insurance (how much to charge to cover average claims?)
Your e-commerce platform sells $200 headphones, and customers keep asking for a damage protection plan. You need to set a price - $15? $25? $40? Charge too little and every broken pair eats your Profit. Charge too much and nobody buys it. The answer is hiding in your historical claims data.
Claims are the payouts an insurance or protection program must cover. Use Expected Value to calculate average claims per customer, then price your Base Fee above that number to build a sustainable financial product.
A claim is a request for payment when a covered event happens - a product breaks, a shipment is lost, or a service fails. From the operator's perspective, claims are the core liability of any insurance or protection program you run.
Every time you offer a protection plan or a service guarantee, you are creating a financial product. Customers pay a Base Fee upfront. In return, you promise to pay their claims later. The fundamental question is: will the Base Fees you collect cover the claims you pay out?
This is an Expected Value problem. If you know the probability of a claim and the typical dollar amount, you can calculate what each customer costs you on average - and price accordingly.
Claims hit your P&L in two ways:
If you are running a protection plan as a Revenue line, mispricing claims is a failure mode that can quietly drain Profit for months before anyone notices. The gap between what you charge (Base Fee) and what you pay out (claims) is your Unit Economics for that financial product.
Operators who understand claims pricing can also make smarter decisions about whether to buy insurance from a third party or cover claims out of their own Budget. The math is the same either way: Expected Value of claims vs. the price of coverage.
Step 1: Estimate Expected Value of claims per customer.
You need two numbers:
Expected Value per customer = probability × average claim size
E[claims] = 0.08 × $180 = $14.40
This means across many customers, you will pay about $14.40 per enrolled customer on average.
Step 2: Price your Base Fee above Expected Value.
If claims cost you $14.40 on average, charging $14.40 hits break-even before you account for operating costs (Labor to process claims, shipping, etc.). You need a buffer for Profit.
A typical approach: Base Fee = Expected Value of claims + operating costs + target Profit.
If processing costs $3 per customer and you want $5 Profit:
Base Fee = $14.40 + $3.00 + $5.00 = $22.40
Step 3: Account for Variance.
Expected Value tells you the average. But in any given month, actual claims could be much higher or lower. If you have 100 customers, expected total claims are $1,440 - but you might see $2,500 in a bad month.
The fewer customers you have, the more Variance matters. With 10,000 customers, the averages smooth out nicely. With 50 customers, a few bad months can wipe out your Cash Flow. This is where Tail Risk becomes real - rare events that cost far more than your average.
Price a protection plan or service guarantee: Any time you offer customers coverage against breakage, loss, or failure, you are in the claims business. Calculate Expected Value before you set prices.
Decide between buying insurance vs. covering claims yourself: If you run a fleet of delivery vehicles, you can buy a policy from an insurer or set aside Budget to cover incidents yourself. Compare the insurer's Base Fee to your Expected Value of claims. If their fee is much higher than your expected claims, self-funding may be cheaper - but only if you can absorb the Variance.
Evaluate whether a protection program is profitable: Pull your historical claims data. Divide total claims paid by total customers enrolled. If that number is close to or above your Base Fee, your Unit Economics are broken.
Budget for Contingent Liabilities: Even if you are not running a formal insurance financial product, any money-back guarantee or service-level promise creates implicit claims exposure. Budget for it using Expected Value.
You sell laptops averaging $900. You want to offer a 2-year protection plan. Historical data from 5,000 past customers shows: 12% file a claim in year 1, 18% file a claim in year 2 (components age). Average claim payout is $340 (mix of repairs and replacements). Processing each claim costs $25 in Labor.
Calculate claim probability over the 2-year period: Not simply 12% + 18% = 30%, because some customers claim in both years. Using your data, 26% of customers file at least one claim over 2 years.
Expected Value of claims per customer: 0.26 × $340 = $88.40
Expected processing cost per customer: 0.26 × $25 = $6.50
Expected Total Cost per customer: $88.40 + $6.50 = $94.90
Set Base Fee with target Profit: $94.90 / 0.70 = $135.57 (targeting 30% of Revenue as Profit). Round to $139.
Sanity check Unit Economics: For every 100 plans sold at $139, you collect $13,900. Expected claims: $8,840. Expected processing: $650. Expected Profit: $4,410.
Insight: The Base Fee ($139) is about 15% of the laptop price ($900). Customers perceive this as reasonable because they compare it to the full replacement cost, not to the Expected Value of their claim. This gap between perceived value and expected cost is where Profit lives in the claims business.
Your company operates 20 delivery vans. An insurer quotes you $4,800 per van per year ($96,000 total). Your incident records over 3 years show: average of 6 incidents per year across the fleet, with an average cost of $5,200 per incident.
Expected Value of annual claims: 6 incidents × $5,200 = $31,200 per year
Insurer's quote: $96,000 per year
Difference: $96,000 - $31,200 = $64,800. You would be paying more than 3× your Expected Value.
But check the Variance: your worst year had 11 incidents ($57,200). Your best year had 3 ($15,600). The range is wide with only 20 vans.
Tail Risk check: one serious incident could cost $80,000+. With 20 vans, one catastrophic year could exceed $100,000.
Decision: if your risk appetite can handle a $100,000 bad year and you have the Cash Flow to absorb it, covering your own claims saves ~$64,800 in an average year. If a $100,000 hit would break your Budget, pay the $96,000 for certainty.
Insight: Insurance companies charge well above Expected Value - that is how they make Profit. The question for an Operator is whether the Variance and Tail Risk are large enough relative to your Cash Flow that paying that spread is worth the certainty. Businesses with more capacity to absorb losses can capture the difference.
Claims are the payout side of any insurance or protection financial product. Your Base Fee must exceed the Expected Value of claims plus operating costs, or you lose money on every customer.
Expected Value gives you the average, but Variance determines whether you survive month-to-month. Fewer customers means more Variance, which means you need a bigger buffer or more Cash Flow to absorb bad months.
The decision to cover your own claims vs. buy a policy comes down to comparing the insurer's Base Fee to your Expected Value of claims, then asking whether your risk appetite and Cash Flow can handle the Tail Risk.
Pricing a protection plan based on gut feel instead of calculating Expected Value from historical data. 'Charge $29 because it sounds reasonable' is how operators lose money for months without realizing it.
Ignoring Variance when the customer base is small. Expected Value works great with 10,000 customers. With 200 customers, a bad month can cost 3-4× the expected amount, and you need enough Cash Flow to survive it.
You run a SaaS platform and offer a 99.9% uptime guarantee: customers get a full month refund for any month you miss the target. You have 400 customers paying $250/month. Historical data shows you miss the target about 2 months per year, and when you do, roughly 60% of customers are affected. What is the Expected Value of your annual claims from this guarantee? What does this tell you about your effective Revenue?
Hint: Calculate the expected number of refund events per year, then the expected dollar amount per event. Remember each affected customer gets $250 back.
Claim events per year: 2 months of downtime. Affected customers per event: 400 × 0.60 = 240. Cost per event: 240 × $250 = $60,000. Annual Expected Value of claims: 2 × $60,000 = $120,000. Your gross annual Revenue is 400 × $250 × 12 = $1,200,000. Expected claims eat 10% of that, making your effective Revenue closer to $1,080,000. If you built a Budget around the full $1.2M, you have a $120,000 hole in your P&L.
A competitor offers an identical laptop protection plan to yours (from the worked example) but charges $89 - below your calculated Expected Total Cost of $94.90. Assuming their claim rates and costs are similar to yours, what are three possible explanations for how they can do this without losing money?
Hint: Think about what could differ in their Cost Structure, scale, or business model that changes the Expected Value calculation.
1) Lower claim probability: They might use stricter Underwriting criteria (e.g., excluding certain damage types), reducing their effective claim rate below 26%. 2) Lower average claim cost: They might repair more and replace less, or have better pricing on parts through Vendor Negotiations, cutting average payout below $340. 3) Scale advantage on Variance: With 500,000 customers vs. your 5,000, their actual claims track Expected Value much more closely, so they need less of a buffer above break-even. They can price closer to Expected Total Cost and still have predictable Cash Flow. Any of these could make $89 viable.
You are deciding whether to cover your 50-person company's dental benefits yourself or buy a group policy at $120/employee/month. Industry data suggests the Expected Value of dental claims is about $85/employee/month. But you have never tracked claims for your specific team. Should you cover the claims yourself in year one? Why or why not?
Hint: Consider what you know vs. what you do not know. Industry averages might not match your specific workforce. What is the cost of being wrong?
You probably should NOT cover claims yourself in year one. The $35/employee/month spread ($120 - $85 = $35, or $21,000/year for 50 people) looks attractive, but you are using industry-average data, not your own claims history. Your actual workforce could have very different claim patterns. With only 50 employees, Variance is high - a few expensive procedures could blow past your Budget. The smarter move: buy the policy in year one ($72,000), but track every claim carefully. After 12 months of real data, recalculate your company-specific Expected Value and Variance. If the data confirms claims run well below $120/employee/month and your Cash Flow can handle a bad quarter, switch to covering claims yourself in year two. You are spending $21,000 in year one to buy information - and the Value of Information here is high because the downside of being wrong is uncapped Contingent Liabilities with no historical basis for estimation.
Claims pricing is a direct application of Expected Value - your prerequisite concept. Where Expected Value taught you to collapse uncertain outcomes into a single number, claims shows you how to price against that number to build a sustainable financial product. The key extension is that Expected Value alone is not enough: you also need to understand Variance and Tail Risk to decide how much buffer to build in above break-even. Downstream, claims connects to Underwriting (the process of evaluating which risks to accept and at what price), Contingent Liabilities (claims obligations that sit on your Balance Sheet as potential future payouts), and Risk Tolerance (how much Variance your business can absorb before it threatens Cash Flow). If you are running a P&L with any kind of guarantee or protection offering, claims math is how you keep that line item from quietly eating your Profit.
Disclaimer: This content is for educational and informational purposes only and does not constitute financial, investment, tax, or legal advice. It is not a recommendation to buy, sell, or hold any security or financial product. You should consult a qualified financial advisor, tax professional, or attorney before making financial decisions. Past performance is not indicative of future results. The author is not a registered investment advisor, broker-dealer, or financial planner.