What Have Authoritative Sites Measured Recently?

On 2026-08-27, Ahrefs published “AI Overviews Cut CTR by 23.1% in France.” The number comes from a controlled comparison in the French market, and the conclusion is specific: when an AI summary appears at the top of search results, the share of visits that can still be clicked away falls by more than a fifth. I suggest foreign-trade business owners remember three things: first, this is a test of one specific market; second, the size of the decline varies with keyword type and industry; third, what it measured is click-through rate, not inquiries.

Right after that, on 2026-10-01, Ahrefs offered another number: 83% of branded searches trigger an AI Overview. This one deserves more attention from foreign-trade people than the previous one. What does it mean? Even when someone searches specifically for your company name, Google puts an AI summary up first, and only then comes your official website. If your brand information is scattered and inconsistent across outside sources, that summary may come out a mess.

Semrush’s “What is zero-click marketing?” of 2026-09-11 pulls these phenomena into one framework: clicks in search are already declining, and the marketing goal should shift from “bringing people to the page” to “putting the right information out there before the user decides.” They do not say clicks are useless; they say that treating clicks as the only KPI will point you in the wrong direction. I agree with that judgment, and it matches the record I keep along the line of AI Overview click declines.

What Does Zero-Click Actually Mean for Foreign-Trade Sites?

Do not panic yet. In foreign-trade B2B, the path to a deal is long to begin with: from the moment a German buyer first sees you to the moment an inquiry arrives, weeks or even months can pass. We used to treat the “first click” as the starting point; now the starting point has moved earlier, to a single sentence inside an AI answer. A user might remember your product name and, two weeks later, search your brand name directly and come in.

So the accounting in the zero-click era has to be done differently. We used to count “how many clicks this keyword brought”; now we have to count “under this question, did I appear in the AI answer, and did I appear correctly.” The former lives in Search Console; the latter can only be checked by manual sampling. That is why, in my note about the traffic decline, I keep stressing: do not just stare at the total traffic curve — look at whether the mix of keywords has changed.

A more practical point: what zero-click cuts most are often the questions that “one sentence can finish.” Take “what is FOB” — the answer is standardized, so AI just gives it. But a question like “can you do mixed loading in a 40-foot container for this spec of yours” cannot be answered by AI; you still have to find a supplier. For factories in niche categories, that is actually good news — the more specific you are, the harder you are to dismiss with a single sentence. For a related judgment, see traffic getting harder is not entirely the fault of AI summaries.

Our Chinese-Market Test: Which Keywords Get Covered Up Most Easily?

I ran a round of AIO coverage testing on Chinese-language queries myself, with a sample of 500. The overall trigger rate was 80.9%, with 84.1% for the US region and 77.8% for the Hong Kong region. What is really interesting is the difference by keyword type: tutorial queries trigger an AIO 96.7% of the time, while technical queries trigger one only 62.9% of the time. The full record is in our Chinese AI Overview coverage test.

This difference directly guides topic selection for foreign-trade sites. If you write how-to keywords like “how to clear customs” or “how to calculate CIF,” you are most likely colliding head-on with AI summaries, and users leave after reading the summary. If you write about how your product is selected under specific working conditions, about parameter limits, or about common reasons for repairs, the AI summary cannot cover it all, and users will still click in to verify. In other words, the more standardized the knowledge, the more dangerous; the more it carries your own experience, the safer.

There is another cold set of numbers: this site’s indexing rate is 100%, but in 28 days it had only 12 impressions and 0 clicks. These numbers are not meant to scare anyone; they show that “being indexed” is already the lowest bar and confers no advantage at all. We used to say “get indexed first and figure out the rest later”; now indexing is just the ticket in — impressions and citations are the game. You can go through the AI citation readiness checklist item by item to see which one is weakest.

Branded Keywords Cannot Escape Either, and That Is Actually Good News

When 83% of branded searches trigger an AIO, the first reaction is that this is bad: a customer searches my name and sees Google’s summary first. But look at it another way — that summary is actually a free “identity verification sheet.” If AI says you make CNC precision parts, are in Dongguan, and hold ISO certification, and that is true, then that summary is building trust on your behalf. Conversely, if AI gets it wrong — mixes up your main category, for example — that is a real loss.

So the approach to branded keywords has to shift from “hold the number one ranking” to “feed the right sources.” When AI describes a brand, it looks at your official website, your Google Business Profile, industry directories, news coverage, and even other people’s review pages. You need to make sure these places say the same thing about who you are, what you do, and where you are. In how to fix it when AI gets your brand wrong, I wrote out the specific order: fix the most authoritative place first, then close the gaps step by step.

For small and mid-sized foreign-trade sites without brand recognition, this has to be done earlier. You do not have many external mentions, so what AI can grab is your own website and a few directory pages. At that point, one clear company profile page and one consistent product naming system are worth more than ten more generic articles. Start with the relationship between brand authority and AI search to sort out your priorities.

In the Zero-Click Era, Which Three Things Should You Change on a Page?

Put the conclusion in the first paragraph, with no buildup

An AI summary has to judge what a page is about in an extremely short space. If your answer is buried on the third screen, the odds of it being excerpted drop. Answer the question in the headline directly in the first paragraph, then unfold the background — that is the first thing I change when editing a draft.

Add applicable conditions and numbers to every section

“Short lead time” is useless; “25 to 30 days for standard specs, plus one week in peak season” is useful. Sentences with conditions and ranges are more likely to be selected by a summary and less likely to be distorted. Here you can follow the approach in structured data and AI search and mark up the machine-readable parts.

Make the page able to answer one question on its own

Do not write one page as an entire product manual. One page should solve one kind of procurement question, and the headline is that question. What gets cited is usually a single answer paragraph that stands on its own, not a sprawling collection page. On how to organize it, I broke it down in page structure and citation probability.

A Low-Cost Implementation Checklist

Three steps, no money spent, and you can finish the first round within a week.

1. Sample 30 of your keywords and test them for AIO: split them into tutorial, product, and branded keywords, manually check each one for an AI summary, and note whether the summary mentions you.
2. Pick the 3 pages most heavily covered by summaries and rewrite their openings: move the conclusion up, add the numbers, narrow the question — and do not touch the whole site at once.
3. Unify your brand identity wording: write your website intro, directory listings, and social media bios with the same set of phrases; the company name, main business, and location must match in all three places.

If you want to judge more systematically where to start, read the Search Console diagnosis and whether free tools are enough. Zero-click is not the finish line; it has only moved the contest from “grabbing clicks” to “winning the retelling.”

The Questions Bosses Ask Most Often

Does zero-click mean no traffic and wasted effort?

No. Clicks in search are declining, but the chance for your brand to be seen is still there. What matters is whether your information makes it into the answer, and whether users come back to find you after seeing it.

Will AI Overviews swallow up all our organic traffic?

Testing in the French market showed a 23.1% drop in click-through rate, but a decline is not the same as zero. The impact varies widely by keyword type, so start by seeing which category your own keywords fall into.

If a foreign-trade site has no branded keywords, is there still an opportunity in zero-click?

Yes, but the goal has to change. Stop fixating on clicks; first write clear answers to procurement questions and keep your company identity consistent, so AI can describe you accurately in its answers.

Should we just block AI crawlers?

This is not black and white. Blocking reduces the chance of being read; not blocking means accepting that your content will be summarized. Figure out which layer your customers mostly encounter you at, then decide.