CostToKeep Methodology
CostToKeep produces estimates and decision-support information, not predictions or guarantees. We combine available product specifications, government data, manufacturer documentation, observed market information, user inputs, and documented calculation methods. The amount and specificity of evidence varies by appliance and model.
Data Types We Use
- Manufacturer Information: Official product specifications, owner manuals, support documents, and verified warranty information when available.
- Observed Market Information: Dated observations such as compatible OEM part prices and published service-market ranges. These values can change over time and should not be treated as permanent prices.
- Category Benchmarks: Broader product-category information used when exact model-specific evidence is unavailable. Category-level data is identified as such.
- User Inputs: Information entered by the user, such as appliance age, technician quote, replacement price, electricity rate, maintenance estimate, or repair allowance.
- Calculated Estimates: Results produced from documented formulas using available data and user inputs.
- Insufficient Data: When available evidence is too weak to support a useful estimate, CostToKeep leaves the metric unavailable rather than inventing a value.
Government Data
Public datasets and regulatory information such as ENERGY STAR product specifications, Department of Energy lifespan analysis, EIA electricity-price data, and CPSC recall information.
Evidence Fallback
Exact variant → Model family → Series → Manufacturer/type → Refrigerator type → Category → Insufficient data
When broader information is used because exact model-specific evidence is unavailable, CostToKeep identifies that fallback in the result or supporting data notes.
Repair Cost Method
When exact compatible parts data is available, CostToKeep may use observed parts pricing as one input. Labor may come from a user-provided technician quote or supported market information. CostToKeep does not currently assume fixed component repair times when verified time evidence is unavailable.
Repair or Replace
Repair-versus-replace decisions consider repair cost, appliance age, replacement cost, available service-life evidence, warranty, parts availability, and practical user considerations. Percentage rules such as 50% may be useful heuristics, but they do not determine the recommendation by themselves.
True Cost of Ownership
Expected Repair Allowance is an assumption or budgeting input, not a predicted failure probability.
Lifespan
When model-specific lifespan evidence is unavailable, CostToKeep may use a broader category benchmark. For refrigerators, the current DOE category-level analysis reports a 14.3-year mean lifetime for standard-size refrigerators/refrigerator-freezers and 8.8 years for compact refrigerators. These values are category-level averages, not predictions for an individual model.
Confidence Levels
- High: Strong model-specific evidence from credible sources.
- Medium: A combination of model-specific and broader evidence.
- Low: Limited model-specific evidence with greater reliance on broader benchmarks or user inputs.
Confidence describes evidence strength and specificity, not statistical certainty about future appliance performance.
Limitations
CostToKeep cannot predict exactly when an appliance or component will fail. Manufacturing revisions, installation, usage, maintenance, environment, repair history, random component failure, local labor conditions, future parts pricing, and electricity prices can all affect real outcomes.