more defensible one in a budget meeting.

This is also, frankly, the argument for treating content marketing and brand strategy as one job rather than two. A page that only exists to rank cannot survive being paraphrased, but a page that carries a distinct point of view, in a voice nobody else owns, tends to be the one a generative system reaches for when it needs a sentence to borrow. That is not a new idea. It is simply, for the first time, a measurable one.

The comparison that finally makes this legible to a finance team is the one paid search solved years ago: a channel is worth what it returns, tracked over time, against a stated cost, not what it feels like it should be worth. Content marketing roi has never had that discipline applied to it properly, because the data was never there to apply it to. As of 31 August 2026, a first, imperfect slice of it is.

None of this needs a moonlit chase through the undergrowth to execute. Follow the scent of your own impressions data, track it like a fox reading a trail rather than guessing at a quarry, and the content marketing roi story starts writing itself by the time the next report lands. The regulator built the den; folkfox just teaches clients how to hunt properly inside it.

Frequently asked questions#