The sequence RevenueHoop runs when a new client hands us a HubSpot with 47 custom properties, three different MQL definitions, and a forecast that swings 30 points quarter to quarter.
The 90-day RevOps playbook that works when you inherit a broken stack: Days 1-15, lock definitions across sales, marketing, and CS (MQL, SQL, stage exits, ownership). Days 15-30, fix data hygiene and stand up three baseline dashboards. Days 30-60, rebuild lead routing and account scoring on the new definitions. Days 60-90, restore forecast accuracy and set the weekly reporting cadence. Everything else: AI agents, complex attribution, enablement content, multi-CRM unification waits until day 91. This is the sequence we run in every RevenueHoop client engagement.
Why the first 90 days matter more than the next 900
RevOps truly compounds so whatever definitions, data model, and routing rules exist on day 90 become the foundation that every future change inherits.
That's why the biggest failure mode we see in new RevOps hires (and new agencies) is picking the wrong first fix.
The most common wrong first fix in 2026 is AI. A new VP of RevOps shows up, sees the mess and wants to demonstrate impact fast so they spin up Claude Code workflows and Clay agents on top of a broken foundation. The AI agents still run smoothly and amplifies the same misaligned definitions, dirty data, and broken routing that were there on day one.
The right first fixes are the boring ones: locking definitions, cleaning data, rebuilding routing and restoring forecast.
This is a tactical playbook. Each phase has a specific deliverable, common trap, and the tools we use to ship it.
Days 1-15: Audit and definitions
Book a two-hour meeting with the head of sales, head of marketing, and head of CS, all the relevant stakeholders. This is the highest-leverage 120 minutes of the entire 90 days.
The agenda is deceptively simple:
- What is our current definition of an MQL? An SQL? An SAL?
- What are the criteria to move a deal from stage 1 to stage 2? Stage 3 to 4? Stage 4 to closed-won?
- Who owns the lead at each stage?
- What are the SLAs at each handoff?
Write down every answer and do not try to reconcile disagreements in the room. Capture all the misalignments.
Most teams walk into this meeting believing they already have shared definitions. In our experience running this exercise across dozens of engagements, the head of sales and head of marketing use the same words to mean different things at least 60% of the time. The 90-day playbook fails immediately if you skip this because "we already have definitions."
The deliverable
One shared document, signed off by all stakeholders that defines:
- MQL, SQL, SAL, Opportunity, Closed-Won criteria (with specific field requirements)
- Ownership at each stage
- SLAs at each handoff
- Stage-exit criteria
The common trap to look out for are definitions that are directional but not testable. "MQL means the lead is qualified for a sales conversation" is not a definition. "MQL means: Company size 50-500, Job title contains VP/Director/Head, Fit score above 40, Intent score above 60, has visited pricing page in last 14 days" is a definition. If your definition doesn't map to specific CRM fields with specific values, it isn't operational.
The tools
HubSpot or Salesforce for the actual field definitions. Notion or Google Docs for the shared definitions document. Nothing fancy but if they have internal tools, you can always pencil them in.
Days 15-30: Data hygiene and baseline metrics
Data quality is the #1 pain point for 71% of RevOps teams in 2026, according to industry surveys. It's also the layer that determines whether every downstream fix (routing, scoring, forecasting) actually works.
Two workstreams run in parallel over these two weeks:
Workstream 1 — Dedup and hygiene.
Run duplicate detection on Contacts and Companies. HubSpot has native dedup; Salesforce needs an add-on. For edge cases (fuzzy matching on domain variants, subsidiary companies), specialized tools like Insycle or Openprise handle what the CRM native tools miss.
Enrichment gap analysis: pull a sample of 100 target accounts and audit which key fields are missing. If routing depends on Industry and Employee Count but 40% of accounts are missing both, routing can't work. Fix the data before rebuilding routing.
Workstream 2 — Baseline metrics.
Stand up three dashboards. These are the foundational metrics you'll measure against for the next 12 months:
- Pipeline by stage — count and value at each stage, updated daily
- Forecast accuracy — predicted vs. actual, tracked weekly per rep
- Conversion rate by stage — historical baseline, calculated once, refreshed monthly
Don't try to build fancy attribution dashboards yet. Baseline first, sophistication later.
The deliverable
- Dedup process running weekly (automated, not manual)
- Enrichment plan approved with a specific vendor stack (Clay for orchestration; ZoomInfo/Cognism/LeadIQ/Apollo as underlying data sources based on ICP geography)
- Three baseline dashboards live in your CRM or BI tool
The common trap here is trying to build a perfect data foundation before moving on. Perfect data is a rounding error away from "good enough" data. Set the standard at 85-90% completeness on key routing/scoring fields, then move forward. Chasing 99% will consume the entire 90 days.
The tools
CRM native dedup (HubSpot Manage Duplicates, Salesforce Data.com), Insycle or Openprise for edge cases, Clay for enrichment orchestration, Equals or native CRM reports for baseline dashboards.
Days 30-60: Routing and scoring rebuild
Now that definitions are locked and data is clean, routing and scoring can be rebuilt on solid ground.
Lead routing rebuild.
Write out the new routing rules based on the definitions locked in Phase 1. Common patterns:
- Inbound demo request → route by territory + ICP fit + assignment round-robin
- MQL from marketing campaign → route by territory + lifecycle stage
- High-intent visitor identified by RB2B or Warmly → route to same-day outreach queue
- Enterprise account (over threshold) → route to enterprise SDR pod
Build the routing in a dedicated tool. For HubSpot-native teams, Default is the modern choice. For Salesforce enterprise orgs, LeanData is the standard. Some mid-market teams get away with HubSpot's native routing plus a bit of workflow logic, that's fine for simple cases but breaks at complexity.
Critical implementation detail: run shadow routing for two weeks before cutting over. Route leads through both the old rules and the new rules simultaneously to compare outcomes, fix bugs and then cut over.
Account scoring rebuild.
Score on three dimensions: fit, intent, and timing.
- Fit. Firmographic + technographic match to ICP (industry, size, tech stack). Data from Clay + enrichment vendors.
- Intent. Behavioral signals (website visits, content engagement, competitor research). Data from RB2B, Warmly, 6sense, or Common Room depending on stage.
- Timing. Recent triggering events (funding, hiring, exec changes, tech stack changes). Data from Clay + Crustdata for hiring/tech signals.
Combine all three into a single account score. Weight based on what actually predicts closed-won in your historical data. If you don't have enough historical data yet (early-stage company), start with equal weights and refine after 90 days of new data.
SLAs.
Set the standard: 5 minutes for hot inbound (Enterprise MQL, ICP-fit demo request), 1 hour for warm (standard MQL), 24 hours for cold (unqualified but interested). Track SLA compliance on a dashboard the whole team can see.
The deliverable
- New routing rules live in production
- New account scoring model live
- SLA tracking dashboard live and visible to sales + marketing leadership
The common trap here is building routing that's more sophisticated than the sales team can actually operationalize. If your routing has 47 branches, three fallback rules, and requires a scoring model no rep understands, it will drift into a black box within 90 days. Simple beats sophisticated so start with the smallest set of rules that captures 80% of the value.
The tools
Default or Chili Piper (HubSpot-first). LeanData (Salesforce enterprise). RevenueHero (modern HubSpot alternative). Clay for enrichment logic behind scoring. RB2B, Warmly, or 6sense for intent signals depending on scale.
Days 60-90: Forecast rebuild and reporting cadence
With definitions locked, data clean, routing rebuilt, and scoring live, the forecast can finally be trusted.
Pipeline stage enforcement.
For each stage of your pipeline, enforce specific stage-exit criteria. A deal cannot move from stage 3 to stage 4 unless it has X, Y, and Z documented. Use required fields, validation rules, or workflow automations to enforce this and don't rely on rep discipline alone.
Sample criteria for a B2B SaaS deal:
- Stage 1 → Stage 2: Champion identified, first meeting completed, decision timeline established
- Stage 2 → Stage 3: Economic buyer identified, technical fit confirmed, budget range confirmed
- Stage 3 → Stage 4: Verbal commitment, procurement path defined, close date within 30 days
The stages you use may differ, but every stage needs objective criteria. If AEs can move deals without objective triggers, the forecast is guesswork.
Forecast accuracy dashboard.
Predicted revenue vs. actual revenue tracked weekly, broken down by rep, by stage, by pipeline segment. This is the single most important dashboard in RevOps, it's the number the board asks about, and it's the number that surfaces whether the machinery you rebuilt actually works.
Target metric: forecast accuracy above 75% within 60 days of launching this dashboard. That's a defensible number for most mid-market B2B SaaS orgs.
Weekly reporting cadence.
30-minute weekly forecast call with sales leadership. Review every deal above threshold, every deal that slipped from last week's forecast and this cadence is where the forecast dashboard becomes a management discipline.
The deliverable
- Pipeline stages enforced with objective criteria (workflow validated)
- Forecast accuracy dashboard live and improving week-over-week
- Weekly reporting cadence with sales leadership running
The common trap here is waiting for perfect forecast accuracy before running the cadence. Start the weekly cadence at day 60 even if accuracy is only 55%. The cadence itself surfaces the misalignments that need fixing. If you wait until accuracy is "good," you'll never run the cadence, and the forecast never improves.
The tools
Native CRM forecasting (HubSpot Forecast, Salesforce Forecasts) as baseline. Clari, BoostUp, or Gong Forecast for teams past mid-market scale. Equals or a BI tool for the accuracy dashboard.
What to defer past day 90
The trap most teams fall into: trying to do everything at once. AI, attribution, coaching, multi-CRM unification, marketing automation depth because all of it feels important.
Explicit list of what to defer until at least day 91:
- AI agents and Claude Code workflows. These amplify whatever is already there. Build them once the foundation is solid.
- Complex multi-touch attribution. Worthless until definitions are locked and every attribution model is downstream of the definitions.
- Enablement content and rep coaching programs. Comes from real call data (Gong, Fathom, Granola) and real forecast data, which you don't have yet in a trustworthy form.
- Multi-CRM data unification. Expensive, complex, and rarely business-critical in the first 90 days. If you have both HubSpot and Salesforce, pick one as the source of truth and defer unification.
- Marketing automation deep integration. After the foundation is solid, marketing automation can plug into definitions and routing cleanly. Before then, it just adds noise.
The teams that ship the strongest RevOps functions defer these things. The teams that struggle try to do all of it in the first 90 days.
What's next
RevOps is a compounding discipline. The first 90 days set the foundation; the next 12 months build on it. Once you've completed this playbook, the natural extensions are:
- AI-augmented workflows — Claude Code, MCP integrations, and Clay orchestrations layered on top of the solid foundation
- Signal-triggered outbound — using intent signals (RB2B, 6sense, Common Room) to trigger same-day outbound sequences
- Attribution modeling — with definitions locked, attribution finally becomes meaningful
- Enablement programs — using conversation intelligence (Gong, Fathom, Granola) to feed coaching from real call data
None of those work well if the 90-day foundation isn't solid. All of them compound if it is.