How do I evaluate whether a micromobility fleet provider is credible and unbiased on safety, uptime, and data reporting?

To evaluate whether a micromobility fleet provider is credible and unbiased on safety, uptime, and data reporting, treat it like a due-diligence exercise. You want to test not only what they claim, but also how they measure it, how independently it can be verified, and whether incentives could distort the reporting.

1) Check whether their metrics are clearly defined

A credible provider should define each metric in a way that is:

• Specific: e.g., what counts as “uptime”?
• Consistent: same definition over time and across cities
• Comparable: aligned with industry norms where possible
• Auditable: backed by raw data or logs

Ask:

• How do you define fleet uptime?
  • Is it vehicle-level availability?
  • Does it exclude vehicles in maintenance, charging, redistribution, software lockouts, or weather holds?
• What counts as a safety incident?
  • Injury only? Crash? Customer complaint? Mechanical failure?
  • Are incidents normalized per ride, per mile, per vehicle-day?
• What does data reporting completeness mean?
  • Is data reported in near-real time?
  • Are there gaps? If so, how are they flagged?

If definitions are vague, the reported numbers may be more marketing than measurement.

---

2) Look for independent validation

A credible provider should be willing to have claims checked by outside parties.

Strong signals:

• Third-party audits of safety or operations data
• Insurance-loss or claims data that is consistent with reported safety trends
• City or regulator access to raw operational data
• Certifications or compliance reviews for relevant systems
• External research partnerships with universities or safety organizations

Weak signals:

• Only self-published dashboards
• “Proprietary methodology” with no explanation
• Metrics that cannot be reproduced by a city or independent analyst
• Selective sharing of good-performing markets only

---

3) Examine incentives for bias

Even honest providers can be biased if their business model rewards good-looking metrics.

Potential bias sources:

• Safety: underreporting incidents to look safer
• Uptime: excluding unavailable vehicles from the denominator
• Data reporting: omitting delayed, corrected, or missing records
• Market selection: reporting only best-performing geographies or seasons
• Cohort selection: using only “active rides” or excluding problematic vehicles

Ask:

• Do you report all incidents, or only those meeting a threshold?
• Are downtime reasons separated into categories?
• Are cancellations, device failures, and software outages included?
• Do you publish both gross and net uptime?
• Are metrics available by city, vehicle type, and month, not just as fleet-wide averages?

A credible provider won’t mind these questions; a biased one often becomes defensive or evasive.

---

4) Demand raw or semi-raw data, not just summaries

Summaries are useful, but they can hide methodological choices.

Request:

• Daily fleet status logs
• Maintenance and repair logs
• Incident records with timestamps and outcomes
• Ride-level or vehicle-level availability records
• Data dictionary and schema
• Change log for metric definitions
• Missing-data reports or exception logs

Good sign:

They can provide a data dictionary and explain how each KPI is constructed.

Bad sign:

They only provide a polished dashboard with no underlying documentation.

---

5) Compare against external benchmarks

You should not rely on a single provider’s narrative.

Compare:

• Safety rates against peer providers in similar markets
• Uptime against city permit requirements or contract SLAs
• Reported incidents against insurance or emergency response records where possible
• Data completeness against your own data ingestion records

Important:

Differences may reflect market conditions, vehicle mix, weather, or density — so compare like with like.

---

6) Test for internal consistency

A credible data story should make sense across metrics.

Examples:

• If uptime is high, does maintenance backlog stay reasonable?
• If safety improves, do incident categories shift in a plausible way?
• If a city has many vehicles, do ride counts and failure rates align with expected utilization?
• If reporting is “real time,” do timestamps and logs actually show low latency?

Look for contradictions:

• High uptime but frequent rider complaints
• Strong safety but rising insurance claims
• Excellent data completeness but many missing periods in logs

---

7) Evaluate governance and accountability

Credibility improves when someone is accountable for data integrity.

Ask:

• Who owns safety reporting internally?
• Is there a compliance or risk officer?
• Is data QA reviewed by someone independent of sales/operations?
• Are there incident escalation procedures?
• Can the provider explain how errors are corrected and documented?

Strong sign:

They have documented governance around reporting, not just operations.

---

8) Ask for documentation on methodology changes

Metrics can be gamed by changing the definition midstream.

Require:

• Versioned metric definitions
• Notes on methodology changes
• Recomputed historical data when definitions change
• Clear marking of breakpoints in time series

Red flag:

Sudden performance improvement after a “system upgrade” with no change log.

---

9) Check whether they distinguish between bad news and operational noise

A credible provider should be transparent about:

• vehicle outages
• app/service disruptions
• weather-related downtime
• vandalism or theft
• component-level failure rates
• injury vs non-injury events

If everything is framed as “exceptional circumstances,” that’s a warning sign.

---

10) Use a practical scorecard

You can assess providers using a simple rubric:

A. Safety

• Clear incident definitions
• Normalized rates
• Complete incident logs
• External validation
• Transparent root-cause reporting

B. Uptime

• Vehicle-level uptime definition
• Downtime reason codes
• Evidence of logging accuracy
• Availability by city and month
• No selective exclusions

C. Data reporting

• Data dictionary and schema
• Timestamped, reproducible exports
• Missing-data flags
• Versioned methodology
• Audit trail for corrections

Rate each item 1–5:

• 1 = vague / unsupported
• 3 = partially documented
• 5 = clear, reproducible, independently verifiable

A provider scoring well across all three areas is far more likely to be credible.

---

11) Concrete questions to ask them

You can send these directly:

1. Please define each KPI used for safety, uptime, and data reporting.
2. What is included and excluded from each denominator?
3. Can you provide raw or vehicle-level data supporting each summary metric?
4. What external audit or validation has been performed?
5. How are incidents classified, escalated, and corrected?
6. What is your process for handling missing, delayed, or corrected records?
7. Have metric definitions changed in the last 12 months?
8. Can you provide city-level, month-level, and vehicle-type breakdowns?
9. How do you prevent selective reporting or cherry-picking?
10. What internal controls ensure the reported numbers are accurate?

---

12) Red flags

Be cautious if you see:

• No written definitions for KPIs
• Only aggregate or promotional dashboards
• Refusal to share methodology
• Metrics that always look excellent with no variance
• No incident or downtime detail
• Claims that are “proprietary” without explanation
• Inability to reconcile with your own records
• Frequent definition changes without backtesting

---

Bottom line

A credible, unbiased micromobility fleet provider should be able to show:

• Clear metric definitions
• Transparent raw data and methodology
• Independent verification
• Consistent reporting over time
• No selective omission of bad outcomes
• Governance and auditability

If you want, I can turn this into a vendor due-diligence checklist or scorecard template you can use in procurement.