Our First-Page Inventory Grew 12x. Clicks Stayed at Zero.

A dashboard that only shows a single trend line will tell you, cheerfully, that things are working. Ours told us exactly that: our named commercial keywords eligible to be clicked — sitting inside Google's first page of results — grew twelvefold in eleven days. Query count up 74%. Impressions on that segment up over 1,400%. Share of our total search visibility carried by that segment up twelvefold. Every instinct built by years of SEO reporting says that's the chart you screenshot for a stakeholder update.

Then we checked the other number next to it. Clicks on that same segment, before and after: zero, and zero.

This is the fourth and closing post in a series that has used only our own site's anonymized Google Search Console and Google Analytics data to show, in public, exactly what building organic visibility for a new brand in Japan actually looks like — including the parts that don't work yet. The first post showed that building content supply doesn't by itself build brand-name search demand. The second showed that a program can look dead on named keywords while it's actually working through the long tail Search Console never discloses by name. The third built a diagnostic method for a single page that ranked well and still got zero clicks. This post is about what happens when that zero-click pattern isn't confined to one page — it's showing up precisely in the segment of our data that grew the fastest and should, by any reasonable expectation, have been the first place new clicks showed up.

Our front-page inventory grew 12x. Clicks stayed at zero.

We track, at intervals, how many of our named (Search Console–disclosed) queries hold a position inside the first page of results — position better than 10 — because that's the threshold below which a listing is realistically visible to a scrolling searcher at all. On 2026-08-07, that inventory was thin: 19 distinct queries met that bar, generating 124 total rows in our pull, and only 24 total impressions actually fell inside position < 10 — 0.64% of that day's trailing-28-day, site-wide impression total.

Eleven days later, on 2026-08-18, we pulled the same read at the same threshold, over the same trailing-28-day methodology (window shifted forward to 2026-07-19–2026-08-15). The inventory had grown to 33 distinct queries, 139 rows, and 377 impressions inside position < 10 — 7.76% of that day's site-wide 28-day impressions. Query count up 74%. Impressions on that segment up roughly 1,471%. Share of total visibility carried by first-page named queries up roughly twelvefold.

For context, our site-wide 28-day impression total over the same rough period moved from 4,160 to 4,857 — up about 17%, itself a derived comparison across two reads five days apart rather than a strict period-over-period figure, and noted here as such. The point isn't that overall traffic grew. It's that the segment closest to being clickable — named, first-page, commercial-intent queries — grew far faster than everything else around it. This is a materially different finding from what we expected walking in. Our working assumption through the first three posts in this series was that impression growth was mostly happening in the anonymized long tail, diluting any signal from the named layer. That assumption doesn't survive this read. The named, first-page layer is where the growth concentrated, and it's exactly the layer that should be easiest to convert into a click, because a searcher can see the actual keyword match sitting in front of them.

Those 377 impressions clicked zero times. Not "few." Zero. And they weren't spread thin across 33 keywords — 327 of the 377 impressions, 87% of the segment's total, came from just two keywords: one commercial keyword generating 173 impressions at a weighted average position of 6.37, and a second generating 154 impressions at 8.76. Two mid-single-digit-to-low-double-digit positions, comfortably inside the first page, together carrying the overwhelming majority of this segment's growth — and zero clicks between them.

One more check worth stating plainly: the weighted average position across this whole 28-day window was 15.90, essentially flat against where it's held before. This growth was not the product of the whole site's rankings jumping up a tier. It was concentrated new density at the front-page layer specifically, sitting inside an otherwise stable overall position picture.

What didn't move

Under the same authoritative 28-day methodology, site-wide clicks read 21 on 2026-08-13 and 21 again on 2026-08-18. Not a small increase. Not a decrease. The exact same number, five days apart, while impressions climbed from 4,160 to 4,857.

Site-wide CTR moved from roughly 0.50% to 0.43% over that stretch. We want to be precise about what that drop actually is: it's arithmetic, not a new negative signal on its own. When the denominator (impressions) grows and the numerator (clicks) holds flat, the ratio falls by construction. Reporting that CTR "declined" without saying why would make a mechanical consequence sound like a fresh finding. It isn't one. The finding is the flat click count sitting underneath it.

Zero clicks at position one, not just position ten

The easiest explanation to reach for when a keyword gets zero clicks is "it's not ranked well enough." We checked that explanation directly, at the daily level, for the two keywords carrying most of this segment's growth — and it doesn't hold.

One of the two spent 2026-07-25 through 2026-08-07 — fourteen consecutive days — at a position between 1.00 and 2.25. Eight of those fourteen days, it sat at exactly position 1.00: the very top organic result, for two full weeks, on a real commercial keyword. That span accounted for roughly 86 impressions. Zero clicks. Position then fell into the 13–23 range after 2026-08-08, before recovering to 6.33 and 5.80 on the two most recent days we have data for.

The other keyword had impressions every single day across the entire 28-day window — no gaps — with position holding almost the whole time in the 3-to-9 band (most recent five days: 3.86, 4.71, 4.12, 4.60, 4.50). Cumulative total: 173 impressions, 0 clicks.

This distinction matters. "We're not ranked well enough yet" is a comfortable explanation because it points to a fix everyone already knows how to attempt: rank higher. What we're looking at instead is a keyword that spent eight separate days at the literal top spot on the page and still didn't get clicked once. Whatever is suppressing clicks here isn't waiting on a ranking improvement. It's happening at position one exactly as much as it's happening at position ten.

The four keywords that should be converting, and aren't

Zooming out from those two, our full set of four commercial, transaction-intent keywords — terms describing our Amazon marketplace-entry consulting service, our cross-border e-commerce agency service, our online marketplace agency service, and our general e-commerce agency service in Japan — produced impressions of 173, 156, 154, and 121 respectively, at weighted average positions of 6.37, 10.55, 8.76, and 13.24. All four: zero clicks, over the same 28-day window.

None of these four sit in the position range where "nobody sees it" is a plausible story. Three of the four average inside the top 10. This is a set of keywords ranked roughly where a mid-funnel, high-commercial-intent search program would hope to be — and every single one of them converted zero of its combined 604 impressions into a click. That's a stronger, more specific claim than "our named keywords underperform," which we already established in Post 2 across the full 139-keyword layer. This is: four keywords, individually chosen because they carry buying intent, individually ranked well, individually still at zero.

What we ruled out today — and what we didn't

Before writing any of this up, we went and manually checked what the actual results page looks like for two of the four keywords, at two geolocation settings each (United States and Japan), for four checks total. In all four: no AI-generated answer summary sitting above the organic results, no paid ads occupying the top of the page, and our own listing holding the top organic position. That rules out the two easiest external explanations — "an AI summary is capturing all the attention" and "ads are pushing us below the fold" — at least for what the results page showed on the day and settings we checked. We're stating that narrowly and deliberately: this was one day's manual check, from an authenticated browser session rather than a strictly logged-out one, and it is not a standing claim about every day or every searcher's view of the page.

There are two things we specifically did not rule out, and we think leaving them out of this piece would make the diagnosis dishonest.

First: the page carrying the single largest share of this front-page growth — 173 of the 377 impressions — had its on-page title edited on 2026-08-13. Google recrawled the page on 2026-08-17. But our manual SERP check on 2026-08-18 still showed the old title and the old snippet in the live results. The recrawl happened; the display layer hasn't caught up to it yet. That means a meaningful share of this zero-click data may still be reflecting what the old, pre-edit listing looked like to a searcher — not the current version of the page. We can't call this favorable or unfavorable to any particular theory. It's unresolved evidence, and it stays unresolved until the display layer catches up and we get a fresh read.

Second: our site has ongoing scanner and crawler traffic. On the day we pulled this data, 46.7% of requests logged at our edge returned a 4xx response — a pattern consistent with automated probing, not human browsing. The keyword that sat at position one for eight straight days without a click is exactly the shape of pattern a rank-tracking tool or a scraper checking its own position would produce, independent of any real person searching. We have not done the work to separate that kind of traffic from genuine search impressions in this data. We're naming it here as a real, live, unexcluded alternative explanation — not a footnote to dismiss, a candidate mechanism we simply haven't tested.

Where the real clicks are

We're not rebuilding the method here — the second post in this series already did that work in full. The short version, applied to this window: the named, disclosed query layer (139 queries, 1,591 impressions) produced zero clicks, and every one of our site's 21 real clicks this period sits inside Search Console's anonymized long-tail aggregate, the same pattern Post 2 documented in detail. That pattern hasn't changed. What's new in this post is the front-page segment specifically, and why it deserved its own look rather than being folded back into that same finding.

What we're doing differently, not just watching the gap

Diagnosing a gap and then publishing an article about it isn't the finish line. On 2026-08-17, we shipped a dedicated, intent-matched landing page built specifically for the two keywords carrying 87% of this segment's impression growth — replacing the assumption that another general explainer post would eventually convert those searches with a page built for what a transaction-intent searcher actually wants to land on. This follows the same logic the third post in this series laid out: a page-type mismatch doesn't get fixed by editing a title, it gets fixed by building the right kind of page.

We want to be honest about what that change has and hasn't proven. A single click or two, on a site at our current traffic volume, is noise — not evidence the fix worked. And per the unresolved display-layer issue above, we can't cleanly separate "the new page isn't converting yet" from "the display layer for the related listing still hasn't caught up" until that catches up and a full measurement window passes after it does. We're stating the change and the reasoning behind it, not claiming a result.

Separately, on 2026-08-16 — the day before that page shipped — we made a resourcing call for the current cycle: new content investment moves toward conversion-side fixes (dedicated, intent-matched pages; the long-tail terms already producing real clicks) rather than toward further expanding the pool of impression-generating supply. The front-page segment covered in this post had already grown 12x without moving a single click. Adding more supply to a segment with that track record, before fixing why the existing supply isn't converting, isn't where the next unit of effort goes.

The transferable point

"Impressions crossed a target" and "named commercial keywords grew in the first page" are both supply-side numbers. Neither one is a conversion number, and treating either as evidence of success on its own is the same mistake dressed up in a bigger, more exciting-looking chart. A twelvefold increase in front-page inventory is a genuinely large number. It is also, by itself, exactly as uninformative about whether the program is working as a flat number would be — because the number that actually answers that question, clicks, didn't move.

That doesn't mean the growth was pointless, and it doesn't mean the honest read is "our content failed." Two live, unresolved confounds — a display layer that hasn't caught up to a real edit, and bot or rank-tracker traffic that hasn't been separated from genuine searches — mean we don't yet know how much of this zero-click pattern is a real conversion problem versus a measurement artifact still working itself out. What a supply-side number like this is actually good for is telling you where to look next, and which alternative explanations to rule out before you draw a conclusion — not handing you the conclusion itself.

FAQ

Q: Why would front-page-eligible impressions grow 12x while clicks stay at zero?
Because impression growth on this segment measures how many queries newly crossed into the first page of results — a supply-side, algorithmic outcome — not whether anyone clicked. In our case, several new keywords entered top-10 positions inside an eleven-day window, which expanded the pool of impressions eligible to be clicked. A bigger pool of clickable impressions doesn't guarantee any of them get clicked; that depends on what the results page actually shows above and around the listing, and on whether the crawler has indexed the current version of the page at all. Both of those need to be checked separately from the position number.

Q: Does ranking well for a commercial keyword guarantee clicks?
No. We had positions in the single digits — one keyword spent eight separate days sitting in position one — and still produced zero clicks across the full window. A strong average position answers only one question: does the search engine consider the page relevant and authoritative for that query. It says nothing about whether an AI-generated summary, an ad, or a directory listing is occupying the space above the listing, and nothing about whether the page format matches what someone searching that term actually wants next. The third post in this series has a fuller checklist for separating those two causes.

Q: If exposure keeps growing and clicks don't, should you stop publishing?
Not necessarily, and it depends on what the new exposure actually is. Growing impressions on queries that were already converting elsewhere is a different situation than growing impressions on a small cluster of commercial keywords that are individually still at zero. In our case the more useful move wasn't to publish less or more — it was to stop treating supply growth as the finish line and put the next round of resourcing into fixing the specific keywords that were already ranked well and still not converting, which is the decision we describe below.

Q: How do you know a new landing page actually fixed a zero-click keyword?
You don't know until three separate conditions are all met: the search engine has recrawled the new or edited page, the results page itself is actually showing the updated title and snippet (recrawl and re-display are not the same event — ours were still out of sync as of this writing), and a full measurement window has passed after both of those are confirmed. On a site at our traffic volume, a single click or two is noise, not proof. We're not claiming our fix worked yet; we're stating what still has to happen before that claim would be honest.

Q: Is a rising impression count on its own a meaningful success metric?
Not read alone. It has to be paired with clicks and CTR from the same window and the same dimensional cut — the second post in this series lays out why mixing date-level, page-level, and query-level reads of the same underlying data produces contradictory-looking numbers. A rising impression count tells you the algorithm is surfacing you more often. Whether that surfacing is turning into anything is a separate question with its own separate numbers, and this post is largely about what happens when those two numbers stop moving together.

Q: Could some of this zero-click data just be bots or rank trackers, not real searchers?
We can't rule it out, and we think it's dishonest to write an article like this without saying so. Our site has ongoing scanner and crawler traffic — on the day we pulled this data, roughly 46.7% of requests to the site returned a 4xx response, which is consistent with automated probing rather than human browsing. A keyword that sat in position one for eight straight days without a click is exactly the kind of pattern rank-tracking tools and scrapers would produce on their own, independent of any real searcher. We have not done the work to separate that traffic from genuine impressions, so we're listing it here as an open, unresolved alternative explanation rather than pretending our zero-click numbers are cleanly attributable to real human behavior.