I ran the same fictitious test lead through three stacks and compared what it actually costs at 1,000 leads/month.
Test lead: {name: Jane, email: jane.test@example.com, message: "Need follow-up for 40 open-house leads/mo"}. Flow in all three: webhook → validate → dedupe by email → CRM create → score → Slack alert → log. Success = CRM record + priority alert; failure = HTTP 400 + error log, nothing silent.
Measured: n8n self-hosted
- ~1 hour to build the 6-node flow (workflow file).
- $0 marginal runtime at 1k leads/mo on an already-provisioned VPS (log disk use TBD after 30 days live — estimate until then).
- Failure behavior proven live in the demo: bad email → HTTP 400 + error log, duplicate → merge.
- You own upgrades, backups, monitoring: honest ~1–2h/month (my estimate).
Cited: Zapier (official machine-readable pricing ref, updated 2026-09-09)
- Each successful action step = 1 task; triggers, polling, and built-ins (Filter/Paths/Formatter) = 0. This flow ≈ 4–5 tasks per lead.
- Free = 100 tasks/mo and two-step zaps only — this multi-step flow REQUIRES Pro.
- Pro 750 tasks = $19.99/mo annual ($29.99 monthly); Pro 5,000 = $89/mo annual.
- Per-1k math: 120 leads/mo ≈ 600 tasks → 750 tier; 1,000 leads/mo ≈ 5,000 tasks → $89/mo annual.
- Polling 15 min Free / 2 min Pro (webhooks instant on paid).
Cited: Make (official page + 3 corroborating 2026 guides, prices as a band)
- Billing unit is credits (renamed from operations in Aug 2025); most actions = 1 credit.
- Free = 1,000 credits/mo, 2 scenarios, 15-min interval. Core/Pro/Teams all base 10,000 credits/mo (~$12/$21/$38 annual) — tiers add features, not volume.
- This flow ≈ 5–6 credits/lead → 1,000 leads ≈ 5–6k credits → Core tier suffices.
Cited: CRM API limits (bind none of the three at SMB volume)
- HubSpot private apps: 100 req/10s burst (Free/Starter), 190 (Pro/Enterprise); daily 250k–1M (official docs).
- Pipedrive: burst by plan + daily token budget (official docs).
- GoHighLevel V2: 100/10s, 200k/day/location, no batch endpoints for most resources (secondary source — treat as approximate).
- Practical edge: HubSpot batch endpoints (100 records/call) matter for migrations, not per-lead intake.
My take (opinion)
- Zapier if the team already lives there, <500 leads/mo, and nobody wants infra. Pay per run for zero maintenance.
- Make if visual routing and error handling matter and someone technical exists.
- n8n self-hosted if per-run pricing stings, you want the workflow file in git, and someone owns the box.
Full disclosure: no clients yet on my side — that's exactly why the demo, the workflow file, and a fix-or-refund guarantee are public. I do fixed-price intake builds ($998–$4,500) and a $79 DIY starter pack for teams with someone technical. Happy to be corrected on any figure above — SaaS pricing moves, and I'll update the post.
Top comments (1)
Great breakdown. The per task math is exactly why I moved to Windsor.ai.
I think In the AI era, LLMs handle the logic the data pipe is what matters. Zapier’s per task pricing gets expensive when you’re just moving data on schedule ( ads, CRM, analytics -> warehouse/BI/LLM). Windsor is fixed monthly, 350+ sources, no per trigger or volume based fees.
For lead intake flows like this, n8n/Make still make sense. But the trigger based pricing model is due for a shakeup.