For Som, from Vin. A proposal to discuss, not a decision yet. Read it alongside the review of Bobby's plan: https://hypertune-reports.pages.dev/aeo-plan-review-2026-10-15-6d6639fc
How it was made. Two AI models (Claude Opus 5.5 and GPT-6 Astra, both on high reasoning) each designed a campaign independently. Each worked from the same inputs: Bobby's 15 Oct plan, our review of it, the live SEO tracker and the house rules. This page combines them. Where they differ, the table at the end says which way this page went and why.
SEO tracker: https://hypertune-reports.pages.dev/seo-gantt-2026-10-02
Run a lean AEO campaign in-house, on top of the SEO plan, not beside it. The same calendar pages get written so AI tools can lift passages from them. Four things are added:
Cost: $150 in tools to 31 Dec, for the Ahrefs Brand Radar $50 package for three months.
Time: about 70 to 85 team hours to 31 Dec. Som about 37 to 44, Hoa about 20 to 25, Austin about 4 to 6, Vin about 3, plus about 11 to 12 developer hours through Austin's Linear tickets. Som's figure already includes the 12 hours planned for F5, so her new time is about 2 to 3 hours a week.
Who does what:
Success by 31 Dec:
That's a small revenue line this year, roughly $150 to $250 of new monthly revenue a month. The near-term value is protecting sales when someone who saw a creator video asks an AI "is Hypertune legit?" before paying.
Google says its AI Overviews and AI Mode need no special optimisation, so for Google the SEO plan already is the AEO plan. This campaign covers what SEO doesn't:
| Measure | Starting figure | Source |
|---|---|---|
| Visits from AI assistants | 88 from 4 Sep to 14 Oct, about 15 a week: 85 from ChatGPT, 3 from Gemini | Google Analytics, AI Assistant channel |
| Signup survey "AI Chat" answers | 7 of 944 answers (0.7%), 29 Sep to 15 Oct. 6 of the 7 paid | Signup survey |
| AI-sourced paying customers | About 11 a month (6 in 17 days) | Survey payers |
| Copilot citations | 167, all to the homepage, time window still to confirm | Bing Webmaster Tools, AI Performance |
| Google AI impressions | 17.6K from 27 Aug to mid Oct, about 2,500 a week. Impressions only, no clicks | Search Console's generative AI report |
| How often AI tools name Hypertune | Not measured yet | Set by the baseline runs, 20 to 29 Oct |
Both Google Analytics and the survey undercount AI visits, because many arrive with no referrer.
"December" means the average of the last two weekly runs in December. Numbers marked provisional are confirmed against the baseline in the 30 Oct report.
| # | Target | Type |
|---|---|---|
| 1 | Named in 35% of answers to the 4 category questions ("best game optimizer"), provisional | Leading |
| 2 | Named in 10% of answers to the 8 problem questions ("how to get more FPS in Valorant"), provisional | Leading |
| 3 | hypertune.gg cited in 50% of answers to the 13 brand questions (comparison, trust, facts), provisional | Leading |
| 4 | Zero wrong price, "free" or partner statements on sources we control (site, help centre, schema, listings, profiles) by Fri 13 Nov | Leading |
| 5 | In December runs: zero answers calling NVIDIA a partner or Hypertune free, and the current price right in 80% of answers that give one | Lagging |
| 6 | Copilot citations in December at 2x October's rate, with 8 or more different pages cited | Leading |
| 7 | Google AI impressions at 4,000 a week in December, with 10 or more pages getting them | Leading |
| 8 | Visits from AI assistants at 35 a week in December | Lagging |
| 9 | "AI Chat" at 1.5% of survey answers in December | Lagging |
| 10 | 25 paying customers in December from AI sources | Lagging |
Two reporting rules:
English, US location, fixed until 31 Dec so runs compare. The ten marked * also run in two extra phrasings. New pages add their question to a separate list (up to 10), kept out of the headline numbers.
| # | Type | Question | Page that answers it |
|---|---|---|---|
| 1* | Category | What is the best game optimizer for PC? | Homepage (C1, C29) |
| 2* | Category | What is the best FPS booster for Windows? | Homepage |
| 3* | Category | Best PC optimizer for gaming in 2026 | C19 |
| 4 | Category | What are the best Razer Cortex alternatives? | C22 |
| 5* | Problem | How do I get more FPS in Valorant? | C12 |
| 6* | Problem | Best CS2 settings for more FPS | C4, C17 |
| 7 | Problem | How to increase FPS in Fortnite on PC | C9 |
| 8* | Problem | Why is my FPS so low on a good PC? | C5 |
| 9 | Problem | How do I fix stuttering in PC games? | C7 |
| 10* | Problem | How do I optimize Windows 11 for gaming? | C3 gaming guide |
| 11 | Problem | Should I turn on hardware-accelerated GPU scheduling? | C11 |
| 12 | Problem | Best Warzone settings for FPS | C10 |
| 13 | Comparison | Hypertune vs Hone: which is better? | C6 |
| 14 | Comparison | Hypertune vs Razer Cortex | C22 |
| 15 | Comparison | Hypertune vs Radiant | None yet. Gap: first comparison page once Bobby confirms Radiant (F3-3) |
| 16 | Comparison | Is Hypertune better than tuning settings manually? | C25 (January); the gaming guide until then |
| 17* | Trust | Is Hypertune safe? | C2 trust page |
| 18 | Trust | Is Hypertune legit or a scam? | C2 |
| 19 | Trust | Hypertune reviews | C2, listings (L2) |
| 20* | Trust | Does Hypertune actually increase FPS? | C13, the new data report |
| 21 | Facts | What is Hypertune? | Homepage, facts page (F5-2) |
| 22 | Facts | How much does Hypertune cost? | C13, facts page |
| 23 | Facts | Is Hypertune free? | C13: paid plans only, no trial mention |
| 24 | Facts | Who makes Hypertune, and is it partnered with Intel, AMD or NVIDIA? | Facts page, C2 |
| 25 | Facts | How do I cancel my Hypertune subscription? | C2 cancel section, help centre (7 ChatGPT visits already land on the account page) |
Hours are Option A (in-house) to 31 Dec. Each has one owner.
New. It replaces the 5-question monthly tracking (F5-4, F5-5). Without fixed questions run often, nobody can tell real movement from AI answers changing on their own. Som 12 h, Hoa 14 h.
Links to F5-0, F5-2, F5-3 and T2. ChatGPT looks up prices and specs on the brand's own site, so one dated facts page beats facts scattered across pages. Som 6 h, Austin 1.5 h, Kamil 4 h.
Inside the existing calendar pages, so no extra pages. AI tools lift passages, not whole pages. Som 2 h to add the rules to the AI content framework and the hypertune-content skill, then about 30 minutes per page for each writer (Som 4.5 h, Hoa 3.5 h).
The rules:
The skill and the Som and Hoa cross-checks enforce these, so it doesn't depend on Bobby's turnaround.
The one new page (suggest tracker item F6). ChatGPT increasingly cites the original source, no competitor publishes per-game gains across hardware, and the report answers "Does Hypertune actually increase FPS?" for every AI tool. Som 6 h, Austin 0.75 h, Kamil 2 h.
Links to L2, L3 and L4; the citation map is new. Hoa 5 h, Som 1 h, Austin 0.25 h.
Links to F1. AI tools merge what profiles say about a brand. Hoa 4 h: an audit in week 3 and a recheck in week 9. She checks about 12 profiles against the fact sheet (X, YouTube, Instagram, Discord, the help centre, schema, AlternativeTo, Product Hunt, Crunchbase, G2 and the download sites) and lists each fix with one owner.
New. Since the 16 Oct price change, AI answers will quote old prices for months unless the sources are fixed. Som 5 h, Kamil 2 h.
Links to R1, R2 and R3. Som 6.5 h, Kevin 3 h.
| Week | AEO work (owner) | Report or milestone |
|---|---|---|
| 19 to 23 Oct | Buy the Brand Radar $50 package (Vin). Load the questions; first run Tue 20 (Som). Pull the starting figures and confirm the Copilot window (Som). Add the writing rules to the framework and skill (Som). Log run 1 (Hoa). Give Som Cloudflare read access; ticket for the schema price (Austin). Approve the questions and provisional targets by Wed 21 (Vin) | Rules applied to the gaming guide (Hoa's draft 19 Oct), the homepage and /technology, and the trust page |
| 26 to 30 Oct | Repeat runs Tue 27 and Thu 29; citation map v1 (Hoa). Baseline analysis; facts page draft to Austin; Cloudflare check (Som) | Fri 30 Oct report: baseline and confirmed targets |
| 2 to 6 Nov | Profile audit and fix list (Hoa). llms.txt; trace the top five wrong statements; tracking spec for Kevin (Som). Sign off the facts page (Austin). FPS data in by Fri 6 (Austin) | Bobby works directories and listicles in map order |
| 9 to 13 Nov | Facts page and llms.txt live by Wed 11 (Austin, Kamil through Linear). Analytics ticket (Austin). Wikidata, separate from hypertune.com (Bobby, L4-1) | Fri 13 Nov report: first movement, our own sources fixed |
| 16 to 20 Nov | Data report draft Wed 18 (Som); sign-off Fri 20 (Austin) | AI figures live in the analytics SEO page |
| 23 to 27 Nov | Data report live Tue 24 (Austin, Kamil through Linear); game pages link to it (Som). Creator brief lines (Austin) | Fri 27 Nov report (light week, US Thanksgiving) |
| 30 Nov to 4 Dec | Second round of fixes (Som). Schema ticket for the Wikidata link and dates (Austin). Listicle pitches use the data (Bobby) | |
| 7 to 11 Dec | Mid-point read against the targets; decide whether to renew Brand Radar on 19 Dec (Vin) | Fri 11 Dec report |
| 14 to 18 Dec | Data report refresh (Som). Profile recheck (Hoa) | |
| 21 to 24 Dec | Automatic weekly run, logged only (Hoa) | |
| 28 to 31 Dec | Final runs logged (Hoa). AEO day-90 review: targets, what worked, keep, scale or stop (Som) | Thu 31 Dec day-90 review |
Bobby does only his existing SEO work. Weeks start on the Monday shown.
| Person | 19 Oct | 26 Oct | 2 Nov | 9 Nov | 16 Nov | 23 Nov | 30 Nov | 7 Dec | 14 Dec | 21 Dec | 28 Dec | Total | A week |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Vin | 0.5 | 0.5 | 0 | 0.25 | 0 | 0.25 | 0 | 0.5 | 0 | 0 | 1 | 3 | 0.3 |
| Som | 6 | 6 | 4.5 | 3 | 4.5 | 2 | 2 | 2 | 2.5 | 1 | 3 | 36.5 | 3.3 |
| Hoa | 2 | 4 | 4.5 | 2 | 2 | 1.5 | 2 | 2 | 2 | 1 | 1.5 | 24.5 | 2.2 |
| Bobby | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 extra | 0 |
| Austin | 0.5 | 0.25 | 0.75 | 0.25 | 0.5 | 0.5 | 0.25 | 0.25 | 0.25 | 0 | 0.5 | 4 | 0.4 |
| Austin (Kamil, through Linear) | 0.5 | 0 | 3 | 0 | 0 | 2 | 0.5 | 1 | 1 | 0 | 0 | 8 | |
| Austin (Kevin, through Linear) | 0 | 0 | 0 | 1 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 3 |
Som's 36.5 hours include the 12 already planned for F5, so her new time is about 25 hours, or 2.3 a week. Astra's version came to slightly more (Som 44, Hoa 20, Austin 6, developers 12), so treat these as a range, not a promise. Cap Som's AEO time at 4 hours a week, and give Hoa the repeatable work.
Bobby takes the measurement (W1), the citation map (W5) and the report section (W8). Som keeps the facts page, writing rules, data report and fixes, and checks his work within 48 hours.
| Person | Total to 31 Dec |
|---|---|
| Vin | 3.75 h |
| Som | 26.5 h |
| Hoa | 6 h |
| Bobby (paid extra) | 26.5 h (October 9, November 8, December 9.5) |
| Austin | 4 h |
| Austin (Kamil and Kevin, through Linear) | 11 h |
At $50, $100 or $150 an hour, Bobby's part costs $1,325, $2,650 or $3,975, plus the $150 for Brand Radar. Astra put his share at 31 hours. Option B saves Som and Hoa about 28 hours, but it puts the 30 Oct baseline in the hands of the person whose tasks most often slip.
Brand Radar check maths. One check is one question on one AI tool for one run. The $50 package gives 2,500 checks a month, plus 150 free in our Lite plan.
| Run | Checks a month |
|---|---|
| 25 questions, 6 tools, weekly | 650 |
| 20 phrasings, 6 tools, weekly | 520 |
| Up to 10 new-page questions, weekly | 260 |
| Claude, monthly (8 checks per question) | 200 |
| Baseline repeats, first month only | 300 |
| Total | 1,930 in the first month, then 1,630 |
The $50 package covers this with about 700 checks to spare. The 150 free checks only buy one run, too thin to tell noise from movement. Daily runs would need the $100 package and add nothing at this size. The full Brand Radar AI ($199 to $699) isn't needed.
| Item | Option A | Option B |
|---|---|---|
| Brand Radar, $50 a month (19 Oct, 19 Nov, 19 Dec) | $150 | $150 |
| AnswerThePublic | free version, $0 | free version, $0 |
| AI tool API testing | not bought | not bought |
| Bobby's extra hours | $0 | 26.5 h × his rate |
| Total to 31 Dec | $150 | $150 + 26.5 h × rate |
Cancel the package before 18 Jan if the day-90 review says stop.
| Item | Why | |
|---|---|---|
| Keep | Fixed questions grouped by type, six AI tools, what gets logged, the words used about Hypertune, citation mapping | Sound method |
| Keep | Self-contained paragraphs, data reports, tracing errors to the source, checking against the approved facts, AEO inside the bi-weekly report | High value, no duplicate systems |
| Change | Baseline in the 30 Oct report, not 13 Nov. Targets set now, confirmed 30 Oct | Four readings before Christmas, not one |
| Change | Monthly runs become weekly runs, averaged over two | Same cost, more readings |
| Change | One owner per task; Kamil becomes "Austin (Kamil, through Linear)"; Hoa added as a writer | House rules |
| Change | Cloudflare instead of server logs | The data already exists |
| Change | Austin's and Vin's review time cut from 6 hours a month each to about 1.5 | Only the facts page and data report need sign-off |
| Change | The AEO check moves into the writers' self-check and the hypertune-content skill | Doesn't wait on Bobby |
| Drop | "About 9 guides" as extra content | They're calendar pages already |
| Drop | Google Ads terms as an AEO result | They're paid clicks; kept only as a source of questions |
| Drop | Press releases this quarter | 99 September pieces already exist |
| Point | Claude Opus 5.5 | GPT-6 Astra | This page |
|---|---|---|---|
| How often to run | Weekly, averaged over two runs | Every two weeks, each run 3 times | Weekly: more readings for the same checks |
| Paying customers target | 25 a month in December, from a real baseline of about 11 | 10 in total by 31 Dec, no baseline | Opus: it starts from the survey data |
| "Named" targets | 35% of category answers, 10% of problem answers | About 17 of 25 questions per tool, brand questions included | Opus, with Astra's rule to report category and problem questions apart from brand ones |
| Hours, Option A | Som 36.5, Hoa 24.5, Austin 4, developers 11 | Som 44, Hoa 20, Austin 6, developers 12 | Opus figures, Astra's shown as the upper end |
| Bobby, Option B | 26.5 hours | 31 hours | 26.5, with 31 as the upper end |
| Comparison pages | Radiant as a gap to fill | Radiant swaps into Hone's November slot, Hone moves later, not an extra page | Both agree Radiant goes first. Astra's swap is the way to do it without adding a page |
| Accuracy | Wrong statements counted | Also: an answer that's silent on price doesn't count as accurate | Astra's rule added |