Orang

Orang Digest, week of October 5, 2026

This Week's Episodes

• RevOps Lab: Colin Gerber, VP RevOps & Strategy at Socure, returns to break down forecasting a fully consumption-based business - the solution readiness doc, the BAR realization process, and comp that keeps AEs honest on ramp.
• GTMnow: Meagan Eisenberg, CMO of Samsara (NYSE), on running an AI transformation inside a public company - a 2.5-week boot camp, 113 agents in production, and headcount held flat while output climbed.
• The LeanScale Podcast: Dan Cook, CEO of PDQ (ex-CRO at Lucid Software), on bolting a sales motion onto PLG without cannibalizing it - the accretion test, the RevOps ratio, and a CRM that started as a Google Sheet.
• The LeanScale Podcast: Luke McKinlay, Head of Finance at Nuvo, on why nobody agrees what ARR is and how to report usage revenue without lying or sandbagging - full write-up in Notion.
• Topline: Manny Medina, CEO of Paid and founder of Outreach, on pricing AI agents once seats stop making sense, and why software margins are heading toward 40-60% - full write-up in Notion.

The 3 episodes below are this week's deep dives; the other 2 are written up in full in the Notion archive.

Executive Summary

• Consumption pricing pushes RevOps past closed-won: at Socure, every booking checks against a Salesforce "solution readiness" object (decisions/month x modules hit x hit-rate x price), targeting 90% of booked revenue realized by month 12. Inflate bookings and you carry a forecast hole for 24 months. Colin Gerber on RevOps Lab.
• AI's payoff in GTM is a production shift, not a new channel: Samsara held marketing headcount flat at 260 while building 113 agents, cutting ABM landing pages from 2-3 weeks to under 30 minutes and a full rebrand go-to-market from six months to 3.5 weeks. Meagan Eisenberg on GTMnow.
• Adding sales to PLG works only if accretive: Dan Cook (LeanScale) carves an enterprise SKU only sales can sell, then proves lift with a control group of untouched accounts. Otherwise it just adds commission cost to revenue already closing.
• Usage revenue breaks ARR and quota math: with 50-60% of revenue variable, Luke McKinlay (Nuvo) books conservatively and pays reps on commit then trues up, since nothing forces a customer to pay month two.

Framework of the Week

Socure's booking-substantiation and realization system, from Colin Gerber (VP RevOps & Strategy) on RevOps Lab - the cleanest answer this week to "how do you forecast when closed-won means almost nothing."

Three connected motions:

1. Substantiate the booking. A custom Salesforce object (one "solution readiness" record per use case) computes expected revenue bottom-up: decisions per month x which modules each decision hits x percentage of traffic reaching each module step x contracted price per module. Example: an onboarding flow where step one fires on 100% of traffic, step two on 50%, step three on 25%, each priced, times monthly decision volume. The rep still calls a number, but if it diverges from the object's output, that gap is resolved before anything books. "Done" = booked and computed numbers roughly agree.

2. Model the ramp (BAR - booked ARR realization). Book at the annualized full-ramp number, but schedule it month-by-month against real seasonality: BNPL and gaming customers freeze code in Sept/Oct and spike Nov-Feb, so the curve reflects go-live month and season. "Done" = a per-opportunity realization schedule, not a flat 1/12th.

3. Review realization monthly with FP&A. Each booked cohort gets rated green (on ramp, graduate off the list), yellow (behind but a path exists, adjust forecast), or red (won't materialize, write down and back out). "Done" = every cohort dispositioned and forecast adjusted.

First step this week: build the bottom-up calc for your five largest open consumption deals and compare it to the rep's number. Measure success by cohort realization at month 12, target ~90%. Failure mode Gerber names: letting bookings drive comp too heavily, which rewards inflated bookings that never realize.

Benchmarks & KPIs Mentioned

• ~90% of booked ARR realized by month 12, cohorted - Gerber's consumption-forecasting accuracy target; hold your own realization curve against it and audit anything realizing at 60%.
• ~70% of variable comp tied to revenue (not bookings) outside the AE team at Socure - a reference point for how heavily to weight realized revenue vs. bookings in consumption comp.
• AE payout over 24 months / 8 quarters, clock starts at go-live - a model for comping ramp in consumption deals so reps stay motivated through slow burn.
• RevOps staffing: ~6:1 seller-to-RevOps early, widening to ~30:1 at scale (hundreds of sellers) - Dan Cook's benchmark; over-invest early because systems and data are the foundation for scale.
• 4x+ sales payback as the gate to add reps - LeanScale/Cook's rule for shifting capital from PLG marketing into sales headcount.
• ~3-6 month sales ramp assumed, 6+ months in reality - McKinlay/Enrico's correction to capacity models; plan ramp longer than your spreadsheet wants.
• Software gross margins heading to 40-60% (vs. the SaaS-era 70-80%) - Medina's claim for the agent era; a planning input if you sell AI with real inference cost.

Across the Shows

Consumption and usage pricing is rewriting RevOps, finance, and comp, and most teams haven't caught up. Colin Gerber (RevOps Lab) and Luke McKinlay (LeanScale) reach the same discipline from opposite seats. Gerber's rule: book pre-launch startups at only their contractual minimum, since founders promising hockey-stick growth leave you owning the forecast hole for 24 months. McKinlay's rule: when 50-60% of revenue is variable, treat it conservatively and protect the downside, because nothing forces the customer to pay in month two. Both reject the idea that bookings stop mattering under consumption pricing: bookings are still the goal, but you must substantiate them against realized behavior. The action for revenue leaders is the same: instrument realization by cohort and stop trusting the signed contract as the forecast.

Manny Medina (Topline) pushes this further: if agents do the work, you can't price seats, and a token is the wrong unit of value since it's an internal lab metric, not something a buyer can inspect. His transparency ladder (show the work, explain the value, then argue ROI) is a useful sales framework, but his sharper claim is the disagreement worth flagging: SaaS's 70-80% margins were an artifact of cheap money, and normal software will run at 40-60% as inference costs stay high, against the default CRO instinct to chase traditional margins. His weaker point is outcome pricing: he admits attribution is "really hard," meaning most agent pricing still sits between effort and output, not true outcomes. Treat outcome-based pricing as a 2027 conversation, not a 2026 plan.

One connective point across Cook and McKinlay (both LeanScale): fewer, better-utilized reps beat stacking headcount. Cook's control-group test and McKinlay's "don't hire until the calendar is full" converge on one discipline: prove accretion before adding cost.

The Deep Dives

The three episodes worth your time this week, with the mechanics behind the takeaways above.

RevOps Lab: Gerber on Consumption Forecasting

Socure prices ~35 modules by API-call consumption, so a signed contract is only potential revenue: the real work starts at closed-won. Gerber's team closed the forecasting gap three ways. First, substantiation: a Salesforce custom object computes expected revenue from decisions/month x modules-per-step x hit-rate-per-step x price, and if the rep's called number diverges, it's challenged before booking. Reps still own a number; the object just keeps it honest. Second, the lag between closed-won and go-live is largely solved with "live trials," where customers are fully implemented in the real environment before signing, moving the integration delay that wrecks consumption forecasts upstream.

Third, the BAR (booked ARR realization) process: book at annualized full-ramp, but model the ramp month-by-month against seasonality (BNPL/gaming code freezes in autumn, spikes Nov-Feb), then run a monthly review with FP&A that rates each cohort green/yellow/red and adjusts the forward forecast or writes down dead bookings. Four sources feed it: Salesforce, Jira, NetSuite, and Tableau as the single pane.

His two warnings: book pre-launch startups at contractual minimum only, and don't over-weight bookings in comp, or you manufacture inflated bookings that never realize. His controversial close: RevOps should own the forecasting system and guardrails, not forecast accuracy - that belongs to sales management, who are actually on the deals.

Takeaways & Implications

• If you're in RevOps: build a bottom-up substantiation object for consumption deals so the booked number is derived from expected usage behavior, not rep optimism. Compare it to rep-called numbers on your largest open deals this week; divergence is your pre-booking conversation and your path to ~90% cohort realization by month 12.
• If you're a CRO: own forecast accuracy yourself and push it to the sales managers closest to the deals - RevOps lays the tracks, but accuracy sits with whoever signs off. Weight comp toward realized revenue, not bookings, to avoid incentivizing forecast holes you answer for over 24 months.

GTMnow: Eisenberg on AI Agents

Samsara's CMO frames AI as different from past shifts: social changed the channel, PLG changed the buyer, Covid changed the venue, but AI changes how the work itself gets produced. The operational core for a revenue leader is the change-management sequence and the output math, not the agent count. Her order: lead by example (she built agents, posted prompts in Slack), evolve "AI power hours" from watch-alongs into hands-on sessions, then run a 2.5-week boot camp - 1.5 weeks self-paced coursework, then daily in-class sessions taught by her own marketers, ending in a build-and-present assignment. She democratized agent-building into every function rather than centralizing an AI team, on the logic that expertise lives in the functions.

The results are the portable part: 113 agents across 260 people, headcount flat, ABM landing pages from 2-3 weeks to under 30 minutes, content from weekly to about 10 a day, and a rebrand that hit market in 3.5 weeks versus six months with an agency. The trade she made explicit with her team: time invested in learning must convert to more output or faster delivery, same people and same timeline is not the deal. She also wired AI into performance: no top calibration score (4 or 5 of 5) without actively building agents, and she interviews for AI fluency by asking what agents candidates have built and what the outcome was.

Takeaways & Implications

• If you lead marketing: sequence adoption as learn-by-doing before cost caps, and democratize agent-building per function rather than centralizing. Expect compressed cycle times (landing pages, content, rebrands) that you must convert into more output or faster delivery, or you just add tool cost.
• If you're a CRO: make AI fluency a scored expectation, not a nice-to-have. Tie top performance ratings to demonstrated building and interview for it by asking what agents a candidate shipped. This is how you keep headcount flat while raising output, but the expectation-setting is yours to own.

The LeanScale Podcast: Cook on PLG + Sales

Dan Cook built Lucid's revenue org from 30 engineers and himself. His first call - an Uber IT lead asking "how much can I get for 15K" - exposed the core PLG tension: hundreds of users already there, no salesperson needed. The question before hiring a single rep: is sales accretive, or just cannibalizing revenue that was already closing? At Lucid the answer was a paywalled enterprise SKU only sales could sell, creating clean attribution. At PDQ the accretive lever was cross-sell across acquired products, not expanding the initial land, and that thesis elongated cycles.

The mechanics worth copying: Smartsheet's "ghost book", a set of accounts no rep may touch, run as a control group to prove sales is additive against the status quo, not just reps against each other. On capital allocation, nail PLG first (SEO, then AEO, then conversion optimization) and gate sales expansion on 4x+ payback. Cook's own forecast was wrong in the right direction when his first four reps 2x'd his pace. On infrastructure: invest in RevOps early (~6:1 seller-to-RevOps, widening to ~30:1 at scale), and build your first CRM from what reps organically log plus what leaders must measure. Lucid's V1 was a Google Sheet, and nailing a repeatable process before systematizing beats an expensive out-of-the-box implementation.

Takeaways & Implications

• If you're a CRO: before layering sales onto PLG, run a control group of untouched accounts to prove accretion, and carve an enterprise SKU only sales can sell so attribution is clean. Gate rep expansion on 4x+ payback, otherwise you just add commission cost to revenue that was already self-serving.
• If you're in RevOps: over-invest early (target ~6:1 seller-to-RevOps) and design your CRM from what reps actually log plus what leadership must measure, not an out-of-box template. Nail the repeatable process before you systematize it; a clean Google Sheet beats a Frankenstein Salesforce you unwind years later.

Also Processed

• The LeanScale Podcast: Luke McKinlay (Head of Finance, Nuvo) on why four people in a room hold four different ARR numbers, and how to report usage revenue without sandbagging or lying: align on one definition.
• Topline: Manny Medina (CEO, Paid; founder of Outreach) on pricing agents by effort/output not tokens, why fixed-price deals stall at renewal, and software margins heading to 40-60%.

Full write-ups are in the Notion archive.

Worth Your Headphones

Topline: "Turning AI Tokens Into ROI" with Manny Medina. The digest captures his frameworks, but the room-energy of a second-time founder riffing live on margins, fundraising regrets, and the Satya Nadella "life is path dependent" green-room moment is worth the hour on its own.