Skip to main content
Main content

Public evidence

EdenRank AI citation evidence.

EdenRank runs its measurement workflow on its own brand. This page publishes the measured window and, when present, the exact prompts, engines, dates, and owned pages used as answer sources.

In the last 60 days, AI answer engines cited edenrank.com in 1152 of 10658 answers we measured (11%) across 10 engines; 4 of those engines produced at least one owned-source citation. Mentions and third-party citations are not counted as owned-source citations here.

All measured answers
11%
1152 of 10658
Answers that could carry a citation
29%
1136 of 3921

measured_runs counts every contract-engine run including answers that structurally cannot cite (parametric, organic fallback); citation_capable_runs counts only runs whose answer surface could carry a citation. Quote either rate only with its own denominator.

Registered randomized case study

A pre-registered, randomized two-arm test on 48 frozen assigned assets across 48 clusters, endpoint assigned_asset_citation_rate. Every readout below is re-verified against the frozen statistical design before it is shown here.

ReadoutEffect95% CIpAnswers (T / C)Verdict
D14primary · 2026-08-05+8.6pp+3.2 to +13.90.00071891229 / 182876caused lift
D28durability · 2026-08-19+11.0pp+5.3 to +16.60.00071892297 / 1821641caused lift

What this does and does not show

Control-arm asset URLs were randomized and reserved but deliberately never published under this protocol, so the control arm's assigned-asset citation rate is zero by construction: this measures whether published assigned assets earn citations, not a controlled change in the brand's overall citation rate.

This result applies only to EdenRank.com under the disclosed protocol and is not cross-brand evidence or a guaranteed client outcome.

Protocol registration hash: b5e252c720a575f0e1012b0b985184a952ab1ed9509d342045692b010c0a24f0

Controlled measurement so far

A different question from the case study above, on a different population: these are before/after samples on hub pages, not the randomized assigned-asset test.

8 of 8 controlled before/after samples on our own hub pages showed no detectable effect (avg -1.2pp, most recent 2026-07-14). We publish this rather than hide it - a null result is a result, and it is why our own hub growth is frozen pending the next readout.

Next: hub-carry-experiment-v1 (randomized two-arm, 17 treatment + 17 control pages, pre-registered), first readout 2026-09-06, durability 2026-09-20. Protocol registration hash: bd0b784e755d548a08c80a7e7b3b1548ba7950b487e62d8b3a681d6ae10f25ae

Owned citations over time

Answers that used an EdenRank-owned page as a source, divided by measured answers in each month.

May 2026
0/118 · 5 prompts
Jun 2026
0/1028 · 32 prompts
Jul 2026
394/4779 · 139 prompts
Aug 2026
704/5221 · 227 prompts
Sep 2026
54/726 · 106 prompts

AI answers vary between runs - these are real results EdenRank captured on the dates shown. “Cited” means the answer linked one of our own pages (edenrank.com or our answer hub), not just mentioned the name.

Want this for your brand?

Use the same workflow to measure mentions and sources, diagnose gaps, prepare approved citation assets, and verify later runs. Start with a free check.