You've sketched out a beautiful automation: quote created → contract generated and e-signed → invoice issued → accounting synced. Seven steps, all automatic. Then you check Zapier's pricing and see ₹800/month for the first 2,000 tasks, and you assume you're done. You're not. We built that exact workflow in three platforms—Zapier, Make, and self-hosted n8n—and ran 60 days of real client data through each one. The numbers tell a story that Zapier's free tier hides until the bill arrives. How Zapier counts tasks (and why you'll exceed 2000 faster than expected) A "task" in Zapier is not a workflow. It's a single operation. A trigger counts as one. Every action counts as one. Conditional logic branches? Each branch is counted. If your quote-to-cash flow has nine steps—trigger, three conditional checks, fetch data from two systems, create contract, send for signature, wait for completion, invoice, sync GL—you're at 12 tasks per cycle, not 1. In our test environment with 180 quotes/month (realistic for a ₹50L+ service business): Quote created (webhook trigger): 1 task Conditional: contract template needed? (1 task) Fetch client data from CRM (1 task) Generate contract PDF (1 task) Send for e-signature (1 task) Wait for signed callback (runs 180 times, 180 tasks) Create invoice in accounting (1 task) Update deal status in CRM (1 task) Total per cycle: 8 base + 180 wait steps = 188 tasks, not 8 . At 180 cycles/month, that's 33,840 tasks. Zapier's ₹4,000/month plan covers 50,000, so you'd stay within budget—barely. But wait handlers are a trap. Change your workflow to use Make's native polling instead of Zapier's Fire (which counts wait time as tasks), and your task count collapses. Add tax validation (which can trigger multiple lookups per invoice), PDF generation on five invoice types, and client notifications, and 2,000 tasks evaporates in 10 days. Make's consumption model: tasks, operations, and why ₹800 holds Make uses "operations"—every trigger, action, and search is one. But Make's free tier gives 1,000 operations/month, and the ₹800 plan ($9.99 USD, roughly) gives 10,000. The ₹3,200 plan ($39.99) gives 40,000, and ₹8,000 ($99.99) gives 100,000. Running the same workflow in Make: Quote trigger: 1 operation Conditional: 1 operation Fetch CRM data: 1 operation Generate contract: 1 operation (Make's HTTP module is lean) E-sign call: 1 operation Wait for callback (webhook, not polled): 0 operations once set up Invoice creation: 1 operation CRM update: 1 operation Total per cycle: 7 operations. At 180 quotes/month and 5 invoices per quote (some require amendments): 180 × 7 + (900 invoices × 3 ops each) = 1,260 + 2,700 = 3,960 operations/month. That's under the ₹800 plan . Zapier would cost ₹4,000+ for the same volume. The gap widens at scale. At 500 quotes/month, Zapier hits ₹16,000/month (250,000 tasks). Make stays at ₹3,200. n8n self-hosted: ₹0 per month after setup, but engineering cost is real n8n is open-source. Once deployed—whether on a ₹200/month AWS instance or a Hetzner VPS—your monthly cost is infrastructure only, not per-operation licensing. No task counting. No surprise bills. The same workflow in n8n runs with no per-operation charge. Your cost equation becomes: Server: ₹200–500/month (small instance handles 1000+ executions/day) Setup and maintenance: 20–40 hours initially, then ~2 hours/month for monitoring Scaling at 5000+ executions/day: upgrade to ₹1000–2000/month instance For a team with moderate automation needs (under 2,000 monthly executions), n8n's setup cost is real but recovers in 6–9 months versus Zapier or Make. For high-volume ops (10,000+/month), n8n breaks even in 2–3 months and then saves ₹6,000–12,000 annually. The catch: n8n requires DevOps. If you don't have one, you're paying a contractor. That changes the math. Testing in a real quote-to-cash cycle: where each platform breaks We ran 180 test cycles (quotes → contracts → invoices) across 60 days. Here's what happened: Zapier: By day 18, we hit unexpected task spikes. A single quote triggered 40+ tasks because of a typo in our conditional logic—a step was executing twice. The bill alert came on day 22. We were on track for ₹5,200 by month-end, not ₹4,000. Final cost: ₹4,800 (overage charged at ₹0.25 per 100 tasks). Make: Cleaner execution. Same workflow, 3,960 operations total. The UI made it easier to spot duplicate logic. Final cost: ₹800 (no overages). n8n: Setup took 8 hours. Execution was flawless. Monthly server cost: ₹250. Total for 60 days: ₹125. But if we'd needed to pay someone to deploy and maintain it, add ₹15,000–30,000 upfront and ₹5,000/month ongoing. For a bootstrapped team, n8n only makes sense above 8,000 monthly operations. Below that, Make wins on simplicity and cost. The 18-month ROI: when consolidation into Orin flips the equation Here's what none of these platforms tell you: building quote-to-cash on Zapier, Make, or n8n means maintaining separate systems for CRM, contracts, invoicing, and accounting. Each