1. GA through 6 Sep
Closed. Do not pull 6 Sep. That day will look thin. I already have GA through 5 Sep 2026.
Scalekit traffic · 29 Jun 2026 to 5 Sep 2026
PostHog says traffic dropped. I am checking if that is real demand, or bots and measurement.
scalekit.com is the marketing site. I used to write www. That meant the same host.
docs.scalekit.com is the docs site.
SG is Singapore. CN is China. HK is Hong Kong. LLM agent bots hit us from those countries through proxies.
| Host | PH users | PH sessions | GA sessions | Same direction? |
|---|---|---|---|---|
| scalekit.com | 3,097 → 2,744 (−11%) | 3,454 → 3,165 (−8%) | 3,697 → 3,379 (−9%) | Yes, but see mix below |
| docs.scalekit.com | 2,264 → 1,700 (−25%) | 2,421 → 1,848 (−24%) | 1,059 → 1,124 (+6%) | No |
docs.scalekit.com: PostHog users −25%. GA sessions +6%. That is not real demand dying. I think this is bots in PostHog, or mix.
Users vs sessions: fewer people spending more time would look like users down and sessions flat or up. That is not this data. PostHog users and PostHog sessions both fell. They moved together.
| Slice | 29 Jun–5 Jul | 6–12 Jul | Change |
|---|---|---|---|
| Singapore | 531 | 84 | −447 |
| United States | 1,026 | 1,101 | +75 |
| India | 435 | 507 | +72 |
Singapore / China / Hong Kong · 505 users on 29 Jun, then ~5–30/dayUS + India held
Daily cut agrees. 505 of those users sit on 29 Jun. Singapore / China / Hong Kong is where LLM agent bots run (proxies). I ignore this for demand.
| Slice | 29 Jun week | 6 Jul week |
|---|---|---|
| Hong Kong | 836 | 516 |
| China | 641 | 452 |
| Singapore | 162 | 250 |
| United States | 260 | 186 |
| India | 154 | 128 |
| GA sessions | 1,059 | 1,124 |
Singapore / China / Hong Kong · 404 → 72 on 8–9 JulUS + India · 69 → 79 that same day
One segment fell (404 to 72). US + India that day went 69 to 79. GA sessions went up that week. Singapore / China / Hong Kong is LLM agent bots on proxies. I ignore this for demand.
| Page | 29 Jun–5 Jul | 6–12 Jul | Change | Share of week 1 |
|---|---|---|---|---|
| /agentkit/quickstart/ | 158 | 77 | −81 | 15% |
| / (home) | 118 | 112 | −6 | 11% |
| /authenticate/fsa/quickstart/ | 62 | 51 | −11 | 6% |
| /authenticate/mcp/intro-to-mcp-auth/ | 25 | 13 | −12 | 2% |
| /cookbooks/java-spring-boot-jwt-timeout/ | 21 | 33 | +12 | 2% |
| /authenticate/mcp/quickstart/ | 17 | 26 | +9 | 2% |
322 docs pages had a session that first week. The top 10% is 32 pages. /agentkit/quickstart/ is the one that dropped. Home held. Site total still went up.
| Week | scalekit.com clicks | scalekit.com impressions | docs clicks | docs impressions |
|---|---|---|---|---|
| 29 Jun–5 Jul | 1,576 | 601,664 | 75 | 18,655 |
| 6–12 Jul | 1,585 | 580,760 | 69 | 15,947 |
| 24–30 Aug | 1,535 | 431,772 | 62 | 12,723 |
| 31 Aug–5 Sep | 1,272 | 390,019 | 62 | 17,074 |
| Ahrefs day | Keywords 1–3 | Top 10 | All keywords |
|---|---|---|---|
| 5 Jul 2026 | 214 | 423 | 466 |
| 30 Aug 2026 | 181 | 411 | 448 |
I already have the GSC queries file. It ends 5 Sep. That is the minus-2 cut. Ahrefs is not a mass deindex. Keywords in 1-3 went 214 to 181. Top 10 only went 423 to 411. I still will not name a reverse.
| Host | PH users | PH sessions | GA sessions |
|---|---|---|---|
| scalekit.com | 2,783 → 2,566 (−8%) | 3,129 → 2,919 (−7%) | 3,810 → 3,589 (−6%) |
| docs.scalekit.com | 707 → 678 (−4%) | 812 → 759 (−7%) | 862 → 704 (−18%) |
| Host | Today | When added |
|---|---|---|
| scalekit.com | No Content-Signal. Only Disallow /authors/. | Not present. |
| docs.scalekit.com | Content-Signal: search=yes, ai-train=no, ai-input=yes |
20 Apr 2026. developer-docs #576. Bundled in an AgentKit docs PR. |
What it means: ask crawlers not to use docs to train models. Search indexing is still yes. Feeding docs into an answer at request time is still yes.
Should it be there? Common on docs sites. It is a preference, not a block. It does not stop Google Search. It was already there in April, so it did not cause the July drop.
| Date | What shipped |
|---|---|
| 13 Aug 2026 | Gemini 3.7 Flash |
| 19–28 Aug 2026 | Google AI Overviews: generative UI, then longer auto-expand |
| 1 Sep 2026 | Claude Fable 5.1 |
| 2 Sep 2026 | Gemini 3.8 Flash |
| 3 Sep 2026 | OpenAI GPT-6 Astra |
The July PostHog cliff is the wrong month for these launches. The Aug–Sep impression slide sits next to the AI Overviews UI change. That is an angle. It is not proof. Google generative search for us does not move much.
Closed. Do not pull 6 Sep. That day will look thin. I already have GA through 5 Sep 2026.
Closed. I already have gsc_site_queries.csv. It ends 5 Sep 2026. I used it for phrase moves. Page-level totals stay the impression number (602k to 390k).
Closed. Used scalekit.com-organic-keywords-history-subdoma_2026-09-07_17-53-10.csv. Keywords 1-3 went 214 to 181 (5 Jul to 30 Aug). Top 10 held 423 to 411. Not a mass deindex.
Closed. Those are the top locations where LLM agent bots run, on proxies. That locks ignore for demand.
Closed. I have the merges from scalekit-inc/developer-docs. I do not need anything else from that repo. I am not calling those merges the cliff.
Closed. Nope. They were not unpublished, redirected, or noindex in 29 Jun 2026 to 12 Jul 2026. They return 200 today.