Start here / 03 Digital authority
Digital authority · baseline against the closing reading
EKU's count in Google's AI answers stayed at 4 across the three readings. Nothing built during the program is on ekupd.com yet. The specialists' counts rose.
How do you know a page worked if nobody froze the starting line?
Readings of 15 July, 22 September and 30 September 2026
| Open buyer term | US searches a month | Google writes an AI answer | Question box | EKU |
|---|---|---|---|---|
| generator monitoring system | 320 | AI answer shown | People also ask | absent |
| generator control panel | 260 | AI answer shown | People also ask | absent |
| power plant control system | 170 | AI answer shown | People also ask | absent |
| remote generator monitoring | 140 | AI answer shown | People also ask | absent |
Volumes: market intelligence dataset, US desktop, 30 Sep 2026, the same instrument as every other volume in this repository. SERP features: second market dataset, US, 30 Sep 2026; its own volume estimates run lower (140, 260, 20 and 90), are kept in the workbook and are never added to the first. All four searches show an AI answer and a question box: the page that wins them needs a direct answer and a FAQ, which the on-page pack already contains.
| What is measured | Instrument and scope | Baseline | 22 Sep 2026 | 30 Sep 2026 | How to read it | Label |
|---|---|---|---|---|---|---|
| AI answers that name EKU when the buyer describes the problem | AI prompt panel, Power Generation | 0 of 5,176 (Jul 2026) | not re-run · see how to read it | not re-run · see how to read it | The July AI panel is a single reading and is not repeated: its instrument stays with the firm that measured. From October, EKU's own measure is the locked question series, a separate instrument that starts its own trend. | Confirmed, single reading |
| AI answers that describe EKU correctly when the prompt names EKU | AI prompt panel | 503 of 744 (Jul 2026) | not re-run · see how to read it | not re-run · see how to read it | EKU is known by name. The file exists; nobody opens it. | Confirmed, single reading |
| AI visibility score across assistants (0 to 100) | Market intelligence dataset, share of answer | EKU 14; Deep Sea 27; DEIF 28 (10 to 17 Aug 2026) | not re-read · see how to read it | not re-read · see how to read it | Read it in the dataset's own screen each quarter; it is not available through the data connection. | Confirmed, single reading |
| Appearances in Google's AI answers (AI Overviews) | Market intelligence dataset, Google US desktop | EKU 4; specialists 13 to 30 (15 Jul) | EKU 4; specialists 27 to 35 | EKU 4; specialists 26 to 41 | EKU did not move. DEIF kept climbing (29, 35, 41). Three of EKU's four appearances are searches for its own name. | Observed |
| Authority score (SEO and backlink metric, 0 to 100) | Market intelligence dataset, root domain | EKU 19; specialists 29 to 37 (1 Jul and 10 Aug) | EKU 19; specialists 30 to 39 | EKU 20; specialists 30 to 39 | One point, inside the 18 to 19 swing this metric shows every month. The gap is unchanged: EKU sits at about half. | Observed |
| Searches where ekupd.com ranks (US organic keywords, all intents) | Market intelligence dataset, Google US desktop | 29 (15 Jul) | 37 | 37 | Not a gain in buyer demand: on 22 Sep and again on 30 Sep, 14 of the 27 organic positions were a product-name collision with home heaters and 4 were buyer terms. | Observed; the reading is a Working Hypothesis |
| Positions 1 to 3, and 4 to 10 | Same dataset | 4 and 2 (15 Jul) | 4 and 7 | 3 and 8 | Top positions are EKU's own name. The 4 to 10 band grows with the heater collision. | Observed |
| Estimated search visits a month (US) | Same dataset, modeled | 207 (15 Jul) | 186 | 202 | Flat. On 22 Sep, 177 of 186 estimated visits (95%) came from searches for EKU's name; the 30 Sep list shows the same pattern. | Observed |
| Referring domains and backlinks | Market intelligence dataset, backlink index | 213 and 925 (1 Jul) | 360 and 1,106 | 448 and 1,280 | Do not present as progress. The 30 newest domains inspected on 22 Sep were automated low-quality sites; the 30 Sep rise was not inspected. | Observed; the reading is a Working Hypothesis |
| Position for "idle management" | Results page check, Google US | #4 (15 Jul) | #4 | #4 | Held. EKU shares that page with truck systems and idle games, so always qualify with "industrial" or "generator". | Observed |
| Position for "maximus controller" | Results page check, Google US | #18 (15 Jul) | #18 | #18 | Positions 1 to 17 belong to an unrelated livestock-controller brand. The 40 searches a month are shared demand, not EKU's. | Observed; the reading is a Working Hypothesis |
| Four open buyer terms, about 890 US searches a month | Market intelligence dataset | EKU absent (Aug) | EKU absent; difficulty 2 to 3 of 100 | EKU absent; 320, 260, 170 and 140 searches; difficulty 2 to 3 | generator monitoring system, generator control panel, power plant control system, remote generator monitoring. Still open, and still the cheapest ground to take. | Observed |
| Clicks from Google, same 31-day windows | Google Search Console, ekupd.com | 314 (14 Jul to 13 Aug, as read in the 23 Sep export) | 350 (22 Aug to 21 Sep) | not re-read · see how to read it | The 13 Aug export read 297 for the first window. Always compare two windows from the same export. | Confirmed |
| Share of visible searches that already contain EKU's name | Google Search Console | 99.2% (nine months to 13 Aug) | 99% (31 days to 23 Sep) | not re-read · see how to read it | The website meets people who already know EKU. Search Console holds data since November 2025. | Confirmed |
| Impressions that reach the Power Generation pages | Google Search Console | not isolated in July | 720 in the last month; 2.7% of all impressions | not re-read · see how to read it | The growth segment is almost invisible in search. The /power-generation page itself showed 0 impressions since 12 Nov 2025. | Confirmed |
| Site health (100-page crawl, 0 to 100) | Site audit | 87 (9 Jul) | 84 (crawl 4 Sep); 87 (crawl 23 Sep) | not re-crawled · see how to read it | Stable. The main error group is invalid structured data on about 68 pages. | Observed; the reading is a Working Hypothesis |
| Website key events configured | Site analytics (property created July 2026) | 0 | 0 | 0 | Analytics collects visits and downloads. Nothing reports a business signal until the events are marked. | Confirmed |
| Published cases with figures | Repository and ekupd.com | 0 | 0 | 0 | The case is written; its figures are with leadership for correction. No confirmed case with figures is published yet. The idle-reduction figures in the case template are model text, with two still to confirm. | Confirmed |
| EKU named in answers to the 37 locked questions (tracked, one AI assistant) | AI answer tracking, one assistant, United States, English | not tracked | not tracked | 0 of 37 (reading of 1 to 2 Oct 2026, after the 30 Sep column date) | First tracked reading, taken with Kanbina Lumina's tool and not repeated. The assistant cited 400 pages, none on ekupd.com. A separate instrument from the July panel and from the manual waves. | Observed, single reading |
| Company | Ranked searches 15 Jul | 22 Sep | 30 Sep | AI answer appearances 15 Jul | 22 Sep | 30 Sep | Authority 22 Sep | 30 Sep | Est. visits 30 Sep |
|---|---|---|---|---|---|---|---|---|---|
| EKU (ekupd.com) | 29 | 37 | 37 | 4 | 4 | 4 | 19 | 20 | 202 |
| DEIF | 692 | 821 | 921 | 29 | 35 | 41 | 34 | 34 | 1,381 |
| ComAp | 640 | 631 | 693 | 22 | 27 | 26 | 39 | 39 | 1,280 |
| Deep Sea Electronics | 911 | 843 | 841 | 13 | 31 | 26 | 37 | 37 | 9,020 |
| Basler | 558 | 537 | 562 | 30 | 31 | 32 | 30 | 30 | 3,438 |
Caterpillar and mtu, the same count, one reading. On 5 Oct 2026 the AI-answer count read 13,676 for Caterpillar (cat.com) and 225 for mtu, Rolls-Royce Power Systems (mtu-solutions.com). These are whole-company websites (machines, marine and rail as well as power), with no July or September reading, so they are a scale reference and not a trend. EKU read 4 on the same day.
Caterpillar and mtu on the buyer searches EKU tracks. Rank tracking of 120 buyer searches, Google US desktop, one location, 8 May and 5 Aug 2026: Caterpillar 3.4% then 3.1% of visibility, top ten for 10 searches; mtu 2.9% then 1.8% of visibility, top ten for 9 searches. EKU read 5.2% on 5 Aug, carried by searches for its own name. A separate instrument, not read in September; the May value is the August value minus the change the tool reports.
Market intelligence dataset, Google US desktop and backlink index. Modeled and directional. Ranked searches and estimated visits are the provider's estimates, not analytics.
Validating the data meant reading every delivered file against the archived readings. These are the differences found and what was done with each one.
| Where it appears | What it says | What is true | What was done |
|---|---|---|---|
| Session kit of 25 Aug (Story Room, Battlecard Arena, Executive Field Brief) | "266 referring domains, +24.9%" and "27 to 33 ranked keywords" | No archived reading supports 266 or 33. The dataset's monthly history reads 213 referring domains on 1 Jul and 277 on 1 Aug; ranked keywords read 27 to 29 (12 Aug). | Corrected in the handover edition of the Story Room and the Arena. |
| Session kit of 25 Aug | "GA4 key events activated" | Analytics has been collecting since July. Key events are specified and 0 are marked. | Corrected in the handover edition. |
| Day 65 readout (3 Sep) and Phase 2 package (AV03) | "11 AI Overview slots where EKU appears" | The 11 are EKU keywords whose results page shows an AI summary. EKU itself appeared in 4. | Read as 4. Corrected in the Phase 2 handover copy. |
| Digital authority (21 Jul) | "643 queries, 16 months" | Search Console holds data since November 2025. The export window label was wrong, the queries are real. | Corrected to "data since Nov 2025" in the English edition here. |
| Digital authority (21 Jul) and later pages | "0 of 1,000 AI answers" next to "0 of 5,176" | Two July samples of different size. The repository uses the full panel: 5,176 answers by problem and 744 by name. | Shown apart, never added. |
| Evidence Pack v1.0 (20 Aug) | Authority score 18; reading note "the only mixed-brand orchestration positioning in the set" | The score swings 18 to 19 month to month (20 on 30 Sep). Exclusivity wording was retired on 23 Sep. | Evidence Pack v1.1 in folder 12 carries both fixes. |
| July and August documents (Commercial landscape, Positioning, session kit) | "empty space", "nobody owns", "unclaimed center" | On 23 Sep a rival's own documentation showed load sharing across controller brands. The white space is now about published proof, not about being alone. | Read those documents with the approved language page in folder 04. |
| Live radar, evidence cut 24 Sep | Generac Link card: "27 Aug 2026, latest release" | The vendor's release notes, read 30 Sep, list 30 Jul, 13 Aug and 24 Sep. | Corrected in the radar update of 30 Sep. |
| Decks before 22 Sep | "0 AI Overview citations of ekupd.com" | The dataset records 4 appearances since July. | Already corrected on 22 Sep; repeated here so nobody revives the old figure. |
| Instrument | What it covers | Dates used | Limit |
|---|---|---|---|
| AI prompt panel | 744 answers by name and 5,176 by problem, Power Generation, several AI assistants, same wording each time | July 2026 | One wave. Set- and date-specific; not market share. |
| Locked 37-prompt tracker | 25 base prompts and 12 Maximus prompts, three assistants, manual capture under fixed conditions | Prompts locked 3 Aug 2026; capture sheet ready | Wave 1 still to be captured. Never edit the prompts between waves. |
| Kanbina Lumina market intelligence dataset | Google US desktop: ranked terms, AI Overview appearances, modeled visits; backlink index: authority score, referring domains; 24 tracked rivals in five tiers | 15 Jul, 10 to 17 Aug, 22 Sep and 30 Sep 2026 | Modeled and directional. The provider recalculates history, so old dates can shift by 1 to 2 points. |
| Google Search Console (EKU account) | Clicks, impressions, queries and pages for ekupd.com | Data since November 2025; read 13 Aug and 23 Sep 2026 | Compare windows from the same export. |
| Site analytics (EKU property) | Visits, downloads; key events once marked | Collecting since July 2026 | 0 key events configured on 30 Sep. |
| Site health crawl | 100-page technical health crawl of ekupd.com | 9 Jul, 4 Sep and 23 Sep 2026 | Health is a technical score, not a ranking. |
| Leadership and field interviews | Leadership interviews; questionnaire answered by four of five sellers | 24 Jul; 3 to 14 Sep; 24 and 25 Sep 2026 | One person's statement never validates a recommendation by itself. |
| Public rival sources | Vendor releases, release notes, datasheets, filings and event agendas, each dated in the radar | 4, 9, 14, 23, 24 and 30 Sep 2026 | Vendor documentation is the vendor's own claim, not an independent comparison. |
| Second market dataset (free tier) | Indexability of the sitemap, page scores, SERP features on chosen terms, an independent estimate of ranked searches and visits | 30 Sep 2026 | A different index from the market intelligence dataset; the two counts are never merged. |
| Own crawl | Titles, descriptions, H1, retired wording, structured data and image alt text on the 44 sitemap pages | 23 and 30 Sep 2026 | Reads HTML as served; three pages re-checked in a rendered browser with the same result. |
| Public YouTube statistics | Subscribers, views and channel age for EKU and four rivals | 30 Sep 2026 | Public counts; no watch-time or audience data. |
| Search autocomplete questions | Questions people type around the buyer terms | 30 Sep 2026 | Text only; no volumes. |
| Short buyer panel (seven prompts) | Seven buyer-problem prompts on three assistants, manual capture | First reading 0 of 7, 2 September 2026 | Counted on its own; never added to the 37-prompt series. |
| Prompt series B | Weighted buyer prompts on three assistants, written for the commercial reading (folder 03) | Written 1 October 2026; first capture still to be taken | Reported next to series A and never added to it. |
| Tracked reading of the 37 questions | One AI assistant through a tracking tool, United States, English | 1 to 2 October 2026, one reading | Kanbina Lumina's tool; not repeated. The denominator of the published percentages is Needs Validation. |
| Buyer-search rank tracking | 120 buyer searches, Google US desktop, one location: visibility of Caterpillar, mtu and EKU | 8 May and 5 August 2026 | Not read in September. Appears only in the Caterpillar and mtu note; never compared with the AI-answer chart. |
Closing
The baseline is frozen and re-read twice with the same ruler. EKU held its own name and one buyer term; it gained nothing on the terms buyers type.
Each month without published pages sends another month of buyers to the specialists. The first real gain will show in the four open terms and in the AI-answer count.
Quote numbers only from this page. Bring it to every monthly review. Use the workbook sheet to add columns.
Validation status. Readings of 15 Jul, 10 to 17 Aug, 22 Sep and 30 Sep 2026. The July AI panel is a single reading and is not repeated: its instrument stays with the firm that measured. From October, EKU's own measure is the locked question series, a separate instrument that starts its own trend. The AI visibility score was read once. Referring-domain growth was inspected on 22 Sep only.
Working foundation for executive review. Not approved external copy. Prepared for EKU Power Drives, 30 September 2026, by KanbinaLUMINA