--- name: revops description: "Design the systems that connect marketing to revenue: lead lifecycle stages, scoring, routing, pipeline stage management, handoff SLAs, deal desk, and RevOps metrics. Use when the user says RevOps, lead scoring, lead routing, MQL, SQL, pipeline stages, deal desk, or leads aren't getting to sales. Not for CRM dedup and record cleanup: use sales-data-hygiene. Not for outbound sequences: use sales-cadence. Not for pricing decisions: use pricing." license: MIT metadata: author: TechTide AI (Alex Cinovoj) provenance: rewritten from patterns in an open SaaS marketing skills pack (MIT) category: Sales & RevOps --- # RevOps Define before you automate: automating a broken process just produces broken results faster. One CRM as the single source of truth, written definitions every team agrees on, and an SLA on every handoff. ## Before starting If `.agents/product-marketing.md` or `.claude/product-marketing.md` exists, read it first. Then confirm: GTM motion (PLG, sales-led, hybrid), ACV range, sales cycle length, current stack, how leads are managed today, and the goal (conversion, speed-to-lead, handoff leaks, greenfield build). Start where the pain is; do not block on missing inputs. ## Core rules 1. **Single source of truth.** Pick the CRM as canonical, sync everything into it. Data in two masters will conflict. 2. **Definitions on paper first.** Stages, scores, and routing rules get written and agreed before any workflow is built. 3. **Every handoff gets an SLA.** Marketing to sales, SDR to AE, AE to CS. Each needs a response time, a tracking field, and an accountable owner. 4. **Alignment is a deliverable.** If marketing calls it an MQL and sales will not work it, the definition is wrong. Fix the definition, not the reps. ## Build order 1. **Lifecycle stages.** Subscriber, Lead, MQL, SQL, Opportunity, Customer, Evangelist, each with entry criteria, exit criteria, and an owner. An MQL requires both fit (matches ICP) and engagement (buying-intent behavior). Neither alone qualifies: a perfect-fit account that never engages is not an MQL, and neither is a student downloading every ebook. 2. **Handoff SLA.** Typical shape: MQL alert fires to the assigned rep, contact within 4 business hours, qualify or reject within 48 hours, rejected MQLs return to nurture with a reason code. Track SLA misses and escalate them. 3. **Scoring model.** Fit attributes (size, industry, role, stack) plus engagement signals (pricing page, demo request, product usage) minus negative signals (competitor domains, personal email, mismatched titles). Backtest against closed-won history before launch, recalibrate quarterly. 4. **Routing rules.** Most specific match first, always a fallback owner, capacity-aware round-robin. Speed-to-lead dominates conversion: minutes matter, and a lead untouched for a day is effectively cold. Log every routing decision. 5. **Pipeline stages.** Required fields per stage, exit criteria, stale-deal alerts at 2x average stage time, stage-skip detection, and no silent close-date pushes: every push needs a reason. 6. **Metrics dashboard.** Three views: marketing (volume, MQL rate, cost per MQL), sales (stage conversion, velocity, coverage ratio of 3-4x quota), executive (CAC, LTV:CAC, revenue vs target). Read references/frameworks.md when you need the full stage table, scoring template with point values, routing method comparison, deal desk approval tiers, or the metrics formula table. ## Deliverables format Ship each artifact as a standalone document the user can implement directly: lifecycle stage doc, scoring spec, routing rules with decision tree, pipeline configuration, and dashboard spec. Add platform-specific notes when the CRM is known. ## Good vs Bad **Bad MQL definition:** "Anyone who downloads two pieces of content." Sales gets flooded with students and researchers, stops trusting the queue, and starts cherry-picking. The funnel metrics become fiction. **Good MQL definition:** "Fit score at or above threshold (ICP match on size, industry, role) AND an intent action in the last 14 days (pricing page, demo request, or trial signup). Competitor and personal-email domains excluded. Sales accepts or rejects within 48 hours with a reason code." Both dimensions, time-bound, negative filters, and a feedback loop. ## Verification After defining stages and scoring, backtest: pull the last 50 closed-won and 50 closed-lost deals and score them retroactively. Expect closed-won deals to have crossed the MQL threshold meaningfully more often than closed-lost. If the model cannot separate them, the weights are decoration; rework before automating. After routing goes live, sample 20 routed leads from the log. Expect every lead assigned within the SLA window and zero leads in the fallback bucket without follow-up. Any unowned lead: fix the rule gap that dropped it. ## Footguns - **Automating before aligning.** Workflows built on contested definitions get bypassed by reps within weeks. Fix: get written sign-off from sales and marketing on definitions first. - **Scoring content downloads as intent.** Research is not buying. Fix: weight pricing, demo, and product actions far above content touches. - **No negative scoring.** Competitors and students inflate MQL counts. Fix: subtract points for disqualifying attributes, hard-exclude where certain. - **Routing without a fallback owner.** Leads that match no rule rot unassigned. Fix: catch-all rule plus a daily unassigned-leads alert. - **Set-and-forget models.** Buyer behavior drifts. Fix: quarterly recalibration on fresh closed-won data. ## Completion checklist - [ ] Stage definitions written with entry/exit criteria and owners - [ ] MQL requires fit AND engagement, with negative filters - [ ] Every handoff has an SLA, tracking field, and owner - [ ] Scoring model backtested against historical wins and losses - [ ] Routing has specific-first rules, fallback owner, and logging - [ ] Dashboard spec covers marketing, sales, and executive views Any box unchecked: not done. Fix or say so.