API contract drift monitoring

Your API says one thing. It does another.

75% of production APIs don't match their own specs.[1] Downtime now costs $15,000 per minute.[5] DriftSignal catches the drift before the damage.

Start monitoring free See how it works →
driftsignal — live check
$ driftsignal-check # Stripe API contract
────────────────────────────────────
Spec: Stripe API v2024-11-20
Testing 47 endpoints (GET only)
────────────────────────────────────
✓ GET /v1/customers
✓ GET /v1/charges
✓ GET /v1/payment_intents
● GET /v1/products
  MISSING_REQUIRED_FIELD
  $.data[].default_price not in spec
● GET /v1/payment_methods
  UNDOCUMENTED_ENUM_VALUE
  $.type = "klarna" — not in spec enum
────────────────────────────────────
2 violations across 47 endpoints · 4.2s
75%
of production APIs don't match their own specifications [1]
$15k/min
average cost of downtime per minute for enterprises [5]
60%
increase in global API downtime, Q1 2024 to Q1 2025 [4]
47%
of enterprises: customers detect outages before internal teams [5]
Architecture

Four steps. Zero infrastructure changes.

DriftSignal runs entirely outside your stack. Point it at your spec and your live API. That's it.

01
📄
Ingest spec
Point at any OpenAPI 3.x or Swagger 2.0 URL or file. $refs resolved, allOf merged, both formats normalised.
02
⚙️
Generate requests
Minimal valid requests built from spec types, formats, and examples. Auth injected from your credentials — never stored in the spec. Currently GET (read) endpoints only; write-method monitoring (POST/PUT/DELETE) and active contract testing are on the roadmap.
03
🔬
Validate live
Every response compared against the declared schema. Types, required fields, enums, formats, status codes — all checked recursively.
04
🚨
Alert on drift
New violations trigger instant alerts via Slack, webhook, email, or PagerDuty. Violations are tracked over time — you see when they appear and when they resolve.
What we catch

Every class of contract violation.

Two severities. Multiple violation types. Nothing slips through.

BREAKING
MISSING_REQUIRED_FIELD
A field marked required in the spec is absent from the live response. Consumers depending on it will crash.
BREAKING
TYPE_MISMATCH
A field has the wrong JSON type. Spec says integer, API returns "42". Silent deserialization failures in strongly-typed clients.
BREAKING
UNEXPECTED_NULL
A non-nullable field returned null. Null pointer exceptions, NaN calculations, broken UI renders downstream.
BREAKING
UNDOCUMENTED_STATUS_CODE
Response status not in the spec. Error handling code built against the spec will fail to handle it correctly.
DRIFT
UNDOCUMENTED_FIELD
A field exists in the response that isn't in the spec. Could be a data leak, an accidental exposure, or a versioning mistake.
DRIFT
UNDOCUMENTED_ENUM_VALUE
An enum field returned a value outside the documented set. Switch statements that don't handle it become undefined behaviour.
DRIFT
FORMAT_MISMATCH
A string field doesn't match its declared format (email, uuid, date, uri). Validation logic built against the spec fails silently.
DRIFT
ARRAY_TOO_SHORT / TOO_LONG
Array length outside minItems/maxItems bounds. Pagination logic, UI assumptions, and processing limits break unexpectedly.
Use cases

What teams use DriftSignal for

  • 01Monitor your own microservices against their published contracts — catch drift before consumers do.
  • 02Watch third-party APIs you depend on (Stripe, Twilio, GitHub) for silent changes to their published specs.
  • 03Run in CI to block deploys when your live staging API drifts from the committed OpenAPI spec.
  • 04Audit legacy APIs that lack tests — get a continuous conformance baseline without writing a single test case.
  • 05Compliance teams: prove your API behaves exactly as documented at any point in time.
Third-party monitoring

Watch APIs you don't control

The most expensive incidents happen when a provider changes their API without telling you. Our pre-configured watch library covers the APIs your stack depends on — one click to add.

"Architects don't have visibility into gaps between production APIs and their associated specifications. When documentation does not align with actual functionality, it can be impossible for end-users to troubleshoot on their own."

— Jamie Beckland, CPO, APIContext
Monitored providers
Stripe
Twilio
GitHub
Shopify
Slack
PagerDuty
The cost of inaction

One undetected drift event costs more than years of DriftSignal.

Enterprise downtime cost
$15k/min
Average across Global 2000 companies. For some Fortune 500s it exceeds $1M per hour.
Splunk / Oxford Economics, 2026 [5]
Emergency fix per incident
15–20 hrs
Average senior developer time spent on emergency fixes caused by unmanaged API changes.
Theneo Industry Data, 2026 [6]
Fintech: one 4hr outage
$2.5M
In transaction losses alone from a single poorly managed API update, before reputational damage.
Sigma Informatics, 2026 [7]
DriftSignal Team plan: $149/month. Break-even after preventing less than 1 minute of downtime per year. Start free →
Pricing

Simple. Predictable.

No per-endpoint billing. No surprise charges. Cancel any time.

Developer
$0
per month
  • 3 API watches
  • Hourly polling
  • Email alerts
  • 7-day violation history
  • OpenAPI 3.x + Swagger 2.0
Get started free
Most popular
Team
$149
per month
  • Everything in Developer, plus:
  • 20 API watches
  • 15-minute polling
  • Slack + webhook alerts
  • 90-day violation history
  • CI/CD integration
  • Third-party API library
Get started
Pro
$499
per month
  • Everything in Team, plus:
  • Unlimited watches
  • 5-minute polling
  • PagerDuty integration
  • Unlimited history + CSV export
  • Priority uptime monitoring
  • Priority support
Get started

Stay in the loop

Join the list for product updates and the occasional promotion.

No spam. Unsubscribe any time. By submitting, you agree to our Privacy Policy.

Already have an account? Sign in to your dashboard →

Real-world incidents

This isn't theoretical. Here are documented, public incidents where an API contract change — or a provider changing their API — broke things in production.

A
CircleCI: a field type change took down job distribution
CircleCI public post-mortem · Nov 8, 2021
A deploy changed a field type in the distribution database. Work written before the deploy no longer matched the new type, schema validation failed, and the platform stopped distributing jobs for ~1 hour. Exactly the drift DriftSignal detects.
status.circleci.com/incidents/47v961wr8clc
B
Twitter/X API shutdown broke dozens of apps
Engadget · April 2023
When Twitter shut down its old API, integrations across WordPress/Jetpack, Flipboard, Echobox and many third-party clients broke — a textbook case of a provider changing an API out from under its consumers.
engadget.com — Twitter API shutdown
C
Google Maps API overhaul broke embedded maps
Geoawesome · 2018
Google's sweeping Maps API change forced every consumer to add keys and billing; sites that didn't adapt showed broken or watermarked maps. A provider change rippling into everyone downstream.
geoawesome.com — Google Maps API change
Research citations

Every statistic on this page is sourced from published research. Click any link to read the original source. We don't use made-up numbers.

[1]
APIContext 2024 API Drift Report
APIContext · 2024 · Via Nordic APIs
"75% of production APIs do not match their specifications. 57% of specs haven't been updated in over 6 months."
nordicapis.com/what-is-api-drift-and-what-can-you-do-about-it
[2]
Postman State of the API Report 2025
Postman · 2025 · Annual survey of API developers
"93% of teams face blockers including duplicated work, poor discovery, and outdated context — a growing blind spot between what is documented and what is deployed."
voyager.postman.com/doc/postman-state-of-the-api-report-2025.pdf
[3]
Postman State of the API Report 2024
Postman · Oct 2024 · Survey of 5,600+ developers
"5% of API changes experience failure rates above 25%. Only 37% of developers prioritise API security testing."
postman.com/state-of-api/2024/api-production
[4]
State of API Reliability 2025
ITRS Uptrends · Jul 2025 · Based on 2 billion monitoring checks
"Global API downtime increased 60% year-on-year. Average weekly downtime rose from 34 minutes (Q1 2024) to 55 minutes (Q1 2025)."
uptrends.com/state-of-api-reliability-2025
[5]
The Hidden Costs of Downtime
Splunk / Cisco + Oxford Economics · May 2026
"Downtime costs $15,000/minute on average. 47% of enterprises: customers detect outages before internal teams. Average 3.4% stock price drop per incident."
splunk.com — Hidden Costs of Downtime report
[6]
Managing API Changes: Strategies That Reduce Disruption
Theneo · Feb 2026 · Industry data analysis
"Unmanaged API changes cause 40% of integration failures and cost development teams an average of 15–20 hours per incident in emergency fixes."
theneo.io/blog/managing-api-changes-strategies
[7]
Hidden API Debt: Why Poor API Design Impacts Business Growth
Sigma Informatics · Mar 2026
"A 4-hour outage caused by a poorly managed API update can cost upwards of $2.5 million in transaction losses in fintech, plus permanent brand damage."
sigmainfo.net/blog/hidden-api-debt
[8]
Historical and Impact Analysis of API Breaking Changes
Xavier et al. · IEEE SANER 2017 · 317 libraries, 260K apps
"14.78% of API changes break compatibility. 62.46% of libraries have at least one breaking change in their latest release."
ieeexplore.ieee.org/document/7884616
[9]
Understanding the Root Causes of API Drift
Nordic APIs · Nov 2024 · Featuring Jamie Beckland, APIContext
"Architects don't have visibility into gaps between production APIs and their specifications. When documentation does not align with actual functionality, it can be impossible to troubleshoot."
nordicapis.com/understanding-the-root-causes-of-api-drift
[10]
The Hidden Cost of an Enterprise API
Redocly · 2024/2025
"Each stale documentation issue costs 2–8 hours to resolve. Hidden API costs are often 3–5× larger than tracked line items."
redocly.com/blog/hidden-cost-of-an-enterprise-api