--- name: channel_partner_program description: "Build and run a reseller and channel partner program: score partners into tiers with criteria anyone can verify, set discount and margin rules that survive the unit economics, run deal registration that stops channel conflict, onboard partners to a first deal, and hold the quarterly review that develops, demotes, and prunes partners on evidence. Use when launching a reseller, distributor, or VAR program, fixing tiers or margins in an existing one, settling a partner versus partner or partner versus direct conflict, or cleaning out dormant partners." --- # Channel Partner Program ## Purpose Give independent companies a clear, written deal: what they earn, what they must prove to earn more, which opportunities are theirs once registered, and what happens when they stop producing. A program works when a partner can predict a vendor decision without knowing anyone at the vendor. Everything here serves that test: published tier criteria scored from records, one discount table checked against real costs, deal registration with timestamps, and reviews that move partners up, down, and out. This skill covers reseller, distributor, and value-added reseller programs for products and product-plus-services offers. It does not cover referral or affiliate programs (no resale, no margin), co-marketing campaigns with a partner, or scoring vendors you buy from. ## Workflow ### 1. Decide the partner job before recruiting Write down, in one short page: which customers partners reach that direct sales cannot or should not, what the partner does in the sale (source, close, implement, support, renew), and which segments partners are authorised for by size, vertical, or geography. Recruit only against that page. A partner signed with no defined job becomes a dormant logo within two quarters, and dormant partners are the main cost this skill exists to prevent. Publish rules of engagement alongside the program: account boundaries between direct and channel, who may pursue what, how registered deals are protected from the direct team, and how conflicts escalate. Put them in writing before the first partner signs, because the first conflict will arrive before anyone feels ready to write them. ### 2. Set tiers from verifiable criteria Score partners with the worksheet in references/tier-and-margin.md. Five inputs, all checkable from records the vendor already holds: trailing revenue, certified staff count, approved registered deals, a current business plan on file, and first-line support capability. Points set the tier: Platinum at 70, Gold at 45, Silver at 25, Registered below that. Run the scorer over the current or target partner base before locking bands: python3 scripts/channel.py score partners.csv Calibrate the revenue bands until the top tier is a small minority and the next tier up is reachable within one planning cycle from the middle of the base. Each tier must change something a partner can feel: discount step, deal registration priority, marketing development funds, lead access, or named manager time. A tier that changes nothing is a label, not an incentive. State the re-tier cadence (annual is common, with quarterly reviews feeding it) and that movement runs both ways. A tier that can only be gained fills up with partners who stopped producing years ago, and every benefit they keep dilutes the partners doing the work. ### 3. Set margin and discount rules from the worksheet Work the margin math per tier before publishing any discount, using references/tier-and-margin.md: list price, tier discount, vendor delivery cost, partner price, vendor margin over cost, and the partner gross left to cover their selling cost. python3 scripts/channel.py margin --list-price 10000 --discount-pct 30 --deal-reg-pct 5 --vendor-cost 4000 Rules that keep the table honest once published: - One table for everyone. Individual exceptions are written, dated, and expire. - Say which discounts stack and cap the total. Deal registration extra, promotions, and volume terms combining by accident is how margin disappears. - Set the deal registration extra as a defined number of points on registered deals, paid for the partner doing the early work of finding and developing the opportunity. - Decide renewal ownership now: renewals earn the discount of the partner holding the account registration, at their current tier. - Re-run the worksheet against current costs once a year. A table set against old costs can pay out of a margin that no longer exists. ### 4. Run deal registration as the conflict system Follow the workflow in references/deal-registration.md. The parts that must be in place before the first registration: - Submission fields that make a registration checkable, with the account website domain as the identity key for duplicates. - A decision within two business days, with declines given from a published reason list. - A protection window sized to your median sales cycle. Published programs commonly run 60 to 90 days for standard deals. Longer windows for larger deals are fine when they carry an evidence checkpoint. - Activity checkpoints (discovery by day 14, a proposal or written next step by day 30) so protection rewards partners who are working the deal, and an extension path based on evidence of progress. - A named conflict owner, a published evidence order, and decisions recorded in the log. First qualified submission wins; exceptions are documented and visible to the partner who lost. Audit the log for collisions whenever you review the program: python3 scripts/channel.py conflicts deals.csv --as-of 2026-10-21 Protection binds the vendor as well as other partners. If the direct team can take a registered account, partners stop registering, pipeline visibility dies, and the program becomes a discount scheme with a logo page. ### 5. Onboard to a first deal, not to a signature Work a fixed checklist for every new partner: 1. Agreement signed, program terms and rules of engagement acknowledged. 2. One named point of contact on each side, recorded where both can find it. 3. Finance set up: tax and payment details, invoicing route, payment cycle stated. 4. Training: product and sales training booked, with the certification target for their first tier step named (for example two certified staff to reach Silver under the worksheet). 5. Brand assets and co-branding rules handed over, with what needs approval stated. 6. First registration target: one real opportunity registered within 60 days, or a dated reason why not. 7. Check-ins booked at 30, 60, and 90 days, then quarterly reviews after that. The 90 day check-in has one question that matters: is this partner moving toward a first deal, and if not, what specifically is blocked? Answers decide whether the partner gets help, more time, or an honest early exit. ### 6. Hold the quarterly review and act on it Score every partner each quarter and sort them into actions: python3 scripts/channel.py review partners.csv --as-of 2026-10-21 - Promote and develop: partners within reach of the next tier get the exact points gap and a dated plan. - Tier review: high-tier partners gone inactive get checked before benefits renew. Tell a partner their tier is at risk the quarter before it moves, with the numbers, never at the moment it moves. - Prune candidates: Registered partners with no revenue, no registered deals, and no activity in 180 days get written notice and one quarter to reactivate by registering a real deal or certifying staff. Remove the ones who do neither. Removal terms protect the customer: support continues, live registrations run to expiry, accrued funds are handled as the agreement states. - Conflict and rules check: review the conflict log and any documented exceptions. A rule with growing exceptions is a rule being replaced in secret. Either enforce it or change it in the open. Pruning is a program health action, not a punishment. Dead partners cost manager time, confuse customers checking the partner directory, and hold tier benefits the active base earned. ### 7. Scale the machinery with partner count Use the worked examples in references/worked-program.md: the same program run at 12 partners and at 60. The short version: at low counts a manager plus a spreadsheet log works; past roughly 30 partners, registration dedupe and conflict checks need to run at submission and weekly, marketing funds work better accrued from revenue than as a fixed pot, and the review becomes a scored pipeline instead of a round of meetings. The rules themselves do not change with size. Only the tooling does. ## Output Contract A program built with this skill produces: - A one-page partner job and rules of engagement document (segments, direct versus channel boundaries, conflict path). - A tier table scored from the worksheet in references/tier-and-margin.md, with the calibration note showing how the current base scored. - A discount and margin table with the per-tier worksheet math behind it, stacking rules, and the review date. - Deal registration rules and workflow from references/deal-registration.md, with the window length, checkpoints, and conflict evidence order stated. - An onboarding checklist per partner, with 30, 60, and 90 day check-ins booked. - A quarterly review sheet from the review command: every partner sorted to hold, develop, tier review, or prune candidate, with the reason shown. ## Operating Rules - Never publish a tier, discount, or benefit you have not costed with the margin worksheet. - Never decide a deal conflict from memory or relationships. Timestamps and the log decide, and the decision gets recorded. - Never let a partner's tier rest on self-reported numbers. Every scoring input traces to a vendor record. - Never announce a demotion or removal that the partner has not already seen coming in a prior review, except for fraud or a contract breach. - Never recruit past the partner job page to hit a partner count. Partner count is a vanity number; active, producing partners are the program. - Present ranges from published guides as starting points to calibrate, never as guarantees of what a market will bear. The worksheet and the partner base in front of you outrank any general range. - Keep partner commercial data inside the program. A partner's revenue, pipeline, and tier score are never shared with another partner.