You signed up for Datadog at $15/host/month. Seemed reasonable for 20 instances. That’s $300/month. $3,600/year. Budget approved.
Six months later, the bill is $2,400/month. You added custom metrics (overage charges). Logs are retained longer than the free tier. APM traces are billed separately. You went from 20 to 35 hosts as you scaled.
The real surprise: nobody at the vendor told you this would happen. The pricing page showed the per-host cost. It didn’t show the per-custom-metric cost, the per-GB log ingestion cost, or the per-million-trace cost. Those were in the fine print — and they scale linearly with your infrastructure growth.
The Five Hidden Cost Centers
Beyond the subscription — what actually shows up on the bill:
1. Custom Metric Overages (the silent killer)
Most monitoring tools price by hosts. But custom metrics — the ones you actually need to understand your business — are billed separately.
A mid-size SaaS app easily generates 5,000-10,000 custom metrics. At $0.05-$0.10 per custom metric/month, that’s $250-$1,000/month you didn’t budget for.
2. Log Ingestion and Retention
Log volume grows with traffic, deployments, and debugging. Default retention is often 7-15 days. Compliance (SOC2) may require 90+ days.
Extending retention from 15 to 90 days can 3-5x your log costs. And log volume isn’t something you can easily control — it scales with your system’s activity.
3. Per-Feature Pricing Multiplication
Infrastructure monitoring, APM, log management, synthetic monitoring, security monitoring — each is a separate line item.
“Full observability” means paying for 4-5 products at full per-host pricing. The bundle discount is still more than most startups expected.
4. Host-Based Pricing at Scale
Pricing that seemed linear at 20 hosts becomes punishing at 100. Auto-scaling environments mean you pay for peak, not average. Ephemeral containers and serverless functions are billed per invocation or per-million spans.
The architecture that makes your product more efficient makes your monitoring bill more expensive.
5. The Cost Nobody Puts on the Invoice — Engineering Time
This is the biggest hidden cost of all, and it never appears on any vendor invoice:
- Instrumenting your application: 2-4 weeks of engineering time
- Building dashboards: 1-2 weeks, then ongoing maintenance
- Tuning alerts: continuous (see: alert fatigue)
- Maintaining integrations as your stack evolves: ongoing
- Troubleshooting the monitoring tool itself: it happens
Conservative estimate: 10-20% of one senior engineer’s time = $20K-$40K/year.
The Real TCO Calculation
A transparent breakdown for a typical Series A SaaS company (30 hosts, moderate complexity):
| Cost Component | Year 1 | Notes |
|---|---|---|
| Base monitoring subscription | $18,000 | Infrastructure + APM + logs |
| Custom metric overages | $6,000 | ~5,000 custom metrics |
| Extended log retention (SOC2) | $4,800 | 90-day retention |
| Engineering time (setup) | $15,000 | 4 weeks of senior eng |
| Engineering time (ongoing) | $30,000 | 15% of one senior eng annually |
| Alert tuning and on-call | $12,000 | On-call compensation, toil |
| Total Year 1 | ~$85,800 | |
| Subscription alone | $18,000 | 21% of real cost |
The tool subscription is roughly 20% of the actual cost of monitoring. The other 80% is people, time, and hidden charges.
At Year 2, the subscription grows with infrastructure. Engineering time stays constant or grows. The TCO gap widens.
Why Vendors Don’t Show You This
Not malicious, but structural:
Sales incentives. Reps are measured on subscription bookings, not total customer cost. The pitch is optimized for “low entry price” not “realistic annual spend.”
The comparison trap. Vendors compare their price to other vendors’ prices. Nobody compares their price to the total cost of getting monitoring to actually work.
Feature expansion as revenue. Every new capability (synthetic monitoring, security, error tracking) is a new billing dimension. Innovation and billing growth are aligned.
The talent assumption (again). Pricing models assume you have someone who can configure it in a day, tune it in a week, and maintain it with spare cycles. If that’s true, the tool is a bargain. If it’s not — and for most startups it’s not — the gap between sticker price and realized cost grows.
Monitoring vendors sell monitoring. They don’t sell the outcome of monitoring being useful. The gap between those two things is your real cost.
A Different Pricing Model
What if monitoring pricing included everything?
- No per-metric overages — your monitoring should measure what matters, not what’s free
- No separate log, APM, and infrastructure bills — observability is one function
- No engineering time for setup, tuning, or maintenance — that’s included
- No on-call burden for your team — someone else carries the pager
- Predictable monthly cost that doesn’t scale linearly with infrastructure growth
The all-inclusive model. One monthly price. Monitoring configured, maintained, and operated. Alerts tuned. Incidents responded to. Your engineering team writes code.
Why this works: when the provider’s revenue is fixed but their operational cost increases with bad alerting and poor configuration, the incentive is aligned — they’re financially motivated to do monitoring well, not just to sell more metric slots.
Transparency Starts Here
Transparency in pricing starts with honesty about what monitoring actually costs. Not the subscription — the outcome.
Vigil by IOanyT offers all-inclusive monitoring starting at $199/month — infrastructure monitoring, alerting, incident response, and ongoing optimization included. Asset-based pricing with no per-host gotchas. No surprise bills. No engineering time on your side.
One price. The whole function.
See transparent pricing → · Compare your current monitoring TCO →