Data Quality Metrics
RateAPI scrapes mortgage, auto, HELOC, personal loan, and credit card rates from 4,300+ credit unions daily. As of July 16, 2026, 97.2% of scheduled scrapes succeed, 94% of rates are verified within 7 days, and rate accuracy measured against manual samples is 99.2%. Anomalous or stale data is never served as current.
Last updated: July 16, 2026
Scraping Frequency
RateAPI scrapes most credit union rate sources daily, and the top 100 institutions by volume twice daily. This continuous schedule keeps data fresh while respecting source website resources.
Validation Process
Every scraped rate passes through multiple validation layers before being served via the API.
Schema Validation
All extracted data must conform to our schema: rate (number), apr (number or null), points (number), term (months), productType (canonical category).
Range Checks
Rates must fall within expected ranges: mortgage rates 3-12%, auto loans 2-25%, HELOCs 5-15%, personal loans 5-36%. Out-of-range values are flagged.
Consistency Checks
APR must be greater than or equal to rate. 15-year rates should be lower than 30-year rates. Jumbo rates should be near conforming rates.
Historical Comparison
Changes exceeding 50 basis points in 24 hours trigger review. Complete rate disappearance triggers investigation.
Cross-Source Verification
When possible, rates are verified against multiple pages within the same institution's website.
Error Handling
When errors occur, we follow a systematic process to minimize impact on data quality.
Scrape Failures
If a scrape fails, we retry with exponential backoff (1h, 4h, 12h). After 3 failures, the source is marked for manual review. Previous valid data is retained with a staleness flag.
Parsing Errors
When page structure changes break our parsers, we detect this via empty or malformed results. AI-assisted extraction attempts recovery. Human review follows if needed.
Data Anomalies
Anomalous data (rate jumps, impossible values) is quarantined and excluded from API responses until reviewed. We never serve unverified anomalous data.
Historical Accuracy
Over the last 30 days, RateAPI achieved 99.2% rate accuracy and 98.7% APR accuracy when scraped values were compared against manually verified samples. We track accuracy by random sampling against source websites.
Accuracy is measured by random sampling and manual verification against source websites. Discrepancies are investigated and corrected.
Quality Guarantees
No Stale Data
Data older than 7 days is marked with a staleness flag. Data older than 14 days is excluded from default API responses.
Source Attribution
Every rate includes the source URL where it was observed, allowing independent verification.
Timestamp Transparency
Every response includes observed_at timestamp showing exactly when the data was collected.
Correction Lineage
When corrections are made, we maintain full audit history. Original values are preserved with correction reason codes.
Data Quality FAQ
How fresh is RateAPI rate data?
94% of rates are verified within the last 7 days and 95% are observed within 24 hours. Most sources are scraped daily; the top 100 credit unions by volume are scraped twice daily.
How accurate is the data?
Over the last 30 days, RateAPI achieved 99.2% rate accuracy, 98.7% APR accuracy, and 99.5% product classification accuracy against manually verified samples.
What happens when a scrape fails?
Failed scrapes retry with exponential backoff (1h, 4h, 12h). After 3 failures the source is flagged for manual review, and the last valid rate is retained with a staleness flag.
Does RateAPI ever serve anomalous rates?
No. Anomalous data is quarantined and excluded from API responses until reviewed, and data older than 14 days is excluded from default responses.
Learn more in our full methodology, see the data behind our mortgage rate benchmark, or browse the credit unions we track.
Questions About Data Quality?
We're committed to transparency. If you have questions about our data collection or find discrepancies, please reach out.