A client called me last month convinced their analytics were broken. Organic clicks in GA4 down a fifth year over year. Rankings in their tracker? Unchanged. Search Console impressions? Flat. No algorithm update they could point to, no lost keywords. Just clicks quietly leaking out of a system that, by every dashboard they trusted, looked perfectly healthy.
This is the first number I show every new client who thinks their traffic dip is a tracking glitch: AI Overviews now cost the German search market 265 million organic clicks every single month. That’s the missing variable. Not a bug in your tags—a structural change in how Google answers questions.
6.6% sounds survivable until you see where it lands. It does not fall evenly. And for B2B companies whose content exists to answer questions, it falls almost entirely on you.
The SISTRIX data, decoded
SISTRIX ran the numbers on the German market in June 2026, and the headline figure—265 million clicks—is the least interesting part. The interesting part is the distribution.
When an AI Overview is present, the click-through rate on the first organic result drops from over 27% to just 11%. That’s a 59% cut to the most valuable position on the page. Across all positions, a typical German search led to an organic click 57% of the time before AI Overviews. With one present, that drops to 33%. Roughly a quarter of the clicks that used to leave Google for someone’s website now never leave at all.
The sector breakdown tells you who pays. Parenting and baby portals lose over 24% of their organic clicks to AI Overviews—informational, YMYL-adjacent content that a generative answer can summarize cleanly. Recipe sites like Chefkoch barely register the hit, losing just over 1%, because nobody reads a three-paragraph AI summary and considers dinner solved; they still click for the actual recipe. Transactional and navigational queries are largely spared for the same reason—Google has no incentive to absorb a click it can monetize.
Even Wikipedia, one of the most-cited sources on the web, is the single largest absolute loser in Germany: SISTRIX estimates 31.6 million clicks gone per month, roughly 5% of its German Google traffic. If being the most-cited source on the internet still costs you 31.6 million clicks, no informational publisher is immune.
Now place B2B informational content on that spectrum. “What is X,” “how does Y work,” “best approach to Z”—the comparison guides, the explainer posts, the buyer-education content that most B2B content strategies are built on. That’s not the recipe end of the curve. That’s the parenting-portal end. With AI Overviews already showing on more than 20% of all German keywords—and that share growing—the question is no longer whether your clicks are being redirected. It’s how many, and on which pages.
Why your dashboard is lying to you
Here’s the mechanism behind the phantom traffic problem. Your rank tracker checks one thing: where your URL sits in the organic results. An AI Overview doesn’t change that. You can hold position one, watch your tracker stay green, and lose more than half your clicks to a box that sits above you and answers the question before the user ever reaches your link.
Impressions and clicks have decoupled. Search Console still counts the impression—your page was eligible, it was in the running. GA4 counts the click that never came. The gap between those two lines is AI Overview interception, and no traditional rank tool will ever show it to you, because the tool is measuring the wrong layer.
Google’s own Search Console Generative AI performance report, which rolled out in June 2026, closes part of this gap. It finally gives site owners impression data for AI Overviews and AI Mode, broken down by page, country, device, and date. For the first time you can ask “did I appear inside the AI answer?” and get a real answer.
But know its limit before you lean on it. The report shows impressions only—no clicks, no CTR from inside the AI answer. It tells you whether you showed up. It cannot tell you whether being cited earned you anything. For that, you need a different number entirely.
The citation multiplier: the data that changes the budget conversation
This is the number that flips the framing. Brands cited inside an AI Overview see a 35% higher organic click-through rate and a 91% higher paid CTR than competitors who appear below the AI answer but aren’t cited in it. The Seer Interactive study behind those figures covered 3,119 search terms across 42 organizations: cited organic CTR of 0.70% versus 0.52% uncited; cited paid CTR of 7.89% versus 4.14%, both on Q3 2025 averages.
Read that twice, because it inverts the entire panic. Being inside the answer is not a consolation prize for losing the click. It is a conversion amplifier. The AI Overview isn’t just taking clicks away from everyone—it’s redistributing them toward the brands it names.
Walk the math for a hypothetical German B2B company with 10,000 monthly impressions on a target query:
- Not cited: 10,000 × 0.52% ≈ 52 organic clicks per month.
- Cited inside the AI Overview: 10,000 × 0.70% ≈ 70 organic clicks per month.
- On the paid side, the same impressions swing from ~414 clicks (4.14%) to ~789 clicks (7.89%)—nearly double.
18 extra organic clicks a month on one query sounds modest. Multiply across a content library of a few hundred indexed pages, each with its own query set, and the citation slot stops being a vanity metric and starts being the difference between a content program that compounds and one that quietly bleeds. SISTRIX names this directly as one of only two real responses available to publishers: lose the click, or—in their words—“optimise to be cited as a source within AI Overviews.” There isn’t a third option where you keep playing the old game and win.
The new measurement stack
Rank position can’t be your primary KPI anymore. It measures a layer that no longer determines your traffic. In the GEO audits I run for German KMU, I map four layers before touching a single piece of content—because you can’t improve what you refuse to measure.
Layer 1—AI impression share
Start with the Search Console Generative AI report. Pull impressions by page, country, and device, and watch which pages are surfacing inside AI features and which are invisible. This is your free, first-party baseline. It won’t show clicks, but it answers the foundational question: are you even in the room?
Layer 2—GEO citation tracking
Third-party tools now monitor whether you’re actually cited as a source across AI surfaces, not just present in search. A few worth knowing, at honest price points:
- Otterly.ai—self-serve monitoring across ChatGPT, Google AI Overviews and AI Mode, Perplexity, Microsoft Copilot, and Gemini. Starts at $29/month with a 14-day trial. A sensible first tool for a KMU.
- Rankscale—tiered by usage credits: Essentials at $17/month (120 credits), Pro at $84/month, Enterprise at $663/month.
- Ahrefs Brand Radar—tracks brand mentions and citations across six AI platforms, starting at $129/month, useful if you already live in Ahrefs.
- OmniSEO and similar GEO suites—heavier platforms for teams managing citation tracking across many queries at once.
Layer 3—brand mention monitoring
Beyond formal citations, track how the assistants describe you when someone asks about your category—Perplexity, ChatGPT, Gemini. Are you named? Described accurately? Mentioned alongside the right competitors, or the wrong ones? This is the reputation layer, and it moves independently of your search rankings. We built Cited for exactly this beat—it tracks how ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews describe your brand and your competitors over time, and turns the gaps into a to-do list. A tracker like that turns “I think we’re invisible in AI search” into a number you can move.
Layer 4—share-of-answer
Pick your top 10 target queries—the ones that actually drive pipeline—and measure how often your brand appears in the AI answer versus your competitors. Share-of-answer is the closest thing GEO has to the old share-of-voice metric, and it’s the one I hand to a marketing director when they ask what “winning” looks like now.
The DACH opportunity window
Here’s the part that should make German B2B founders sit up rather than despair. The DACH region is structurally behind on generative AI—the region’s momentum score runs 8% below the global average, 71% of DACH businesses admit they aren’t moving fast enough, and over half fear a competitive disadvantage because of it. That’s usually framed as a weakness. For the companies reading this, it’s a window.
Three things compound in your favor right now. GEO adoption among DACH competitors is low—most are still optimizing for blue-link rankings that are actively losing CTR share. German-language editorial content is comparatively sparse in the corpora these models lean on, so quality German content is disproportionately valuable to an engine looking for something to cite. And the agencies most German firms hire are still selling position tracking as the deliverable.
Citation slots inside AI answers aren’t infinite. An AI Overview names a handful of sources, not 50. The first movers into those slots own visibility that’s expensive to displace later. In English-language markets that land grab is already underway. In German B2B, the field is still mostly empty—which means the cost of being early has rarely been lower.
Three structural changes that earn AI Overview citations
Distilled from the audit patterns I see repeat across German KMU sites, three things separate the pages that get cited from the ones that don’t.
1. Entity clarity
AI engines cite sources they can resolve into a confident entity. That means a named, real author—not “the team”—backed by Organization and Person schema with sameAs links to LinkedIn and other verifiable profiles, plus consistent brand signals across every platform you appear on. Before: a German consultancy publishing anonymous “editorial” posts with no author and a logo for a byline. After: every post carries a named expert with a linked profile and matching schema. The second site is legible to a model; the first is a blur.
2. Structured depth
Mark up what you publish so the structure is machine-readable: Article schema with honest datePublished and dateModified values, FAQPage markup on product and service pages, and an answer-first paragraph structure that leads with the conclusion instead of burying it under 400 words of preamble. Before: a 2,000-word guide that answers the title question in paragraph nine. After: the same guide answering it in the first two sentences, then earning the rest. AI engines extract the answer from the top—give them somewhere clean to extract from.
3. Editorial authority
This is the one that can’t be faked, and it’s the most durable. Original data, attributed quotes, and first-person observations that an engine cannot synthesize by averaging ten other articles. Before: a roundup post restating what everyone else already wrote—nothing for a model to cite that it couldn’t get elsewhere. After: a post with a number from your own client work, a named source, a “here’s what I keep seeing in the audits I run” that exists nowhere else. Uniqueness is citability. If your content is interchangeable, the engine will cite the original instead of you.
A measurement cadence your marketing team can actually run
You don’t need a data scientist for this. You need a rhythm. Here’s the cadence I leave with KMU clients—print it, hand it to whoever owns marketing, and run it.
Weekly—spot-check
- Run your three core queries in Perplexity and ChatGPT.
- Are you cited? Named? Described accurately?
- Log a yes/no and a screenshot. Five minutes, no tooling required.
Monthly—trend
- Pull the GSC Generative AI report: AI impressions by page.
- Lay it next to your organic click trend in GA4 and look for the decoupling—flat or rising impressions, falling clicks.
- Log this month’s citation-tool data (Otterly, Rankscale, or Brand Radar) so you have a series, not a snapshot.
Quarterly—baseline
- Rerun a full GEO audit and measure citation-slot movement against last quarter.
- Recalculate share-of-answer for your top 10 queries.
- Decide where next quarter’s content effort goes based on which slots are winnable.
What to measure now
The core fact, stated cleanly so you can quote it in your next budget meeting: AI Overviews are draining 265 million organic clicks a month from the German market—but brands cited inside those answers post 35% higher organic CTR and 91% higher paid CTR than uncited competitors. Losing the click and winning the citation are two different games, and only one of them is still worth playing.
Rank position told you where you stood on a page that fewer people are clicking. AI impression share, citation rate, and share-of-answer tell you whether you’re inside the answer that’s replacing it. Swap the KPI before your competitors do—in DACH B2B, most of them haven’t started.
This is the first piece in a longer thread on measuring and winning AI visibility for German B2B—the next ones get specific on schema, on the GSC report, and on building toward individual citation slots. If you want to start with a number instead of a theory, an AI Visibility Audit is the entry point: a baseline of where you’re cited today, where you’re invisible, and which slots are winnable first. Flat rate, starting at 99 €. Measure before you build.