Case studyDecision log

Decision log

Every call behind the program, with the options on the table and why. I made each one; AI researched the market and laid out the options. Search, or browse by topic, and open any decision for the reasoning.

110 decisions

1StackProject setup
Stack: Postgres (Supabase, with pgvector), Next.js on Vercel, Claude API
Options considered
Hosted vs local Postgres; separate app vs portfolio site
Why
Simplest thing that works; hosted database from day one
Link to decision 1
2Synthetic data only, from a fixed seed; one command rebuilds itProject setup
Synthetic data only, from a fixed seed; one command rebuilds it
Options considered
Hand-built fixtures; random data per run
Why
Reproducible evals; nothing resembles any real company's program
Link to decision 2
3Tiers Member, Select, Premier, Elite; Elite is invite-only and unmarketedTiers
Tiers Member, Select, Premier, Elite; Elite is invite-only and unmarketed
Options considered
Three tiers; revenue-banded names
Why
Mirrors how mature programs reward behavior while keeping a quiet top tier
Link to decision 3
4AI-generated SQL runs as a read-only role with a 5-second timeout and a 500-row capProject setup
AI-generated SQL runs as a read-only role with a 5-second timeout and a 500-row cap
Options considered
App-level checks only
Why
The database, not the prompt, must refuse harm
Link to decision 4
5The fictional vendor is Cloud Cruz; Tony is its partner operations lead running the Ecosystem Partner ProgramChanged later by 108Project setup
Later changed by decision 108. Read it for the rule as it stands.
The fictional vendor is Cloud Cruz; Tony is its partner operations lead running the Ecosystem Partner Program
Options considered
Unnamed vendor; Tony as outside advisor
Why
"Context is key": the work reads as real only with a company and a role behind it
Link to decision 5
6Builds live in this public repo, one folder per build, a tagged release per ship dateProject setup
Builds live in this public repo, one folder per build, a tagged release per ship date
Options considered
One repo per build; making the portfolio repo public
Why
One readable history; the shared foundation lives once
Link to decision 6
7Cloud Cruz starts with no partner programProject setup
Build 1 is a 0-to-1 Program Blueprint; the synthetic data becomes Year 1 of that program; each later build maps to lifecycle stages
Options considered
Add a standalone blueprint and keep a mature dataset; describe 0-to-1 in writing only
Why
Shows both halves of the job: designing a program and running it
Link to decision 7
8Seven lifecycle stagesChanged later by 34Lifecycle
Later changed by decision 34. Read it for the rule as it stands.
Seven lifecycle stages: Acquire, Onboard, Activate, Enable, Hand off / Co-sell, Maintain, Optimize
Options considered
(stage definitions still to be decided)
Why
The partner journey end to end, the shape of a bow-tie funnel
Link to decision 8
9Salesforce setup ships inside the Blueprint ("How I set up Salesforce for this"), with a sample of the dataProject setup
Salesforce setup ships inside the Blueprint ("How I set up Salesforce for this"), with a sample of the data
Options considered
Its own build; after the demos
Why
The design and the system that runs it ship together
Link to decision 9
10ScheduleChanged later by 103Project setup
Later changed by decision 103. Read it for the rule as it stands.
Schedule: Blueprint Oct 9, Measurement Oct 23, Propensity Nov 13, Answer Engine Dec 11, Year-1 review Dec 18
Options considered
Keep the Diagnostic on Oct 9; slip everything a week
Why
Design before review; no other published date moves
Link to decision 10
11The Ideal Partner Profile is grounded in public research on developer-platform, AI and data-platform partner ecosystemsMarket and customers
The Ideal Partner Profile is grounded in public research on developer-platform, AI and data-platform partner ecosystems
Options considered
Invent partner types from scratch
Why
Real programs are benchmarks, never presented as Cloud Cruz partners
Link to decision 11
12Benchmark set expanded to the large cloud, AI-lab and data-platform partner programsMarket and customers
Benchmark set expanded to the large cloud, AI-lab and data-platform partner programs
Options considered
Model only the developer-platform programs
Why
Hyperscaler, AI-lab and data-platform programs share the architecture Cloud Cruz should scale down from. Guardrail: public sources only; Cloud Cruz copies no program's names, tiers or numbers
Link to decision 12
13Cloud Cruz is an AI data platformMarket and customers
"the data and AI platform your apps and agents run on." ISVs embed it. Partner types: SIs and consultancies, ISVs (build on, embed, integrate, including ERP and CRM connectors), agencies, cloud marketplaces (AWS, GCP, Azure), resellers
Options considered
Developer platform; AI platform for finance teams
Why
Fits the partner types that drive AI and data platforms (SIs, ISVs, cloud marketplaces), reaches finance software through ERP and embedded tracks, and fits the Foundation data with the least rework. Trade-off accepted: no accountant channel. Research: blueprint/research/
Link to decision 13
14Ideal Customer ProfileMarket and customers
Primary: B2B software companies of roughly 200 to 2,000 employees shipping AI features and agents inside their own products; buyer is the CTO or VP Engineering; trigger is a "ship AI this year" mandate; partners pulled in are ISVs and agencies. Secondary: enterprise data teams (buyer: CDO or Head of Data), reached through SIs and cloud marketplaces
Options considered
Enterprise data teams first; two ICPs from day one
Why
One focused customer makes the partner journey designable; enterprise stays reachable through the partners who already sell there. Follow-on for the Year-1 data: shift the account mix toward mid-market
Link to decision 14
15Recruiting order: services first, then technologyChanged later by 97Launch and targets
Later changed by decision 97. Read it for the rule as it stands.
A 90-day beta (Jan 1 to Mar 31, 2027) with about 25 design partners: about 20 agencies and boutique AI consultancies plus 3 to 5 ISVs as integration design partners, to get the motion down before launch. Launch (Apr 1, 2027) opens the Services path; the Technology path (ISV embed and integrations, including ERP and CRM) opens mid-Year 1 (about Oct 2027); resellers and cloud marketplace co-sell come in the second half
Options considered
Technology first; both paths at launch
Why
Services partners create the first production customers, which every benchmarked program treats as the proof point for Activate; one onboarding journey to perfect during the beta
Link to decision 15
16The program is architected for partner communications (Slack), partner support (Zendesk or similar) and a Support HubSystems
The program is architected for partner communications (Slack), partner support (Zendesk or similar) and a Support Hub
Options considered
(tool choices still to be decided)
Why
Partners need a channel, a ticket path with SLAs and self-serve answers from day one of the beta; the Support Hub is where build 4, the Answer Engine, lives
Link to decision 16
17Partner comms: hybrid SlackSystems
Beta: a Slack Connect shared channel per design partner. Launch: one partner community workspace for everyone (announcements, help, tier channels). Premier and Elite keep a dedicated Slack Connect channel
Options considered
Slack Connect only; community workspace only
Why
High touch for 25 design partners without a model that breaks past 100
Link to decision 17
18Partner support: Zendesk synced to SalesforceSystems
Every ticket shows on the partner record and feeds partner health
Options considered
Salesforce Service Cloud; Intercom
Why
SLA-driven escalations plus one partner view in the system of record
Link to decision 18
19Support Hub: version-controlled program docs answered by the AI Answer Engine (build 4) with citationsSystems
Unanswerable questions open a Zendesk ticket; urgent issues escalate to a person in Slack
Options considered
Zendesk help center; a portal hub page
Why
Self-serve answers partners can trust, with a human path every time; the Answer Engine demo lives where partners would actually use it
Link to decision 19
20PRM: introw.ioSystems
PRM: introw.io, synced two ways with Salesforce. Onboarding, training (LMS), community and comms, deal registration, MDF and commission calculation live there. The demo uses introw's free Starter plan, which takes one partner live, so the beta motion is shown end to end for one design partner
Options considered
PartnerStack; a Salesforce partner portal plus Tipalti; Crossbeam-first with a light PRM
Why
One partner platform that automates the whole journey, built on the CRM, with a real free tier to demonstrate on (introw.io/pricing, introw.io/integrations/salesforce)
Link to decision 20
21Payouts: TipaltiMoney
for accounts payable, partner KYC and payment risk. Commission rule: a commission is earned when the customer's invoice is paid and closed, held 60 days, then released in prorated amounts, annualized at the partner's commission rate. Accruals need a system of record (to be decided)
Options considered
Paying at deal close; PRM-native payouts
Why
Pay only on collected cash; KYC and tax compliance handled by a payables platform
Link to decision 21
22Credentials: CredlyCredentials
is the named production choice. Partners with an existing Credly account receive Cloud Cruz badges in the wallet they already use for their cloud certifications. Training happens in introw; the badge is the third-party validation. The demo issues badges from a free Open Badges 3.0 issuer (to be chosen)
Options considered
Keep certification inside the PRM; Skilljar
Why
Industry-recognized, third-party-verified credentials ride the hyperscalers' existing badge network
Link to decision 22
23Account mapping: Crossbeam from launchSystems
for partner overlap, co-sell targeting and co-marketing, assuming partners use it too
Options considered
Mid-Year 1; Salesforce reports only
Why
Feeds the Hand off / Co-sell stage from day one; partner alliance and partner account managers expect it
Link to decision 23
24Commission accruals live in a Salesforce ledgerMoney
(a Commission Accrual object on the partner and opportunity records: Earned, Held, Released). introw calculates; Tipalti pays released amounts; finance books the GL accrual in the ERP
Options considered
Finance ERP as the ledger; introw only
Why
One view partners, sales and finance all share
Link to decision 24
25Commission model, three mechanisms plus acceleratorsChanged later by 93Money
Later changed by decision 93. Read it for the rule as it stands.
(1) Month-to-month and pay-as-you-go: a percentage of each customer invoice once it is paid and closed, to reward retention and expansion. (2) Annual and multi-year: annualized and prorated, paid as Cloud Cruz is paid, so the company is protected and larger deal structures are rewarded. (3) Lump-sum SPIFFs after the hold, for milestones and thresholds. Accelerators raise the rate above thresholds. The 60-day hold after an invoice is paid applies to all
Options considered
A single per-invoice rule; a single annualized rule
Why
Each deal structure gets the incentive that fits it: recurring usage rewards retention, multi-year rewards commitment, SPIFFs reward milestones
Link to decision 25
26Implementation package, partner program onlyChanged later by 45Money
Later changed by decision 45. Read it for the rule as it stands.
When a partner lands or influences a customer who needs onboarding, Cloud Cruz prepays 5 consulting hours at $150 and 5 hands-on technical hours at $200 ($1,750 per customer) to offset the partner's startup cost, and the customer gets a 90-day free window until launch
Options considered
Standard MDF only; no implementation support
Why
Removes the partner's first-project risk and the customer's pre-launch cost; the benefit is only reachable through the program, which is a reason to join
Link to decision 26
27Credential ladder, badges issued at lifecycle stagesCredentials
(demo issuer: Certifier; production: Credly). Onboard: Sales Accreditation and Technical Accreditation. Enable: Certified Level 1 after the products and services training plus an accreditation. Levels 2 and 3 split by role: a technical track (hands-on implementation) and a business track (sales and project managers: value proposition, uncovering opportunities, how to partner with Cloud Cruz)
Options considered
Sertifier; one flat certification
Why
Badges mark real progress through the lifecycle, and higher levels match the partner's role
Link to decision 27
28Ideal Partner Profile segmentsPartners and scoring
(four), keyed to one test: can the agency build on an AI data platform? Drawn from the public service tags in the mediasense Global Agency Family Tree 2026 (cited, not reproduced). (1) Holding-company data and digital units: invite-only Elite, white-glove onboarding, the route to high-profile commercial clients. (2) Consultancies' digital arms: strategic SIs, the enterprise route. (3) Independent digital and data agencies: Premier and Select targets. (4) Boutique AI consultancies and build studios: the beta cohort and volume tiers for the mid-market ICP
Options considered
Segment by holding company; by service tag alone; by client industry
Why
Follows Cloud Cruz's own ICP and fit score. Provenance guardrail: derived fresh from public research and the decisions in this log; reproduces no company's material, names, criteria or target lists
Link to decision 28
29Fit score (100 points; recruit at 70+ with no category at zero)Partners and scoring
after two gates (an AI or data delivery practice; 3+ mid-market B2B software clients): client-base quality (marquee and ICP-fit clients) 25, propensity to refer 20, propensity to learn 20, delivery maturity 20, average project value 10, customer count 5. Pre-join proxies: referral = activity in other vendors' programs and co-sell history in Crossbeam; learning = certified people and hyperscaler Credly badges
Options considered
The original five signals; custom weights
Why
Adds the three signals that predict a productive partner (will they refer, will they skill up, do their clients matter) while the no-zero rule stops any one strength from carrying a weak partner
Link to decision 29
30Real agencies appear only as segment examplesPartners and scoring
in research and the Blueprint, citing the mediasense tree. The Year-1 data and any demo hit list use invented partners
Options considered
Publish a named hit list; no real names at all
Why
Never implies a real agency is a Cloud Cruz target or partner
Link to decision 30
31Segment names: Anchor, Scale, Specialist, BuilderPartners and scoring
Anchor: holding-company data and digital units (invite-only, white glove). Scale: consultancies' digital arms (the enterprise route). Specialist: independent digital and data agencies. Builder: boutique AI studios and consultancies (the beta cohort)
Options considered
Flagship / Systems / Specialist / Studio; Flagship / Fleet / Harbor / Crew
Why
Names say the role each segment plays for Cloud Cruz, and belong to this program alone
Link to decision 31
32Tier names stay Member, Select, Premier, EliteTiers
Member is the free entry tier; Elite is invite-only and drawn from the Anchor segment
Options considered
A Cloud Cruz-specific ladder; three tiers
Why
Industry-standard vocabulary (the large cloud and data platforms use the same words) that partners read instantly; distinctiveness lives in the criteria
Link to decision 32
33Year-1 tier criteria: counts first, revenue laterChanged later by 92 and 109Tiers
Later changed by decision 92 and decision 109. Read it for the rule as it stands.
Member: free, an agreement and a named contact. Select: 4 certified people (2 sales accreditations, 2 technical), 3 live joint customers, 1 public reference. Premier: 10 certified (3 at Level 2+), 10 live joint customers, 3 references, $175K partner-sourced revenue. Elite: invitation from the Anchor list, 25 certified (8 at Level 3), 25 live joint customers, 6 references, $500K sourced, a joint business plan with an executive sponsor. Promotions reviewed Jan 1 and Jul 1; demotions Jun 30 with 90 days' notice; each partner sees a live scorecard of criteria against actuals
Options considered
Keep attributed-revenue tiers; a 100-point hybrid score; lower or platform-level bars
Why
Calibrated to the entry and middle tiers of the large public programs; a new program lacks the telemetry for consumption-based gates. Research: blueprint/research/platform-programs.md
Link to decision 33
34Stage 0, Develop, precedes AcquireLifecycle
Stage 0, Develop, precedes Acquire (eight stages, one owner each). Develop: the partner development rep (PDR) generates pipeline through targeted outbound and inbound partner-qualified leads, up to a meeting with a director-level-or-above decision maker or a deal-in-hand handoff. Acquire: qualify against the IPP gates and fit score, then sign. End state: an agentic inbound and outbound PDR, built only after the motion is run and understood by hand. Amends decision 8: seven stages become eight
Options considered
Fold Develop into Acquire; treat Develop as a function, not a stage
Why
Instrumenting Develop shows which acquisition channels work (with partner marketing) and makes PDR activity visible; learning the motion first is what makes automating it safe
Link to decision 34
35Lead scoring: fit x intentChanged later by 68Partners and scoring
Later changed by decision 68. Read it for the rule as it stands.
Fit is the IPP score (decision 29); intent is a partner-qualified-lead score, co-designed with Tony later. Crossed, they route nurture, warm, hot (same-day) or urgent. Deal in hand is an automatic urgent fast lane: human-in-the-loop urgency rises, a Slack thread alerts stakeholders, the PDR stays in the loop while a partner manager and the AE join, and the deal registration is submitted during onboarding
Options considered
One combined score; intent only for inbound
Why
Keeps why a lead is hot visible; a partner arriving with a live deal is the highest-value moment in the program and gets people on it immediately
Link to decision 35
36PDR rules of engagementLifecycle
outbound targets director level and above at agencies in the four segments; multi-threaded with 3+ contacts (executive sponsor, delivery or technical lead, sales or growth lead) with tailored messaging; a no-collision check in Salesforce and Crossbeam before outreach, routing through any existing owner; qualified meetings and deal-in-hand leads handed off within 1 business day with context in Salesforce. PDR metrics: qualified leads, meetings held, warm handoffs
Options considered
Fewer rules
Why
Clear ownership prevents two Cloud Cruz people pitching one agency; multi-threading makes relationships survive one contact leaving
Link to decision 36
37Fit line at AcquirePartners and scoring
70+ approved automatically; 50 to 69 goes to a partner manager, who can admit with a logged reason (reviewed at Optimize); under 50 goes to nurture
Options considered
A hard gate at 70; the score as advisory only
Why
Human judgment exactly where it matters, and a rule an agentic PDR can follow later
Link to decision 37
38Anchor path: invite, sign, then planChanged later by 84Partners and scoring
Later changed by decision 84. Read it for the rule as it stands.
Invitation from the Anchor list; standard agreement plus an Anchor addendum; within the first 30 days a joint business plan, executive sponsors on both sides, a dedicated Slack Connect channel and a custom onboarding plan
Options considered
Plan before signing; same path as everyone
Why
White glove without slowing the signature
Link to decision 38
39Agreement: apply, get accepted, then click through terms on first portal loginPartners and scoring
(introw). Anchor and Elite partners add an addendum. Procurement path: partners who can't click through get a $0 invoice with the program agreement routed through their procurement. The same path provisions partner sandbox accounts (not-for-resale), beta tools and pre-launch access so they can get hands-on and pitch their own customers, which also removes the personal-credit-card blocker common with EU and non-US partners
Options considered
A negotiated contract for everyone; tiered agreements
Why
One lightweight path for most partners and a procurement-friendly path for the enterprises that need one
Link to decision 39
40Onboarded means all ofChanged later by 84Lifecycle
Later changed by decision 84. Read it for the rule as it stands.
three named contacts (executive sponsor, technical lead, sales lead), Tipalti KYC and payout details, Slack connected (Slack Connect for beta and Anchor partners, the community workspace for the rest), a partner sandbox account live, Sales and Technical Accreditation badges issued, and a kickoff held. For managed partners (assigned a partner account manager and a partner solutions architect) the kickoff comes first: it orients the team, sets expectations, and walks through a standard 30-day onboarding timeline and an accelerated one, so onboarding feels turnkey. The long tail gets an automated orientation instead
Options considered
A lighter checklist
Why
Firms that know implementations respond to a clear plan; the checklist makes "ready to sell and deliver" measurable
Link to decision 40
41Onboarding time ceilings by segmentLifecycle
Builder 14 days, Specialist 30, Scale and Anchor 45 (more people to accredit, procurement paths). Managed partners work to the 30-day plan or the accelerated plan within their ceiling
Options considered
30 days for everyone; 14 days for everyone
Why
Realistic per firm size while leaving room inside Forrester's 90-day activation window
Link to decision 41
42Onboarding deliveryLifecycle
an automated checklist and LMS in introw with nudges; a weekly live cohort session; a partner manager 1:1 only for managed partners (Anchor and Scale, plus any Specialist assigned a partner account manager)
Options considered
1:1 for everyone; fully self-serve
Why
Scales to 1,500 partners while keeping a human where the relationship is worth it; each automated step is a candidate for agentic help later
Link to decision 42
43Activation is three rewarded milestonesLifecycle
(1) First deal registration: a sign of intent that shows how the partner works with its partner account manager and how co-selling with AEs will go; engagement here opens funding and implementation support. (2) First closed-won deal. (3) First live joint customer: a customer in production on Cloud Cruz with the partner as sourcing partner, partner of record or implementer. "Activated" means milestone 3. Time between milestones and between deals measures velocity and each partner's propensity to keep working with Cloud Cruz, summarized as a cold, warm or hot activation rating
Options considered
Activation as live customer only; first closed-won only; first deal registration only
Why
Rewarding each step keeps partners moving and gives earlier signals than a single finish line
Link to decision 43
44Stalls must be caught during onboarding, not at day 45Lifecycle
Early flags on who is coming in hot and who isn't, from the first days after signing (design to be decided next)
Options considered
A day 45 / 60 / 90 plan alone
Why
By day 45 most of Forrester's 90-day window is gone
Link to decision 44
45Implementation package: a competitive incentive for enterprise dealsMoney
Implementation package: a competitive incentive for enterprise deals. Cloud Cruz offsets the partner's implementation time (consulting and hands-on-keyboard hours, e.g. 5 hours at $150 plus 5 at $200) so partners win business on Cloud Cruz over other platforms. Eligible: every registered deal on enterprise pricing (Cloud Cruz has self-serve, premium and enterprise pricing) above a minimum deal size, with the partner account manager, AE, solutions architect and partner solutions architect all engaged. Competitive migrations off named competitor platforms qualify. Package size scales with deal size. Drawn down from an annual budget; $100,000 for the 90-day beta, to test whether offsetting implementation cost shortens time to live. Refines decision 26
Options considered
Capped per partner per year; unlimited; first customer only
Why
Real, needle-moving deals get real investment; a budget, not a per-partner cap, controls cost; the beta measures whether it works. Still open: the enterprise deal-size minimum and the package bands
Link to decision 45
46Early warning: daily activation health (cold, warm, hot), judged against each partner's own onboarding planLifecycle
Signals: first portal login, contacts added, kickoff booked; checklist pace against the partner's plan (standard 30-day, accelerated or segment ceiling); days to each accreditation; sandbox logins and first builds; cohort attendance and Slack activity (a fair leading indicator, though engagement never counts toward the active-partner rule); shared target accounts in Crossbeam; a deal registration during onboarding. Rules: no first login by day 3, the PDR or partner manager reaches out; 25% behind plan, an automated nudge; 40% behind, cold, and a person acts; no sandbox build by day 10, partner solutions architect office hours; a deal registration or ahead of plan, hot, and a fast track to co-sell and implementation-package review. Visible rules in the beta; weights calibrated on Year-1 data by the Year-1 review and the propensity agent
Options considered
Calendar checkpoints at days 7, 14 and 30; a predictive model from day one
Why
A mechanism, not a slogan, behind "we drive activation"; per-partner plans treat a 45-day Anchor and a 14-day Builder fairly; there is no history to train on in a 0-to-1 program, so rules come first
Link to decision 46
47Credentials count toward tiers only when activeTiers
Level 1 and above expire after 12 months, with recertification on major releases, and a person counts only with an active credential and hands-on product use in the last 90 days
Options considered
24 months with no use rule; never expires
Why
Stops paper certifications from inflating tiers (the active-practitioner rule in public AI programs)
Link to decision 47
48Enablement is free for every tierChanged later by 81Credentials
Later changed by decision 81. Read it for the rule as it stands.
Anchor partners also get custom sessions and exam vouchers for large cohorts
Options considered
Free Level 1 with paid higher levels; tier-based voucher allowances
Why
Keeps the climb fast; the research recommends free tiers until benefits clearly justify fees
Link to decision 48
49Co-sell eligibility: Select and aboveLifecycle
Earned (4 certified people, 3 live customers, a reference), so AEs trust who they bring in; Member partners can still register their own deals
Options considered
Any activated partner; Premier and above
Why
The large cloud programs' co-sell eligibility pattern: referrals follow proven capability
Link to decision 49
50Rules of engagement and AE creditingLifecycle
Ownership: the AE owns the customer relationship; the partner account manager owns the partner relationship; the partner may own the customer too, which is where friction lives. Crediting policy starts with company philosophy set by the CRO, CEO, SVP of Sales or COO; the rules of engagement are then agreed collaboratively around the selling behavior and outcomes the business wants. Recommendation: full quota credit to AEs on partner-sourced and co-sold deals in their territory, with partner commission treated as a program cost, never a deduction from the AE
Options considered
Partial credit; SPIFF only
Why
Bringing in a partner never costs the AE, so AEs pull partners in, close faster and larger, and reach relationships they didn't have. Answers the Scorecard's open gap on AE crediting and channel conflict
Link to decision 50
51Routing: fit plus capacity, round-robin among equalsLifecycle
Each partner's profile is built with an agency Business Model Canvas (Osterwalder's nine blocks, public framework): the PDR and partner account manager fill it in call by call in Salesforce, enriching a public partner directory where customers can self-match. Partner-qualified leads are gated (for example a logged-in customer, or one who has had a conversation). Every customer inbound asks "Are you working with a partner? Can we recommend one?" After selection, how the deal came in and where it stalled feed coaching
Options considered
The AE's choice; tier priority
Why
Fair distribution that still matches capability; the directory turns profile data into self-serve demand; coaching closes the loop
Link to decision 51
52Outcome-based credential ladderChanged later by 64Credentials
Later changed by decision 64. Read it for the rule as it stands.
Outcome-based credential ladder. L1 Foundations: a technical exam plus a sandbox lab (connect a source, ship a first query), and the business L1 is required of technical people too, because they co-sell in pods: product, value proposition, co-positioning, the problem their own firm solves and why Cloud Cruz, capped by a joint plan to win a real target customer written on the Problem Statement Canvas (Marius Ursache, Disciplined Entrepreneurship Toolbox, Apr 26, 2018, https://www.detoolbox.com/articles/problem-statement-canvas/: context, customer type, root problem, emotional impact, quantifiable impact, alternatives and their disadvantages). L2 Delivery: technical, 2 customer go-lives on the packaged 30-day Launch plan with customer sign-off and a usage event; business, 2 signed business cases with a baseline and target ROI confirmed at 90 days. L3 Advisor: technical, runs the Cloud Cruz Readiness Review (data quality, security, cost, reliability, adoption) with 3 reviews and high-risk items closed; business, 3 accounts expanded on proven ROI plus executive value reviews. Customer outcomes are targets measured against each customer's kickoff baseline, never claims
Options considered
A single simpler track; exams only
Why
Each level certifies something that saves the customer time or money, which turns partner capability into ROI: "make the complex simple and get to revenue faster". Public patterns: delivery-based certification levels, packaged and gated implementation plans, and a fixed architecture-review rubric. Research: blueprint/research/certification-and-implementation.md
Link to decision 52
53The implementation package requires at least one L2 technical person on the dealMoney
The credential grants the right to deliver the packaged 30-day Launch plan
Options considered
No link to the ladder; L1 only
Why
Cloud Cruz's money funds proven delivery; one rule controls both cost and quality
Link to decision 53
54Partner health: four plain questions, each green, yellow or redLifecycle
Are they selling? (sold, influenced or supported in the last 90 days) Are their customers winning? (live, expanding, not churning) Are their people skilled? (active credentials) Is their delivery clean? (support tickets on their customers, CSAT, Readiness Review results) Healthy means no red. Shown on the partner scorecard; feeds tier reviews and the propensity agent
Options considered
A weighted health score; active or inactive only; partner NPS
Why
Simple enough to explain in one breath; the no-red rule keeps the same no-zero logic as the fit score; "are their customers winning" keeps customer time and money at the center
Link to decision 54
55Touch cadence by segmentLifecycle
Anchor: a monthly business review plus a quarterly executive review against the joint business plan. Scale and Specialist: a quarterly business review. Builder: an automated monthly scorecard digest plus the community workspace. Every review works from the same live scorecard
Options considered
Quarterly for everyone; on request only
Why
Human time goes where the relationship is largest; everyone sees the same numbers
Link to decision 55
56Inactivity path: 90, 180, 365 days, with tactful languageLifecycle
90 days without activity: marked "at risk" internally only, with automated re-engagement and a partner manager note. 180 days: a re-activation offer (fresh Crossbeam targets, a cohort seat). 365 days: moved to an alumni list with narrower portal access; earned commissions are always paid. Every message the partner sees is encouraging, hopeful and honest, aimed at getting the relationship started so both sides make money
Options considered
Never offboard; tier demotion only
Why
Keeps the roster honest so the active-partner rate means something, without burning relationships
Link to decision 56
57North star: customers live through partnersLaunch and targets
The count of Cloud Cruz customers in production with a partner as sourcing partner, partner of record or implementer, with time to live beside it; revenue, ROI and activation are supporting metrics
Options considered
Partner-sourced revenue; program ROI
Why
One plain number that is literally customers getting value; it leads revenue in Year 1 of a new program
Link to decision 57
58Rules hold for 12 monthsLaunch and targets
(tiers, commissions, packages), with one mid-year adjustment window decided at the Jan 1 review and announced 90 days ahead
Options considered
Change anytime; annual changes only
Why
Partners plan around stable rules; frequent restructures are affordable only at scale
Link to decision 58
59Partner advisory boardLifecycle
8 to 12 partners across the four segments, meeting before each semiannual review, seeded by the beta's design partners, plus a standing feedback channel in the Slack community
Options considered
Surveys only; feedback relayed by partner managers
Why
Partners shape the program they depend on (public precedent: partner advisory boards at finance-software companies)
Link to decision 59
60Pricing plansChanged later by 93Market and customers
Later changed by decision 93. Read it for the rule as it stands.
Self-serve: monthly, card on file, under $12K a year. Premium: annual, $12K to $50K ACV. Enterprise: annual or multi-year, $50K+ ACV, sales-led, implementation-package eligible
Options considered
Enterprise from $100K; enterprise from $25K
Why
Fits the mid-market ICP and the Foundation's deal sizes; sets the implementation-package minimum (decision 45)
Link to decision 60
61Implementation package bandsChanged later by 96Money
Later changed by decision 96. Read it for the rule as it stands.
$50K to $150K ACV: $1,750 (5 consulting hours + 5 hands-on). $150K to $500K: $3,500 (10 + 10). $500K+: $7,000 (20 + 20). Consulting at a maximum agreed rate of $150 an hour and technical implementation at $200, locked for the pilot. The $100,000 beta budget funds about 57 base packages, and funding increases as the pilot winds down to drive more conversions and data points
Options considered
A flat $1,750; a percentage of ACV
Why
Investment scales with the size of the win while staying explainable as hours of real work; locked rates make the pilot budget predictable
Link to decision 61
62Salesforce sampleSystems
all 25 beta design partners plus 25 launch partners across the four segments, with their accounts, deals, deal registrations, accruals and tickets (about 1,500 records, within Developer Edition limits); the full 1,500 partners stay in Postgres
Options considered
Beta cohort only; load until storage runs out
Why
Tells the 0-to-1 story end to end in the system of record, curated and explainable
Link to decision 62
63The agentic partner development rep lives inside build 3Team and AI
The Propensity and tier agent (Nov 13) extends to score partner leads (fit x intent) and draft PDR outreach, with a person approving every send
Options considered
A new build after December; future roadmap only
Why
Build 3 already scores partners; Year-1 data supplies the "run it by hand first" evidence (decision 34) without moving any date
Link to decision 63
64Readiness Review: four pillars, always including AICredentials
(Tony's rule; pillars chosen within it): Trust (data quality, security, reliability), AI quality (agent answers accurate and evaluated), Cost (no bill shock), Adoption (people use it). Replaces decision 52's five-pillar list
Options considered
Five pillars; six with AI; three
Why
Simpler to explain, and AI quality is always part of readiness for an AI data platform
Link to decision 64
65Year-1 targets (by Mar 31, 2028): outcomes, not countsLaunch and targets
North star: 400 customers live through partners. 250 partners activated (a live joint customer). 40% of partners activated within 90 days of signing (monthly cohort rate). About 1,000 partners signed is a funnel input, not a headline. Reviewed Jan 1 and Jul 1
Options considered
Conservative (about 500 signed); stretch (about 1,500 signed); signed partners as the headline
Why
Decided after a simulated leadership room (brief: blueprint/decision-briefs/2026-09-25-year-one-targets.md). Signed partners is a vanity metric; outcomes are what the business and the field care about
Link to decision 65
66The partner team scales on flat headcount: 9 people all yearChanged later by 110Team and AI
Later changed by decision 110. Read it for the rule as it stands.
(head of partner ops, PDR, 2 partner managers, partner account manager, partner solutions architect, enablement, partner operations, partner marketing; AEs and solutions architects sit in sales). Scale targets: 44+ customers live through partners per team member; 60% of partner questions answered by the Support Hub without a ticket; 70% of long-tail onboarding completed with no human touch; 60 to 80 managed partners. L2 and L3 proof-of-work reviews use the Readiness Review rubric plus sampling and a weekly cap. The Year-1 data holds headcount at 9 (the Foundation's planned extra hire is removed)
Options considered
A growing team (the Foundation's 14 to 15)
Why
Scale comes from systems, not hires: builds 3 and 4 (agentic PDR, Support Hub, triage) become the operating model, with a person approving every state change; cost per customer live falls all year
Link to decision 66
67Region to start: the United States, the Americas and English-speaking companiesMarket and customers
Applies to customers and to partner recruitment in Year 1
Options considered
Global from launch
Why
Focus: one language for enablement, support and the Support Hub, and time zones a flat team can cover; expansion is a later decision
Link to decision 67
68Intent score (100 points)Partners and scoring
Completed application 30; responded to outreach or booked a meeting 20; stated need and timeline 20 (now 20, this quarter 10, later 0; a typed judgment read from the application or reply); webinar or cohort attendance 10 each up to 20; directory visits or content downloads up to 10. Deal in hand is automatic urgent (a typed judgment on free text; a person confirms low-confidence calls). Signals older than 30 days count half. Refines decision 35
Options considered
The first draft (application, outreach, events, visits, downloads)
Why
Actions weigh more than browsing; stated urgency is the best prioritizer; the half-life keeps intent current. Chosen after a leadership room (brief: blueprint/decision-briefs/2026-09-25-intent-score-and-jev.md)
Link to decision 68
69A vendor-neutral typed-decision layer, with Jev (TypeSafe's System One model) as its first enginePartners and scoring
Code owns the workflow and the point weights; typed judgments (choice, score, yes/no, each with a probability) fill only the signals plain code can't read. The same layer later serves support routing and deal-registration triage. Designed now, run by hand in the beta, live in build 3 with the agentic PDR. Conditions: a rules-only fallback; every judgment logged in Salesforce with value, confidence, model and date; calibration checked against outcomes; a vendor and privacy review before any real data; pricing verified
Options considered
A points table only; a general-purpose LLM prompt-and-parse
Why
Explainable (our weights, model judgments per signal), low-confidence calls go to a person, and it adapts to the market without depending on one brand-new vendor
Link to decision 69
70Automation scales by designTeam and AI
Capacity-aware routing (each rep gets the top N leads by fit x intent daily; N starts at 8); event-driven, idempotent scoring written to Salesforce, batch work on the scheduler; people handle hot and urgent leads, the agent drafts warm outreach and runs nurture; quality guards (human-review rate at or under 10%, a random 5% weekly audit, weekly calibration); a monthly spend cap on the judgment API
Options considered
A fixed "hot" threshold with humans on every lead
Why
Tested at Year 1 (about 70 leads a week) and ten times that: compute is not the constraint, people are; routing by capacity keeps human time flat while governance stays cheap
Link to decision 70
71Outreach approval: a person approves every AI send in Year 1 and in the demoTeam and AI
as the founding guardrail requires. Policy-level approval (approved templates, audiences and limits, plus a sampled audit) is allowed only when volume makes per-send approval the bottleneck, and only by a new explicit decision from Tony
Options considered
Every send forever; policy-level from day one
Why
Keeps the founding guardrail intact now while naming exactly how, and by whom, it could change at scale
Link to decision 71
72Commission rates confirmedMoney
Member 10%, Select 15%, Premier 20%, Elite 25% of collected revenue on partner-sourced deals; marketplace partner of record 5%
Options considered
Flatter steps (10/12/15/18%)
Why
Tony's call after a mentor-mode leadership room (brief: blueprint/decision-briefs/2026-09-26-commission-rates.md); in a real market range (public partner programs pay 10 to 25% by tier) and each tier step is worth earning
Link to decision 72
73Commission term and rate lockMoney
The tier rate pays for 24 months from customer go-live, then a lower renewal rate (number set later). The rate at close holds for that deal's whole term: promotions apply to new deals, and a demotion never cuts pay on customers a partner already landed
Options considered
Pay forever; rate follows the current tier
Why
Bounded cost, rewarded retention, and no pay disputes at tier reviews
Link to decision 73
74Influenced deals count but don't pay commissionChanged later by 94Money
Later changed by decision 94. Read it for the rule as it stands.
They count toward tier and attributed revenue and can earn a SPIFF
Options considered
Pay influence at the tier rate
Why
Avoids paying twice alongside the AE's full quota credit (decision 50) while still rewarding co-selling
Link to decision 74
75Self-serve referrals and clawbackMoney
Self-serve customers referred through a tracked partner link are paid automatically; a commission reverses if the customer refunds within 90 days
Options considered
Manual handling of small referrals; no clawback
Why
Keeps a flat team out of small-commission admin and protects against refunded deals
Link to decision 75
76Accelerators and SPIFFs are mechanisms now, numbers laterMoney
(set at the Jul 1 review with Year-1 data). SPIFFs fire on the north star: first customer live, first Enterprise go-live, and customer retained at 12 months
Options considered
Numbers now; SPIFFs on activity
Why
Every incentive points at customers live through partners; numbers set on evidence
Link to decision 76
77A co-marketing fund, separate from commissions, for Premier and aboveMoney
Partners apply with a short plan for a campaign of $3,500, $5,000 or $10,000; Cloud Cruz approves each plan and pays against proof of performance (events, content, paid media). Elite partners also get specialized campaigns built with Cloud Cruz partner marketing. Replaces the room's idea of reinvesting commission as credit
Options considered
Commission reinvested as co-marketing credit at 1.2 times; an automatic allowance by tier
Why
Tony's call: marketing money follows a plan and results, commissions stay simple cash, and three fixed sizes are easy to explain and budget
Link to decision 77
78Contribution pointsMoney
Partners earn points for approved contributions: community participation, LinkedIn posts, and published articles that follow Cloud Cruz's language guide. Partner marketing approves each one before it counts. Select and above need a minimum each review period, rising by tier (numbers set later with Year-1 data). A higher score also raises the partner's rank in the partner directory, so contribution and influence bring referrals
Options considered
Points as a booster only; no contribution requirement
Why
Tony's call: rewards partners who grow the community and market Cloud Cruz, ties it to tiering, and turns the directory into the payoff
Link to decision 78
79Stage owners, clocks and one metric eachChanged later by 101Lifecycle
Later changed by decision 101. Read it for the rule as it stands.
Owners: Develop and Acquire the PDR (the partner manager supports Acquire: the meeting, admissions at fit 50 to 69, the Anchor plan), Onboard the partner manager or automation, Activate the partner account manager, Enable the enablement lead, Hand off / Co-sell the partner account manager with the AE, Maintain the partner account manager or automation, Optimize the head of partner operations. New clocks: an Acquire decision within 5 business days of the meeting; deal registrations answered within 2 business days; activation within 90 days of signing. Metrics: qualified meetings held, fit score of accepted partners, days to onboarded, share activated within 90 days, active certified people, co-sold win rate, partners with no red, customers live through partners
Options considered
Partner manager owns Acquire; shared ownership of Acquire
Why
Tony's call: the PDR who opens a lead owns it through signing, so partners never start over, and one owner per stage holds (decision 34)
Link to decision 79
80Sales and Technical Accreditations never expireCredentials
Levels 1 to 3 still last 12 months (decision 47)
Options considered
Expire at 12 months like the Levels
Why
They mark onboarding; the 90-day product-use rule already keeps inactive people from counting toward a tier, so an expiry adds only admin
Link to decision 80
81Level exams carry a fee, offset by yearly vouchersCredentials
Level 1 $100, Level 2 $150, Level 3 $150. Free exams a program year (reset Apr 1, any level, recertification included): Member 5, Select 10, Premier 10, Elite 15. Extra vouchers for people on the accelerated onboarding plan, issued automatically; Anchor extras stay. Courses, labs and accreditations stay free. Amends decision 48
Options considered
Free exams at every tier; a per-level or one-time voucher pool
Why
Tony's call: a priced exam gets preparation and focus, especially when the company pays, so the credential holds its value; vouchers keep it free to committed partners
Link to decision 81
82Contribution points live in introw; the partner directory lives on the Cloud Cruz siteSystems
Partners submit contributions in introw, partner marketing approves them there, and the total syncs to Salesforce for tiering. The public directory is built from each partner's Salesforce profile (the agency Business Model Canvas) and ranked by points and results
Options considered
Everything in Salesforce with Experience Cloud; everything in introw
Why
Partners already work in introw; the directory reads one source, and no extra license
Link to decision 82
83Exams: introw for learning, Pearson VUE for Level examsCredentials
introw hosts the courses, study material and the two accreditations. Level 1, 2 and 3 exams are proctored through Pearson VUE, which also issues the vouchers; a pass triggers the Credly badge, which writes the credential to Salesforce
Options considered
Exams in introw's LMS with voucher codes
Why
Tony's call: a proctored exam matches the paid-exam rule (decision 81), so a credential means something
Link to decision 83
84Slack Connect: design partners during the beta, then Premier and Elite onlySystems
Everyone else, including Anchors below Premier, uses the community workspace. Amends decisions 38 and 40
Options considered
Beta, Anchor, Premier and Elite
Why
One rule tied to tier; fewer channels for a flat team to staff
Link to decision 84
85Salesforce: an Opportunity Partner junction objectSystems
One row per partner per deal, with a role: Sourced, Influenced, Implementer or Partner of record. The Sourcing Partner lookup stays on the Opportunity (the partner that is paid). Only a Sourced or Partner of record row can carry a commission accrual; an Influenced row can earn a SPIFF (decision 74). Customer records stay standard: Accounts (Customer record type) and Contacts, with every opportunity on the customer account and owned by the AE
Options considered
Single lookups for sourcing partner and partner of record; Salesforce's standard Opportunity Partner list (no custom fields)
Why
Real deals often have more than one partner (one sources, another implements); one row per role answers "who did what" and keeps commission rules enforceable
Link to decision 85
86The Year-1 schema replaces 001_schema.sql in placeProject setup
The database is reset, never migrated, so one readable schema file beats a migration that drops and alters a dozen tables. The Foundation schema stays in the v0-foundation tag
Options considered
A new 003_year_one.sql on top of 001
Why
A reader sees the whole model in one file; the history lives in git
Link to decision 86
87The data covers Jan 1, 2027 to Mar 31, 2028Project setup
the 90-day beta plus Year 1
Options considered
Year 1 only (Apr 1, 2027 on)
Why
The beta cohort, the $100,000 package budget and the launch ramp all show in the data
Link to decision 87
88SPIFF amounts, set at the Jul 1, 2027 reviewMoney
first customer live $500; first Enterprise go-live $2,500; customer retained 12 months $1,000. Accelerator and renewal-rate numbers stay open (no deal reaches 24 months in Year 1)
Options considered
$250 / $1,000 / $500; no SPIFFs in Year 1
Why
Small next to a 20% commission, and every SPIFF pays for the north star (decision 76)
Link to decision 88
89Contribution points, effective Jul 1, 2027Money
community answer 5, LinkedIn post 10, published article 25. Minimum per half-year review: Select 20, Premier 50, Elite 100; Member none
Options considered
Same points with half the minimums
Why
A few approved posts a quarter keeps Select within reach while the top tiers lead the community (decision 78)
Link to decision 89
90The 40% activation target is an exit rateLaunch and targets
The monthly cohort rate (activated within 90 days of signing) climbs from about 20% for the beta and launch cohorts to 40% for the latest eligible cohorts (Oct to Dec 2027), with about 250 activated in total
Options considered
A flat 40% (about 300 activated); a flat 30% with 40% shown as missed
Why
About 665 partners are 90+ days old by Mar 31, so a flat 40% contradicts the 250 target; a rising rate is the believable improvement story (decision 65)
Link to decision 90
91During the 90-day beta, Cloud Cruz's partner solutions architect stands in as the Level 2 technical personLaunch and targets
on implementation-package deals and pairs with the partner, whose people earn Level 2 from those go-lives. From launch, decision 53 applies as written
Options considered
Waive Level 2 in the beta; keep the rule and spend nothing
Why
Level 2 takes two go-lives, which no partner can have in a 90-day beta, so the rule as written left the $100,000 pilot budget unspendable; this keeps the quality bar and funds the pilot
Link to decision 91
92Promote when qualifiedChanged later by 109Tiers
Later changed by decision 109. Read it for the rule as it stands.
A partner is promoted within 30 days of meeting a tier's criteria (checked on the 1st of each month) instead of waiting for Jan 1 or Jul 1. Demotions stay at the Jun 30 review with 90 days' notice. Amends decision 33
Options considered
Open co-sell below Select; keep twice-yearly reviews
Why
Year-1 data showed only 5 partners at Select by Jan 1 and 14 by Mar 31, with co-sell gated at Select (decision 49): partners who earn it now get it quickly, and the bar stays the same
Link to decision 92
93Annual and multi-year contracts are billed quarterly in advanceMoney
Annual and multi-year contracts are billed quarterly in advance; self-serve bills monthly by card. Commissions on annual plans arrive as each quarter is paid. Refines decisions 25 and 60
Options considered
Annually in advance
Why
Common for mid-market software, and steady commission timing that matches "paid as Cloud Cruz is paid"
Link to decision 93
94Influence is rewarded with work, not cashMoney
(A) Services routing back: on an influenced Enterprise win, the AE offers the implementation to the partner that helped close it, and the customer can still choose. (B) The implementation package extends to influenced Enterprise deals, paying for logged hours. Influence still counts toward tier and attributed revenue and pays no commission. Replaces "can earn a SPIFF" in decision 74
Options considered
A flat influence SPIFF gated on AE confirmation; platform credits; a tier-credit multiplier; gated cash for top tiers
Why
Tony's call after research (blueprint/research/influence-rewards.md): most large programs pay cash only on sourced deals and reward influence with tier credit or funded work; this never pays twice and gives partners the services revenue they want
Link to decision 94
95Launch holds Apr 1, 2027, with a go/no-go check on Mar 15Launch and targets
Four must-haves: at least 2 design partners each have a customer live; commissions are earning in the ledger and a test payout has run through Tipalti; median onboarding under 30 days; 90% of deal registrations answered within 2 business days. If one is missed, launch goes ahead narrower (Builder and Specialist only), never later. The beta passed all four
Options considered
The must-haves set the date; Apr 1 with no check
Why
Partners and the team plan around a known date, and the check still proves the motion works before partners pour in
Link to decision 95
96The unspent beta package budget carries into Year 1, topped up $50,000 at the Jan 1, 2028 reviewMoney
The beta approved 9 packages ($22,750) of its $100,000, because most beta wins were Premium, below the Enterprise line. The remaining $77,250 funds packages from Apr 1, and the top-up makes the Year-1 budget $127,250. Refines decision 61
Options considered
Widen eligibility to large Premium deals; close the beta budget and set a new Year-1 line
Why
No rule changes; the money follows Enterprise deals as they arrive, and the Jan 1 review sees the real pace before adding more
Link to decision 96
97The Technology path, resellers and cloud marketplace co-sell all open Oct 1, 2027, confirmed at the Jul 1 reviewLaunch and targets
The review checks readiness for each: ISV embed terms and integration docs; marketplace listings live on AWS, GCP and Azure; partner solutions architect capacity. One that isn't ready waits; the others open. An ISV with a deal in hand can join early by exception. Refines decision 15
Options considered
Technology Oct 1, resellers and marketplace Jan 1; each opens when ready, no date
Why
The same pattern as launch: a date partners can plan around, with a check that keeps it honest
Link to decision 97
98Design partners keep recognition, not rulesLaunch and targets
A permanent "Design partner" label in the partner directory and first seats on the advisory board. They start at Member and earn tiers like everyone else. Their Slack Connect channels close Mar 31, 2027, with 30 days' notice, unless they reach Premier (decision 84)
Options considered
Select for Year 1; nothing special
Why
Credit for shaping the program without giving anyone a tier they didn't earn (decisions 33, 92)
Link to decision 98
99Partner solutions architect capacity: 12 open package dealsTeam and AI
The partner solutions architect leads up to 12 open implementation-package deals at once, largest first. Above 12, the partner's own Level 2 technical person leads delivery and the partner solutions architect reviews it at go-live. ISV and technology onboarding stays automated, with 4 integration reviews a month for the ISVs with the largest pipeline
Options considered
12 with new packages queued; a cap of 8
Why
Open package deals reach 24 by Mar 2028; Level 2 already proves delivery (decision 53), so work moves to partners who earned it on flat headcount (decision 66)
Link to decision 99
100Guard metrics, and resellers paid under existing rulesLaunch and targets
Two guards are checked monthly: median days to onboarded at 30 or fewer, and 90% of deal registrations answered within 2 business days. If one is red for a full calendar month, the next opening pauses until it's green for a full month. Resellers earn the tier commission when they source and 5% as marketplace partner of record (decision 72); a dedicated reseller discount is a Year-2 decision. Considered and dropped: three separate opening dates with resellers in Year 2, which would regenerate every record for no gain a reader would see
Options considered
Three staggered openings; a reseller discount now
Why
Tony's call: keep the demo simple and defensible. The guards answer "why open three motions at once on a 9-person team," and the existing commission rules already cover resellers
Link to decision 100
101Two books: managed and long tailChanged later by 110Team and AI
Later changed by decision 110. Read it for the rule as it stands.
The partner account manager owns the 80 managed partners (Anchor, Scale and assigned Specialists) from Activate through Maintain; 80 is the cap (decision 66). The head of partner operations is accountable for the long tail of about 920 partners. The two partner managers work it by exception, acting on partners whose health turns cold or hot (decision 46), and automation does the rest. Partner operations studies where long-tail partners stall and fixes the cause across the board with partner marketing and lifecycle marketing (onboarding content, nudges, campaigns), not one partner at a time. Amends decision 79
Options considered
Automation owns the long tail with no named person; the partner account manager owns all 997
Why
Tony's call: one named person is accountable for every partner, human time goes to the exceptions, and a stall seen twice becomes a fix for everyone
Link to decision 101
102The program's calendar moves 18 months laterProject setup
The program's calendar moves 18 months later, so every program date falls after the build began: beta Jan 1 to Mar 31, 2027; launch check Mar 15, 2027; launch Apr 1, 2027; reviews Jul 1, 2027 and Jan 1, 2028; Technology, resellers and marketplace open Oct 1, 2027; Year 1 ends Mar 31, 2028. The data is regenerated on the new calendar (story, signings and cohort rates unchanged). Figures that moved: customers live through partners 445 (was 440); partners activated 264 (was 265); beta packages 8 for $19,250 (was 9 for $22,750), so the carry-over is $80,750 and the Year-1 package budget $130,750, with $117,250 approved (decision 96); open package deals peak at 28 (was 24; decision 99); Select partners 16 at year end. Salesforce date formulas now compare against a Program as-of date setting instead of TODAY(). Tony began building the demo on Sep 23, 2026
Options considered
Keep the 2025 to 2026 calendar with a note; a relative calendar with no years; a whole-week shift that kept every number but landed on odd dates
Why
Tony's call: a program designed in 2026 should run on dates that come after it; clean dates beat identical numbers, and every rule stays exact
Link to decision 102
103Builds show status, not future datesProject setup
Each build is Shipped, Building now or Planned; a date appears only once a build ships. The case study opens with four reading paths by role: partner programs; partner operations and RevOps; implementation and enablement; AI in partner operations Replaces the dated schedule in decision 10.
Options considered
Keep public ship dates; ship every remaining build this week
Why
One calendar on the page, the program's own; an unfinished date reads as unfinished work; each build ships when its guardrails hold, and a reader starts from the finished work that matches their role
Link to decision 103
104"Ask the data" ships inside Build 2, the Measurement LayerTeam and AI
A visitor asks a question in plain words; Claude writes a read-only SQL query against the metric layer, runs it, and shows the query beside the answer. Guardrails from the founding rules: the read-only database role, a 5-second statement timeout, a 500-row cap, a per-visitor rate limit and a daily spend cap, set as a Vercel AI Gateway budget
Options considered
Dashboard only, with the question box moved to Build 4; a fixed menu of example questions first
Why
Tony's call: anyone on the partner team can get an answer without waiting on an analyst, the guardrails already exist in the Foundation, and showing the query keeps every answer checkable
Link to decision 104
105Two clean revenue measures, never double countedMeasurement
Partner-sourced revenue goes to the one sourcing partner, with ties broken by the rules in foundation/config/attribution.yaml. Partner-influenced revenue lists every partner who helped, split by touch weight. At program level each deal counts once in each measure, so sourced plus influenced never exceeds real revenue
Options considered
Full credit to every partner on a deal; partner of record only
Why
Tony's call: totals a finance team can reconcile to invoices, while every partner who helped still shows up (decisions 74, 85, 94)
Link to decision 105
106Program cost is shown two waysMeasurement
Paid to partners: commissions (SPIFFs included), implementation packages and the co-marketing fund. Fully loaded: adds the 9-person partner team, tools and exam vouchers. Cost per customer live leads with the fully loaded figure. AE pay stays out of program cost (decision 50)
Options considered
Fully loaded only; paid to partners only
Why
Tony's call: the fully loaded number is the one a CFO would ask for, and the paid-to-partners view shows where the partner money goes. Research: blueprint/research/measurement.md
Link to decision 106
107The measurement page leads with five numbers, north star firstMeasurement
customers live through partners, partner-sourced revenue, partner-influenced revenue (shown separately and never added to sourced), fully loaded cost per customer live, and partners activated. Marketplace deals are their own bucket, credited to the partner of record, so every won dollar sits in exactly one bucket
Options considered
Three numbers; revenue first
Why
Tony's call: outcomes first and revenue second, the way the large cloud programs grade partners, while keeping the co-sell work of decision 94 visible
Link to decision 107
108Tony leads the partner ecosystem, not just its operationsProject setup
The role owns partner strategy, go-to-market and the program, with partner operations as one part of it. The 30-day Launch plan, backed by the implementation package, is the solution play partners sell. Amends decision 5
Options considered
Partner operations lead
Why
Tony's call: the job already covers strategy, partner economics and go-to-market, so the title says so
Link to decision 108
109Demotions happen at the Jul 1 review, with 90 days' noticeTiers
the same date as the program's mid-year review, so partners hear about tiers once, on the program calendar. Amends decisions 33 and 92
Options considered
Keep a separate Jun 30 review
Why
Tony's call: one review calendar is simpler for partners and for a team of 9
Link to decision 109
110The top seat on the partner team is head of partner ecosystemTeam and AI
owning Optimize, the long tail, the program reviews, the launch checks and the guard metrics. Partner operations stays its own seat on the team. Amends decisions 66 and 101
Options considered
A separate head of partner operations reporting to the ecosystem lead
Why
Tony's call: it matches decision 108 and keeps the team at 9
Link to decision 110