List-price math only. Real Datadog bills depend on per-hour billable-host sampling, indexed-event sampling vs raw GB, and contract-specific commit discounts. The deep audit reconciles your actual Datadog usage report against indexed-event filters and ranks reduction wins by dollar impact.
| Product | Unit rate | Your usage | $ / mo |
|---|---|---|---|
| — | |||
| Product | Pro rate | Enterprise rate | Billed by |
|---|---|---|---|
| Infrastructure | $15 / host / mo | $23 / host / mo | billable host |
| APM | $31 / host / mo | $40 / host / mo | billable host |
| Log ingest | $0.10 / GB | $0.10 / GB | raw GB ingested |
| Log indexing (15-day) | $1.27 / GB | $1.27 / GB | indexed GB / 15 days |
| Log retention (30-day) | 1.5x indexing | 1.5x indexing | retention tier multiplier |
| Synthetics browser | $5 / 10K tests | $5 / 10K tests | per test run |
| RUM sessions | $1.50 / 1K sessions | $1.50 / 1K sessions | per recorded session |
One-page checklist of the biggest Datadog levers — index exclusion filters for routine 200-OK access logs, debug-level filtering by environment, log sampling for high-volume endpoints, retention right-sizing per index, custom metric audit (the silent $1/100 charge), APM service map vs span-search cost gap, Synthetics test consolidation, RUM sample rate tuning, Network Performance Monitoring opt-out, container-host vs node-agent counting, the annual-commit negotiation lever, host autoscaling billable-host trap, and the dev-environment cluster billing mistake. PDF sent to your inbox.
monthly = hosts × tier_host_rate + apm_hosts × apm_rate + log_gb_day × 30 × $0.10 + (log_gb_day × 30 × index_pct%) × $1.27 × retention_mult + tests_K × $0.50 + sessions_M × $1500
Example A (default): Pro tier, 10 hosts, 50 GB/day log ingest, 0% indexed, 15-day retention, no APM, no synthetics, no RUM. Infrastructure: 10 × $15 = $150/mo. Log ingest: 50 × 30 × $0.10 = $150/mo. Log indexing: 0% indexed = $0. Monthly total: $300. Real teams typically index 5-15% of ingested logs via exclusion filters — start there.
Example B (typical 50-host shop): Pro tier, 50 hosts, 200 GB/day logs, 15% indexed, 30-day retention, 40 hosts with APM, 20K synthetic browser tests, 2M RUM sessions. Infrastructure: 50 × $15 = $750/mo. APM: 40 × $31 = $1,240/mo. Log ingest: 200 × 30 × $0.10 = $600/mo. Log indexing 15% × 30-day = (200 × 30 × 0.15) × $1.27 × 1.5 = $1,715/mo. Synthetics: 20K × $0.0005 = $10/mo. RUM: 2M × $0.0015 = $3,000/mo. Monthly total: $7,315. The same workload with 100% indexing would add $9,715/mo — index filters are the highest-impact single lever.
Datadog bundles 5+ products with two billing models stacked together: per-host (Infrastructure, APM, NPM, DBM) and per-volume (Logs, Metrics, Synthetics, RUM). A team starting at 10 hosts on Infra Pro pays $150/mo. Add APM and that becomes 10 × $31 = $310/mo on APM alone, on top of $150. Add 100 GB/day log ingest = $381/mo just for ingest before indexing. By the time a 50-host team enables Infra + APM + Logs + Synthetics, the bill routinely hits $5,000-15,000/mo. The killer is that ingest is unlimited but indexing is per-event — teams ingest everything at $0.10/GB but pay $1.70 per million indexed events, and at 5,000 events per GB that becomes $8.50/GB indexed, 85x more than ingest. Most teams discover this on the second invoice.
Pro at $15/host/mo is the cheapest for pure pricing, but Enterprise at $23/host/mo unlocks more custom metrics per host (200 vs 100), live container map, and other features. For most teams under 100 hosts using standard integrations, Pro is enough and saves $8/host/mo, or $9,600/year on 100 hosts. The break-even is custom metrics: if you would otherwise pay $1,920/year for 200,000 custom metrics on Pro, Enterprise's included pool starts to matter. Pro APM is $31/host/mo vs Enterprise APM at $40/host/mo, so adding APM actually flips the savings — at 50 hosts with APM, Pro saves $5,400/year over Enterprise. The honest answer is most teams should stay on Pro until a contract negotiation forces an Enterprise commitment.
Logs almost always cost more than hosts past a certain scale, and most teams underestimate by 5-10x. Ingest is $0.10/GB, which sounds cheap. The trap is indexing at $1.27/GB. At 100 GB/day ingest across 30 days, ingest alone is $300/mo. If you index everything, that becomes 100 × 30 × $1.27 = $3,810/mo, or 12.7x the ingest cost. A typical 20-host team pays $300/mo for hosts but $2,000-5,000/mo for logs. The lever is index exclusion filters: most teams index 100% of ingested logs when they only need to query 5-15%. Configuring filters to skip routine 200-OK access logs, debug-level entries, and high-volume noise sources cuts indexing by 60-80%. Retention also matters — the default 15-day retention is included, but every step above (30/45/60/90 days) adds 1.5-3x multipliers.
APM at $31/host/mo on Pro (or $40 on Enterprise) more than doubles your per-host cost — Infra Pro $15 + APM $31 = $46/host/mo. For a 50-host shop that is $27,600/year for APM alone, on top of the $9,000 Infra bill. The value question is whether you actually use distributed tracing, code-level profiling, error tracking with stack traces, and the service map. If your team only opens the service map once a week and never opens a span search, you are paying $7.75 per dev per month for read-only graphs. The honest break-even is roughly one prevented sev-2 incident per quarter — at $5,000-25,000 per major incident in lost revenue and engineer time, APM justifies itself if it surfaces one root cause faster than logs alone could. Keep APM if you run microservices, have measurable revenue tied to checkout latency, or debug perf regressions monthly. Drop it for monoliths, internal tools, or anywhere logs + metrics suffice.
No. Host counts, log volumes, APM enabled hosts, Synthetics test counts, and RUM session numbers all run locally in your browser. The page fires an anonymous pageview beacon and CTA-click events so we can measure whether the calculator is useful — no host names, no usage data, no account identifiers, no email (unless you submit one to the cheat-sheet form), no IP stored raw.