Hypertune: 90-day progress report

Prepared 21 September 2026 by Vinson Leow for Cosmic Wire leadership, reviewed with Austin Copeland the same day. Covers 21 June to 21 September. Sources: the Hypertune analytics dashboard (values as at 21 Sep, 16:39; the tiles below reproduce it), Stripe, Klaviyo, Search Console, the site itself, and the decisions recorded on calls with Austin between 10 and 21 September.

How to read the numbers. Every comparison in this report is the last 31 days (22 Aug to 21 Sep, the dashboard's default view) against the 31 days before (22 Jul to 21 Aug) and the 31 days before that (21 Jun to 21 Jul). Rates are shown as their value in each window; changes are shown as percentages. Nothing is compared to a single day.

1. The short version

Cash collected doubled across the quarter: $26,119 in the first window, $37,577 in the second, $52,755 in the third (Stripe reports $53,409 gross and $49,159 net for that last window; the dashboard tile is being reconciled to Stripe and Stripe's figure is the one to quote). MRR is $48,773 today against $35,924 at the start of the quarter, up 36%. Revenue churn fell from 64% to 25% and subscriber churn from 54% to 20%, most of it after the trial came out of the default checkout on 5 August.

The pricing changes (1 and 21 July) and the trial change (5 August) are what moved the business. Visits and sign-ups roughly doubled in the last window, with the press campaign, creator activity and the win-back email all landing in the same weeks; the tracking does not yet separate them, so no share is attributed here. New paying customers rose 7% on the prior window.

The last three weeks went into measurement and the website before paid advertising starts: a company-wide dashboard, a speed and conversion review now largely implemented on the test site, Google Analytics and Search Console installed. Paid campaigns are gated on tracking parameters reaching Stripe, which is the first item on Kevin's list.

2. The main numbers

Live today (not window-based): MRR $48,773, ARR $585,281. Paying 7,519 (5,026 monthly, 2,493 yearly, 33%). Past due 347. Trialing 361 (mostly year-long creator comps; the true 7-day trial count is about 100, to be split on the dashboard). Pending cancellation 895 (11.9%).

Two cautions on the table. Cash collected is the dashboard's figure in all three columns so the trend is like for like; Stripe shows $53,409 gross and $49,159 net for the last window and the tile is being reconciled. Churn in the first two windows counts trial expirations as churn (the dashboard's base includes trialing accounts); trials were removed on 5 August, so part of the fall is definitional and part is real. The cohort table in section 2.4 is the clean read.

2.1 Money and subscribers

Money · 22 Aug to 21 Sep

Subscribers · live counts, not filtered by date range

Realised price is 93.4% of list: $4,292 given away on 9.3% of invoices through promotions and creator comps, up from 0.4% of invoices at the start of the quarter as the $59.99 annual promotion and creator comps came in. Revenue by calendar month: June $30,113, July $30,959, August $50,464, September to date $43,700 (21 days).

2.2 Pricing mix

Plan and term · paying subscribers

58.3% of paying subscribers are still on the legacy $4.99 / $47.88 pricing sold before 1 July. Tier 1 ($9.99 / $69.99) holds 2,480 (33%). Regional Tier 2 and 3 hold 658. As legacy subscribers churn and new ones arrive at Tier 1, MRR rises even at a flat subscriber count.

2.3 Churn and why people leave

Why they leave · 22 Aug to 21 Sep

Revenue churn 25.1% ($10,489 of $41,769 MRR at window start), subscriber churn 19.5% (1,384 of 7,095). Leavers pay above-average prices, so revenue churn runs above subscriber churn. Cancellation survey, 730 responses: did not give me performance 38%, too expensive 20%, not using it enough 19%, other 17%, bug 5%.

2.4 Retention by cohort

Month-1 retention ran between 24% and 43% for every sign-up cohort from December to July. The August cohort, the first signed up without a default trial, is at 57.8%. This is the clearest single piece of evidence in the quarter: the trial was bringing in people who left within a month. Older cohorts flatten at 13 to 18% by month 6, which is the durable base.

2.5 Conversion and channels

Funnel · 22 Aug to 21 Sep · stages come from different systems, indicative not strict

Channels · 22 Aug to 21 Sep

57,844 landing visitors became 1,734 new paying customers, 3.00%. YouTube is the one channel that demonstrably converts: 9,636 site entries to 395 new subscribers (4.1%), up from 35 a quarter ago. X: 2,466 entries to 8. Untagged traffic (direct, search, anything without a tag) is 26,358 entries, most of them existing customers signing in. The funnel stages come from different systems and are indicative, not a strict sequence.

2.6 Product usage and support

Are they using it · of 7,862 paying

40.3% of paying subscribers signed in or opened the app in the last 30 days, 55.9% in 60 days; 12% re-ran the optimiser. Optimisation is set-once, so a low re-run rate is expected and is not a churn signal. Measured gain on before-and-after benchmark pairs: median 2.6%, mean 5.7%. The public claim is "up to 60%", the 99th percentile of measured runs, not a typical result. Product usage measurement is being reworked (telemetry decision, 18 Sep).

The Games They Play panel is built from game mentions in Discord, the only game signal available today. It reflects Discord-active users, not the paying base, and is being relabelled.

Support · AI-triaged Discord tickets, 22 Aug to 21 Sep

49 Discord tickets triaged in the window. Performance complaints (8 tickets) have a 0% resolution rate and the highest frustration score alongside billing. Four tickets expressed intent to cancel; two are unresolved.

3. The last 90 days: what happened and what it did

The pricing and trial changes moved the business permanently. The September activity moved visits and sign-ups for a month; how much of that becomes paying customers, and from which source, is what the tracking work is for.

4. Website conversion: done, found, to be done

Done

• Speed work implemented on the test site (itss.hypertune.gg): advertising and analytics tags now load after the page is interactive, the hero video downloads after the page loads, animations download after the page. GTmetrix grade C to B, performance 61 to 88, blocking time from about 1 s to 101 ms. Each of these is a trade: the video appears about a second later, and visitors who leave inside the first seconds are not counted by the ad pixels.
• Microsoft Clarity (heatmaps and session replay) installed. Google Analytics 4 tag installed and the Search Console property verified.

Found

• Six purchase buttons with four different labels and two destinations, no purchase button in the header, and nothing to buy at the bottom of the page. Sign-in method today: Google 67%, Discord 25%.
• The site makes performance and anti-cheat claims that have no written source or vendor confirmation. They stay until sign-off, but they are a risk once paid traffic arrives.
• Accessibility and copy: low-contrast text on the pricing card, heading order, button labels that do not match their visible text, one typo.

To be done (agreed 21 Sep)

• Purchase button in the header, visible in every window, and the pricing card repeated at the bottom of the page.
• The whole 50% widget clickable, not just its button.
• Game logos re-exported at the right size; hero video poster image so the page never shows a blank hero.
• The small fixes above.
• Button labels and a single destination: to be settled on a longer call, then A/B tested. Not changed now.

5. Dashboard and measurement: done, found, to be done

Done

• Company-wide analytics dashboard v1 live 17 Sep: money, subscribers, churn two ways, cancellation reasons by plan, retention cohort grid, funnel and channels, deal profit and loss, payment recovery, support triage, product usage, games audience. Subscribers are counted three ways and never merged: paying, past due, trialing. Custom date ranges work.
• CPM deal terms and per-video tracking links made mandatory for new creator deals.
• Go-to-market decision rules agreed: every channel test has a budget, a timeline and a success measure before it runs; creator spend has a sign-off gate.

Found

• 58% of revenue has no known source. Only 15% of recent subscriptions carry a campaign tag, and whether tags survive account creation into billing is unconfirmed. No advertising can be measured until that is confirmed or fixed.
• Cash Collected on the dashboard ($52,755) does not tie to Stripe ($53,409 gross, $49,159 net) for the same range.
• Trialing (361) mixes year-long creator comps with real 7-day trials.
• The 60-day LTV tile shows one June cohort on legacy pricing ($11.45) regardless of the date range. The planning LTV is $25 until measured post-July cohorts exist.
• Smaller definitional defects: two different "paying" totals on one page, funnel stages in the wrong order, channel totals that do not reconcile with the funnel, an "owed" figure Austin has flagged as wrong, the games panel mislabelled. Listed and with Kevin.

To be done

• Confirm tracking parameters (plan, campaign tags, click IDs) reach Stripe through account creation. First priority.
• Events between the button click and the purchase: sign-up viewed, started, failed, completed; purchase completed. Today only the click is recorded.
• Cash Collected reconciled to Stripe; Trialing split; the defect list above.
• Search Console and GA4 pulled into the platform daily. Ad spend, CAC and ROAS section (empty until ads run).

6. Creator campaigns

Creators are the anchor channel: 43.3% of invoiced revenue in the last window is creator-attributed ($26,264 of $60,725), up from 28.6% a quarter ago, and YouTube is the only source with a proven conversion rate.

Two kinds of arrangement run side by side.

Affiliates (revenue share). 174 standard affiliate accounts, 98 of them with revenue to date, earning a percentage of each payment their customers make. All-time: $114,688 attributed revenue from 9,918 customers, 2,993 still active. Commissions owed today $9,398; total paid out through the platform to date $40,348 (commissions and platform-paid deal fees together).

Paid deals. 22 creators on paid terms, tracked on the Deals P&L page. $63,933 committed, $31,075 paid, $32,859 still owed. Net revenue attributed so far $19,066 from 966 customers, programme return 29.8% with most CPM deals still accruing.

Past break-even: Mykhailo Zborovskyi (+$3,715), Helli (+$869), Kaan Demir (+$429), aloneintokyo, Hunter Schline. Carrying the spend: Yuki Aim ($15,000 paid, $62 revenue so far) and the Marvel Rivals team ($13,000 committed, month one, $0 revenue yet).

Austin is supplying the confirmed creator list with deal type, amount paid, date paid and start date; the Deals P&L page will carry start and end dates once it has them. Until then the figures above are the platform's and the dates are not shown.

7. SEO and AEO

Bobby (Robert Lally) is engaged. First working session 11 September. A meeting this week goes through all of his proposed changes and the timelines attached to them; his Hypertune-specific report follows that, then a presentation to Austin.

What he found. Effectively zero on-page SEO today: no keyword targeting in the copy, weak and duplicated meta descriptions, no structured data, video testimonials with no descriptions or creator names, leftover Comp Ready branding. Domain authority 16. Every ranking query is the brand name; 21 pages indexed, 347 not. Google's index still holds old copy promising a "full refund, no questions asked"; re-indexing is to be requested.

Direction. A manual on-page pass coordinated with Kamil first, then an ongoing mentions-and-links programme (journalist and blogger outreach beyond gaming press, the Reddit threads already discussing Hypertune, listicles for AEO, directory listings), knowledge panel and Wikidata off the press coverage. AEO (being cited by ChatGPT, Perplexity and Google's AI answers) is small today and growing; ChatGPT already sends about 130 visitors a quarter unprompted. Real results inside the first year, not the first month. Dependency: Search Console into the platform (Kevin) so branded vs generic search and the backlink record can be tracked on the dashboard.

8. Email, Google Ads and Facebook

A full strategy for email, Google Ads and Facebook and Instagram follows in a separate report in the coming week. The position today:

Email. 15,487 mailable profiles. Six live flows, agency-built, unchanged since 23 July. Email-attributed revenue is about 1% of cash collected. The 8 September win-back to 6,601 cancelled users returned 15 payments; the plain-text version signed by Austin produced 9 of the 15. First change, already specified: the payment recovery flow rebuilt to cover Stripe's full 7-day retry window (three emails and one in-app notification, no re-firing on every retry). A third of payment attempts fail and 46% of past-due accounts recover today.

Google Ads. No campaigns have run. Before spend: account access confirmed, connected to the GA4 property, conversion import from the purchase event, and tracking parameters confirmed into Stripe.

Facebook and Instagram. Never tested (26 visitors lifetime). Pixel already on the site. Plan: start soon after Google, with creator video as the creative, at a test budget with a stop-loss.

Sequence, decided 10 and 15 September: website conversion work and measurement first, then paid channel tests each with a defined budget, timeline and success measure, then referral mechanics. Reddit last.

9. Social media: an AI-native content process

Today social is run by hand and the numbers show it: X sends traffic that does not convert, Instagram and TikTok are near zero. The plan is to rebuild content creation and analysis as one process rather than a posting schedule.

• Research: pull what is already converting (the creator videos and settings guides behind the 4.1% YouTube conversion, the support-ticket themes) and generate content briefs from that, not from a calendar. Daily scan of what is trending across channels to decide what to quote, reply to and ride.
• Creation: AI-assisted drafts in the founder's voice from a documented voice guide, reviewed by a second model before anything is published, then a human sign-off. Every claim in a post comes from the evidence ledger.
• Distribution: every post carries a tracked link so it shows up in the funnel by source.
• Analysis: weekly read of which posts produced site entries and subscribers, fed back into the briefs. Content that does not produce entries is cut.

The aim is that what three or four people do in a marketing team, one person does with the process. It is being documented this week and starts on Hypertune once tracked links are in place.

10. Next 30 days

Notes on the numbers

• Windows are 31 days each, matching the dashboard's default view, so the three columns are like for like.
• "Paying" excludes trials and past due, and includes subscribers who have asked to cancel at period end.
• "Unattributed" revenue has no known source. It is not organic; it is a measurement gap that the tracking work closes.
• The measured press figure is site visitors from press referrers. Agency reach and impression estimates are not used.
• LTV of $25 is a planning number until measured 60-day cohorts on post-July pricing are available. The dashboard's $11.45 is one June cohort on legacy pricing and should not be quoted as the business LTV.