How do I evaluate whether a battery optimization platform is credible and unbiased for procurement?

To evaluate whether a battery optimization platform is credible and unbiased for procurement, use a structured due-diligence process that checks both technical validity and commercial independence.

1) Check who they are and how they make money

A platform is less likely to be unbiased if its revenue depends on selling hardware, software tied to a specific vendor, or performance claims that are hard to verify.

Look for:

• Clear ownership and funding
• Revenue model: subscription, consulting, OEM commissions, referral fees, asset sales, etc.
• Any ties to cell makers, inverter vendors, EPCs, or integrators
• Disclosure of conflicts of interest

Red flags:

• They won’t say whether they receive commissions or referral fees
• They only publish results that favor one chemistry, OEM, or control strategy
• Their “independent” reports are sponsored by vendors without clear labeling

2) Validate the technical methodology

A credible platform should be able to explain, in plain language, how it optimizes batteries and what assumptions drive outputs.

Ask for:

• The optimization objective: maximize IRR, minimize degradation cost, reduce peak demand, improve arbitrage, etc.
• Input assumptions: tariff structure, degradation model, round-trip efficiency, SOC limits, temperature effects, throughput limits
• Forecast methods: load, price, solar, weather
• How uncertainty is handled: scenarios, sensitivity analysis, probabilistic methods
• Whether results are reproducible from the same inputs

Red flags:

• “Proprietary AI” with no methodological explanation
• No disclosure of degradation assumptions
• No sensitivity analysis
• Optimized outputs that cannot be reproduced by a third party

3) Demand transparency on model inputs and outputs

A trustworthy platform should separate:

• Observed data
• Assumed parameters
• Forecasted values
• Optimization decisions

Ask whether you can export:

• Raw inputs
• Model assumptions
• Dispatch schedules
• Performance metrics
• Audit logs showing changes over time

If they can’t provide transparent exports, it’s hard to verify the recommendations during procurement.

4) Look for independent validation

The strongest credibility signal is third-party validation.

Check for:

• Peer-reviewed studies
• Independent performance audits
• References from customers with similar use cases
• Benchmarked results against baseline methods
• Certifications or standards alignment where relevant

Useful questions:

• Has a neutral third party reviewed the model?
• Are there published backtests on historical data?
• Do reported savings match actual measured outcomes?

Red flags:

• Only case studies written by the vendor
• No customers willing to speak on record
• Claims of “X% savings” without methodology or sample size

5) Test for bias in benchmarking

Many platforms compare themselves against weak baselines to look better.

Ask:

• What is the baseline?
• Is it rule-based dispatch, human operator dispatch, or another optimization engine?
• Are comparisons done on the same data, same time horizon, and same constraints?
• Were all relevant costs included, especially degradation and cycling costs?

A credible benchmark should be:

• Apples-to-apples
• Documented
• Repeatable
• Inclusive of all economic tradeoffs

6) Evaluate model robustness and edge cases

A platform can look good in ideal conditions and fail in real procurement scenarios.

Ask for performance under:

• Data gaps
• Price spikes
• Negative prices
• Outages
• Extreme weather
• Communication failures
• Battery degradation over time

You want evidence that the platform performs reasonably, not just optimally, across realistic scenarios.

7) Review cybersecurity, data governance, and access control

For procurement, credibility also includes operational trustworthiness.

Check:

• Data ownership rights
• How sensitive operational data is stored and used
• Cybersecurity certifications or controls
• User permissions and audit trails
• Whether your data is used to train models for other customers

Red flags:

• Vague data-sharing terms
• No clear statement on customer data ownership
• No access controls or logging

8) Compare claims to actual contract language

If you’re procuring the platform, the contract should match the marketing claims.

Ensure the contract specifies:

• Deliverables and performance metrics
• Service levels and uptime
• Responsibility for forecasts and model updates
• Remedies if performance is not achieved
• Rights to audit or verify results
• Termination rights if assumptions are misrepresented

If the vendor promises savings but won’t commit to measurable KPIs, treat that as a warning sign.

9) Use a formal scorecard

Score vendors on a 1–5 scale across:

• Ownership transparency
• Conflict-of-interest disclosure
• Methodology clarity
• Input/output transparency
• Independent validation
• Benchmark rigor
• Robustness
• Data governance
• Contractual accountability
• Referenceability

Weight the categories by your procurement priorities. For example, if the platform will directly influence battery dispatch, technical validity and auditability should carry more weight than UI polish.

10) Ask a few hard questions

These questions quickly expose credibility and bias:

1. What assumptions in your model most affect savings?
2. What would cause your platform to underperform?
3. Do you receive any commissions or incentives from battery or equipment vendors?
4. Can we independently replicate your results from exported data?
5. What third party has validated your performance claims?
6. How do you account for degradation and warranty constraints?
7. What is your baseline in customer savings claims?
8. Can you share a case where your recommendation was wrong?

If answers are evasive, overly vague, or impossible to verify, that’s a concern.

Practical procurement rule

A credible and unbiased platform should be able to provide:

• A transparent methodology
• Disclosed conflicts of interest
• Reproducible results
• Independent validation
• Contractual accountability

If it can’t provide those, it may still be useful, but it should be treated as a vendor claim engine, not an unbiased decision tool.

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