- How to tell the difference between a site that has gone invisible and a site that has lost rankings
- The first-party numbers from a five-week stretch where traffic collapsed and one metric moved the other way
- Why treating volume and rate as separate dials changes what you actually do in a bad month
Quick answers
- Q: Why would the conversion rate rise while search traffic disappears?
- A: We can't pin it to one cause. Over the same window we were fixing the receiving end — a Japanese-language signup destination and reworked CTAs — so that may be part of it. What's certain is the narrower fact: the volume metrics and the rate metrics did not move together.
- Q: What's the first number to look at in a month where traffic drops?
- A: Not session count. Look at indexed page count, whether any manual action exists, and average position. If those are healthy and only position is falling, the site isn't broken — it lost rankings, which is a different problem with different fixes.
Every figure below comes from the analytics and Search Console data for Kingfin's Japanese site (kingfin-jp.com). These are one site's measurements, not evidence of what will happen to yours, and doing the same things carries no promise of the same result. We also can't confirm from the outside why a search engine moved the way it did. Read this as a record of how it went in one case.
One morning, search traffic stopped arriving
On July 19, 2026, search traffic to Kingfin's Japanese site effectively stopped. Trailing-seven-day sessions from search went from around ten to essentially zero, and stayed there. Nothing had been deleted. Nothing had moved. Every article was still live at the same URL it had always been at.
The instinct in that moment is to hunt for a cause. A penalty, maybe. Thin content. Some setting quietly broken. That's most of what the following week went into. The honest summary of what we found: nothing was broken.
Five weeks later, it still hadn't come back
As of August 22, 2026 — five weeks in — the picture looked like this. Not only had nothing recovered, average position had kept sliding.
| Metric | Before the drop | As of Aug 22, 2026 |
|---|---|---|
| Search sessions (trailing 7 days) | ~10 | Near zero, five weeks running |
| Average position in Google Search | 24.1 | 29.6 (worse, not bottoming out) |
| Total sessions (28 days) | ~241 | 173 (-28.2% vs. prior period) |
| Indexed pages | — | 475, no manual actions |
That last row is the one that matters. 475 pages indexed and no manual action means the site had not become invisible to the search engine. It was visible, and it had lost position. Those two situations get treated as the same problem constantly, and they aren't. The first one is a technical failure you can go and repair. The second one is the absence of anything to repair.
Five weeks of zero also does something to your willingness to keep publishing. The thought that nothing you write is reaching anyone was not a paranoid one — it was arithmetic.
The number that went the other way
Then a different metric over the same 28 days did something we weren't expecting. Among sessions arriving from social, the share that reached the signup button rose from 3.75% to 9.38% — roughly 2.5x.
Fewer people arriving. A higher share of them acting. Both at once.
| What you're looking at | Movement | What it means |
|---|---|---|
| Search traffic (volume) | Flat at zero | Rankings lost. Not directly under your control |
| Total sessions (volume) | -28.2% | The search collapse working through the numbers |
| Social conversion (rate) | 3.75% → 9.38% | The receiving end is getting better |
Before the drop, the problem was "people arrive and don't convert." Five weeks later it was "people convert, but almost nobody arrives." Same complaint on the surface — completely different next move.
Why did only the rate move? (We can't say for sure)
We can't isolate a cause. Over the same period we were reworking the receiving end — a Japanese-language signup destination, replaced CTAs across the site — so that may well be part of it. With a sample this small, that stays a guess.
What isn't a guess is the shape of it: the rate metric improved independently while the volume metric was flat on the floor. You can't reach into the rankings and turn them. You can absolutely change what a visitor sees, where they hesitate, and where they leave. Knowing those are two separate dials changes what you do in a bad month.
Volume and rate are separate dials
The default reaction to a traffic collapse is to stop everything. Traffic's gone, so publishing is pointless, so publishing stops. It happened here too — article output went dark for nine days.
But if rate work stops along with volume work, every leak in the funnel is still there when volume returns. Put the other way round: a quiet month is the cheapest time to fix rate, because with few enough visitors you can actually follow what each one does.
- Work on the path to signup — the fewer visitors there are, the more closely you can trace each one
- Non-search traffic routes — social, email, referrals: anything that doesn't depend on position
- Recording your rates — without the prior-period data, you can't judge anything when traffic returns
Go look at your own conversion path first
OlympTrade, the platform Kingfin promotes, takes users from signup through to trading in Japanese. Seeing the actual signup screen the way your readers see it makes it much easier to work out where they're stopping. Trading carries the risk of losing your principal, and income amounts are not guaranteed.
Sign up freeHow to read a rate when the sample is small
About that 2.5x. It needs a caveat, because with a small denominator a rate swings on almost nothing.
Four conversions in 100 sessions becoming eight doubles the percentage. The real change is four events. Judge a change on the percentage alone and you'll book noise as a win. When a rate moves, read the numerator with it — say how many became how many, out loud if it helps.
This is the same lens worth using on anyone else's affiliate results. A percentage published without the raw count looks larger than life in both directions.
A checklist for the month traffic falls
Here's the article in a form you can run against your own site.
Search traffic still hasn't returned. The rate at which arriving visitors act went up anyway. That's what was left over from not shutting everything down. Start by writing down whether the thing falling on your site right now is volume or rate — that one split is most of the work.
Frequently asked questions
Disclosure: This article is informational and educational content from Kingfin's English editorial team. All figures are first-party measurements taken from kingfin-jp.com's own analytics and Search Console data and are not evidence that any other site will see similar results. The causes of search ranking movements cannot be confirmed from the outside, and the interpretations offered here are one reading of the data. Kingfin's affiliate program promotes OlympTrade, an FX and binary-options trading service that is not registered as a financial instruments business in Japan. Trading always carries the risk of losing your principal, and we cannot promise that anyone will definitely earn or that gains are guaranteed. Results vary from person to person; only ever trade with money you can afford to lose, and make your own decisions at your own responsibility.
