12 · Workshops and evening events: the owned-format craft

First principles (Priya Parker, operationalised)

The Art of Gathering doctrine, translated to fieldcraft:

Purpose before format. A dinner is not a purpose. "Get 10 Heads of Ops comparing notes on where AI actually broke in production" is. The pack's brief field forces it; if the purpose is generic, cancel the venue search.
Exclusion creates the room. "If everyone is invited, no one is invited." The guest list is the product — curation is the host's first act of generosity, not elitism.
Author the rules. Off-record by default, no pitching (by anyone, including us), one conversation at a table, phones away. Stated in the invitation and restated in the welcome — guests relax when the frame is explicit.
Give the room a job. Micro-roles: one guest kicks off the topic, another brings a war story at the main course, the host holds the closing round. People given an excuse to talk, talk.
Logistics never eat purpose. The venue hunt matters less than the question the table will work on. One hour on the discussion prompt beats three on the wine list.

The executive dinner

The math (why it's INTO's workhorse — sources in research-foundations): 80–90% show rates; 30–50% of attendees converting to qualified opportunity (vendor range — ceiling); $1–3K cost per qualified meeting, the lowest of any field format; $5–20K fully loaded. Senior buyers who decline vendor pitches accept peer conversations — the format is the access model.

Numbers that govern:

10–12 seats for the first editions (8 minimum — thinner feels sparse; 15 maximum — louder stops being one conversation). Above ~10, split across two round tables. Rounds beat longs.
Mix ≈ 60–70% prospects / 20–30% clients / 10–15% INTO (2 execs max). Clients aren't filler — a prospect seated next to a client hears validation no pitch can deliver. Client seats also serve S1 (relationship objectives). Client attendance is itself information governed by confidentiality rules: no client's presence is disclosed in advance to others without their OK.
Invite ~1.5× the seats across the campaign, expecting yes→seat decay of 20–30% even at exec show rates; confirm 12 for 10.

The invitation ladder (highest yield first): exec-to-exec personal note from the host (T-4w, one follow-up) → deal-owner/CSM personal invites → hand-written/physical card for the top tier → never a marketing blast. The invitation names the topic, the peer set ("11 operations and product leaders"), the rules (off record, no pitches), and the ask ("bring one place AI has actually failed you").

Venue: genuinely private room with a door; acoustics checked in person at the same weeknight (chapter 05 rule); ≤10 min from wherever guests already are (the conference venue, the business district); a table shape that lets everyone see everyone.

Conversation design (the Ortus-school mechanics):

First 15 minutes: welcome, the frame ("tonight is off the record; nobody sells; here's the question"), and one thought-provoking opener each guest answers in 60 seconds — everyone's voice in the room early.
The host moderates ~20% of airtime, guests 80%; nobody holds the floor past five minutes; the host's job is moving airtime, not lecturing.
One themed question per course works: where AI broke / what worked / what you'd do first with a clean slate.
INTO never pitches. The follow-up carries the commercial conversation; the dinner buys the right to have it.
Closing round: one takeaway each, and the host names the continuation ("we do this quarterly — you'll hear from me").

Follow-through (where the ROI actually lives): debrief within 24h; every guest in the CRM with the conversation note; personalised outreach within 48h referencing what they said; promises fulfilled first; 3–5 follow-up meetings is the target from a 10–12 seat table. "The dinner does not close the deal — the system around it does."

Cadence + co-hosting: quarterly per city compounds — repeat guests bring peers, and the table develops a reputation. Co-host with a complementary non-competing partner to split cost and double the draw (lead-sharing terms agreed in writing first); a client speaker anchoring the theme is the highest-credibility draw.

The workshop / masterclass

Format defaults: the 90-minute masterclass (the field's converged session length) for authority + pipeline; the half-day working session for late-funnel co-creation with one account. Cap at 12–20 — big enough for peer energy, small enough that INTO hears every company's problem.

Design backwards from the artefact. Attendees leave holding something they made: for INTO, a maturity self-score — an AI-readiness rubric applied to their own organisation in-session. The self-assessment produces the gap; the gap produces the conversation. This is the consulting-workshop deep logic (David C. Baker): show how you think, don't solve it for free.

Structure (90 min): frame + one sharp claim (10) → the rubric taught through two real case patterns (25) → guests score themselves, INTO circulates (25) → peer comparison in pairs/trios (15) → readback + the offer (15). Hands-on every 15–20 minutes; INTO keeps a copy of every worksheet (with consent) — it's the qualification record.

Gating = qualification. Free-and-open workshops no-show at 40–60%; paid at ~10%. For exec audiences, invitation-only + "request a seat" application achieves the paid-signal effect without procurement friction. The application asks two questions that pre-qualify: current AI state, and the workflow they'd fix first.

The conversion ladder (CEF-native): workshop → the standing dated offer made in the room ("a complimentary 45-minute readout of your self-assessment within two weeks") → readout call → Path A (Discovery & Definition) or Path B for those with a defined brief. The readout is the assessment-as-door-opener move; the paid diagnostic (Path A) is our productised version of it. Speed rules apply: the readout offer converts at multiples when booked within 24–48h of the workshop.

Materials: the rubric one-pager, two case patterns (anonymised per brand rules), the worksheet, and a follow-up pack (slides + notes) that doubles as the tier-B sequence artefact.

Evening side-events at conferences

Counter-programming someone else's show — the highest-ROI entry point at a major conference because you control the room, the list, and the talk track:

Night selection: never head-to-head with the official party or keynote night; take the under-programmed evening (often night one) or the pre-conference evening. Experienced attendees triage to 3–5 events a night — be the curated one, not the loudest.
Walking distance from the main venue or attendance halves. Book venues absurdly early — good private rooms in conference cities go 6–8 weeks+ out, earlier for flagship weeks.
Overbooking: curated dinner formats behave like exec dinners (confirm 1.2×); open-format drinks no-show like free events (40–60%) — overbook 2×, run the reconfirmation ladder (confirmation → T-7d → T-48h reconfirm ask → day-of), and the host's personal T-48h note remains the strongest lever.
A reason beyond the bar: a 20-minute fireside, a client speaker, one themed question. Generic open bars collect badge tourists and other vendors.
Co-sponsoring halves cost and doubles the list; agree lead-sharing upfront.
And when not hosting: RSVP the team across other people's side events — the side-event calendar is free hunting ground (chapter 04's stack).

Run-of-show discipline (all owned formats)

Minute-by-minute grid, one named owner per line, contingency rows written ("speaker no-show → host's reserve talk"; "AV fails → printed one-pagers + louder voice"). Venue access ≥2h before doors (dinners 1h); AV checked 60 minutes before the first guest; speaker check-in call at T-1w. The ROS is printed — wifi fails at exactly the wrong moment.