State of the Business: September 2026

One-page monthly snapshot. Refreshed on the 1st of each month from meetings, decisions, inbox, metrics, and the git log.

Where each entity stands

• Hard Calls, renamed Anyhouck (2026-09-04): Newsletter repositioned from decision quality (the how) to leverage (the what). Domain anyhouck.com bought, redirects to michaelhouck.co/newsletter. Brand structure settled 2026-09-02: Michael Houck is the brand, Hard Calls is the ethos (a format inside his newsletter, not a standalone publication), Hockey Stick is its own vertical again (reversing the Aug 8 fold-in). Active subs drifted down across the month: 58,129 (W36) to 57,047 (W37 reading). The Aug decline that triggered concern has not reversed; still not root-caused.
• Megaphone (+ Studio): Studio price raised to $5,000/mo for both Megaphone and JOLT (2026-09-22), from $3,995. Prompt engagement format overhauled 2026-09-01 (one-liner + thread, thumbs up/down, human tag on engagement) to fix near-zero client engagement; no evidence yet the fix shipped into the actual cron jobs or moved the needle (see Decision Follow-Through). Notion replaced by the client portal for editing/approvals (2026-09-22). Three clients churned in one week: Pierce Freeman, Brentan Alexander, and Fasset/Rafi confirmed 2026-09-21; Roebling churned separately mid-month (2026-09-16, contract ended Sep 20). New client Adam Gillman closed and paid Sep 19, kicked off Sep 22. Manse and Pally (Haz Hubble) in active proposal/kickoff stage end of month.
• JOLT!: Highest build velocity of the month (64 GitHub commits in W38 alone). Equity and comp structure with Tatem finalized 2026-09-21 after two same-night negotiation calls; OA due to be signed by Sep 25. New clients: Decade Labs (TikTok-first, stealth founder) and Manse, both kicked off week of Sep 22-24. TikTok breakout alert moved to weekly cadence (decided Sep 1); not confirmed whether the cron itself was actually changed.
• Lore: Two-tier pricing set 2026-09-10 ($2,500 / $4,500 per month). Aeon proposal (website rebuild + comprehensive retainer) scoped 2026-09-24, due out to the client over the Sep 27-28 weekend. Realworld GEO engagement ongoing. Pipeline script tests broke 2026-09-26 (real regression, unpatched at month end).
• Hockey Stick: Relaunched as its own vertical under the Sep 2 brand ruling, reversing the Aug 8 fold-into-Hard-Calls decision. Same offer and pricing as before ($12,500 base, $2,495/mo retainer), fronted by Max Kelly. Still pre-launch; no revenue this month.
• Rye Valley (holding): Payments/contracts platform migrated from Stripe + PandaDoc to Agree.com (decided Sep 16, contract signed Sep 23 with fees fully waived). Ben Gusberg’s departure confirmed: leaves all recurring roles Oct 19 after 2.5 years, for an EIR role at qsbsrollover.com. Blake absorbs Augment AM; Ben retains Clipwing review until a JOLT-specific AM is hired. People count drops from 9 to 8 effective Oct 19.

The month in numbers

Four weekly leverage snapshots (W36 Sep 6 through W39 Sep 27). Kick weekly figures restate materially 7-14 days after close; treat all weekly NOI as directional, the monthly settled figure as the only reliable read.

• Net operating cash flow by week: W36 -$29,460 (payroll week), W37 -$3,390, W38 +$30,318 (inflated by roughly $50k of likely non-operating/transfer flows, never fully confirmed), W39 -$10,952 (another heavy payroll week, $26,772 to Odyssey Trust/Thera). Sep MTD through day 27: +$56,000 operating NOI on $191,687 cash in against -$135,686 cash out, a clear improvement over August's settled +$2,760 (restated to +$2,071 in the Sep 27 report). Capital cash flow for the month ran -$91,445, unconfirmed as owner draw vs. miscategorized expense.
• Revenue per person (T4W annualized, 9-person base): swung from ~$126,681/yr (W36) down to ~$90,413/yr (W37) up to ~$147,663/yr (W38) up to ~$169,768/yr (W39). The swings are driven by which settlement week sits in the trailing 4-week window, not real revenue change; Ben’s Oct 19 departure will mechanically lift this metric by roughly $18-21k/yr once the denominator drops to 8.
• Meta Ads CPL: held in the $1.66-$2.27 range most of the month (W36 $1.66, W37 $1.92, W38 token expired/no data, W39 $2.27), a genuine step-change down from August's $2.95-$6.85 range. Cause still unconfirmed; nobody has documented what changed.
• web_traffic_stats.py: broken the entire month, now 7 consecutive weeks with zero PostHog website traffic data collected. Chronic, unresolved, flagged every week.
• Brain build velocity: 45 (W36) -> 29 (W37) -> 27 (W38, plus 118 tracked GitHub commits across repos that week, led by JOLT’s 64) -> 32 (W39). Steady cadence, no stall.
• Inbox grew every week with no net filing: 148 (Sep 13) -> 172 (Sep 13 reading) -> 208 (Sep 20) -> 237 (Sep 27). Oldest item now dates to Jul 29, over two months unfiled.

Open strategic threads

1. Hard Calls/Anyhouck subscriber decline, still unexplained after two months. First flagged in August, continued through September (58,129 to 57,047). No root cause identified, no beehiiv send-level diagnostic has been run despite being recommended in the Sep 6 scoreboard.
2. web_traffic_stats.py, 7 straight weeks broken. Every weekly scoreboard and two agent-manager reviews have flagged this with zero action. This is the single most repeated open item in the entire month’s record.
3. Ben’s Oct 19 departure needs a named handoff owner before he leaves. Handover notes were promised “next week” as of Sep 18; no handover document has surfaced in the brain as of month end, with under three weeks of runway left.
4. Capital cash flow -$91,445 for September is unconfirmed. Could be fine (owner draws) or a real miscategorization distorting the NOI picture. Nobody has pulled the Kick transaction detail to confirm.
5. Curator auxiliary model has been retired for 8 consecutive weeks, blocking all skill consolidation including the pending hermes-config-ops fold-in. A single hermes config set command would unblock it; still unactioned.
6. Inbox backlog has nearly tripled since mid-August (88 to 237) with no triage pass scheduled. It is functioning as a running capture log, not a queue that drains.

Biggest risks

1. Three Megaphone clients churned in one week (Pierce, Brentan, Fasset/Rafi, Sep 21), on top of Roebling churning Sep 16. Four churns in a single month is the highest client-loss rate recorded in the vault. The Sep 1 prompt-engagement overhaul was meant to address exactly the disengagement pattern behind these churns; there is no evidence in the brain that the fix actually shipped before the churns were confirmed three weeks later.
2. Ben’s departure risk is live and under-covered. No handover document exists yet for a person leaving in under three weeks, holding Clipwing review and Augment account knowledge that Blake has not yet absorbed.
3. Chronic data blindness compounding. Zero website traffic data for seven straight weeks, capital cash flow unconfirmed, per-entity revenue split still not available from Kick. The business is making pricing and staffing decisions (Studio price raise, Jolt equity) without the traffic or entity-revenue signals that would normally inform them.
4. Meta CPL improvement is undocumented and therefore fragile. A genuine cost efficiency gain (sub-$2 CPL most of the month, down from $3-7 in August) has never been traced to a specific creative or audience change. If it reverts, there will be no record of what to restore.

← Voice-mismatch and AI-sounding complaints are a leading churn indicatorSales/discovery call prep (live since 2026-07-30) →