reserve price

A reserve price is the minimum bid you'll accept in an auction. Below that floor, you keep the item unsold - even if a Buyer is offering above your cost.

The naive approach: set your reserve at break-even. If it costs $20K to serve a customer, accept anything above $20K. This leaves money on the table - sometimes a lot of money.

Roger Myerson's optimal auction framework proves that the Revenue-maximizing reserve comes from a quantity called the virtual value. For a Buyer whose value is v:

Virtual value(v) = v - (1 - F(v)) / f(v)

where F(v) is the probability a random Buyer values the item at v or less, and f(v) measures how concentrated Buyers are near that value.

The intuition: to justify including a Buyer who values the item at v in your Allocation, you must concede surplus to every Buyer above v. Why? Any higher-value Buyer could claim to value the item at v and pocket the difference - that's their Informational Advantage. The term (1 - F(v)) / f(v) is the marginal cost in surrendered surplus of lowering your threshold to include one more Buyer type.

Virtual value is what a bid is actually worth to you after accounting for that cost. The optimal reserve is where virtual value crosses zero. Below that threshold, accepting the bid costs you Expected Revenue - even though the Buyer is paying real dollars.

Every time you sell something competitively - ad slots, inventory, enterprise contracts, even internal Allocation of scarce capacity - you're running an auction, whether you call it one or not.

Without a reserve, Buyers capture too much surplus:

• •Automated ad slot auctions without floor prices let Buyers extract most of the value your platform creates
• •Enterprise sales that accept any offer above cost systematically undercharge high-value Buyers
• •Competitive bidding processes without minimum price thresholds attract low-value bids that waste your time

The P&L impact is concrete. With 2-5 Buyers and moderate value dispersion, the right reserve lifts Revenue 15-30% versus no-reserve auctions. (With 10+ Buyers, competition does most of the work and the reserve rarely binds.) For a business running $2M/year through competitive selling with a thin Buyer pool, that's $300K-$600K from a single decision rule change.

The counterintuitive part: sometimes the optimal move is no sale. Setting a reserve means some auctions end with no winner. An empty ad slot, an unsigned contract - it feels like waste. But the Expected Value math accounts for this. The Revenue you gain from higher prices on successful sales more than compensates for the deals you lose.

Step 1: Estimate the Buyer value distribution

You need a model of how Buyers value what you're selling. Sources include historical bid data, Pipeline Close Rates at different price points, and Conjoint Analysis.

Two common shapes:

• •Uniform [0, H]: Buyer values spread evenly from zero to some ceiling H. Reasonable when you have little data or the item is unfamiliar to Buyers.
• •Exponential with mean μ: Many low-value Buyers, a few high-value ones with a long tail. Common in ad slots and enterprise sales where most Buyers cluster at modest willingness-to-pay.

Step 2: Compute virtual values and find the reserve

For uniform values on [0, H]:

• •Virtual value = 2v - H
• •Reserve (zero cost) = H / 2

For exponential values with mean μ:

• •Virtual value = v - μ
• •Reserve (zero cost) = μ

The uniform result is striking: if Buyer values range from $0 to $1,000, your floor should be $500 - the midpoint - regardless of your cost.

Step 3: Incorporate your costs

If delivering costs you c, the reserve shifts up. Instead of setting virtual value = 0, set virtual value = c:

• •Uniform [0, H] with cost c: reserve = (H + c) / 2
• •Exponential mean μ with cost c: reserve = μ + c

Notice the reserve is always above your cost. The gap between reserve and cost is the correction for Buyer Informational Advantage.

Step 4: Apply the reserve

In an auction where the winner pays the second-highest bid, announce the reserve up front. The winner pays whichever is greater: the reserve or the second-highest bid. This format incentivizes Buyers to bid their true value, because their bid only determines whether they win - not what they pay. When you're Pricing to one Buyer at a time (no competing bids), the reserve is your price.

Step 5: Account for estimation error

The entire framework assumes you know F(v). In practice, you're estimating it from noisy Pipeline data. The reserve is asymmetrically sensitive to that estimate: setting the floor too high costs you entire deals (the Buyer walks away and you get zero), while setting it too low costs you only marginal surplus on deals that would have closed anyway. When uncertain, err slightly below the formula output.

Key insight: the reserve doesn't depend on the number of Buyers. Whether 2 or 200 Buyers show up, the optimal floor is the same. More competition raises the winning price through bidding - it doesn't change where you draw the line.

Apply Myerson reserves when all four conditions hold:

1. 1)You control the rules. You set the auction format, the minimum prices, the Allocation criteria. If you're the one placing a bid (Buyer side), this framework tells you what the seller should be doing - useful for anticipating their strategy.

2. 2)Buyers have private values. Each Buyer knows what they'd pay, but you can't observe it directly. This Informational Advantage is the entire reason virtual values exist - they correct for it.

3. 3)You can estimate the distribution. You need enough historical data - past bids, Pipeline Close Rates at different price points, or market research - to model the range and shape of Buyer values. Without distribution data, you're guessing.

4. 4)Walking away is an option. If you must sell today (inventory approaching Obsolescence, contractual obligations, Cash Flow emergencies), the reserve drops - possibly to zero. The value of a reserve comes from credibly refusing bad deals.

Where this shows up for Operators:

• •Setting floor prices on ad slots (the single most common application)
• •Pricing enterprise tiers when you negotiate one deal at a time
• •Running internal Capital Allocation with Hurdle Rates (the Hurdle Rate IS a reserve price on your capital)
• •Liquidating end-of-life inventory via auction

The entire reserve calculation rests on Expected Value - the reserve maximizes expected Revenue across all possible Buyer outcomes, which sometimes means accepting no sale at all. Reserves connect directly to Pricing: the optimal posted price for a single Buyer IS the Myerson reserve, making this the theoretical foundation for any Pricing decision where the Buyer knows more about their willingness-to-pay than you do. The concept maps cleanly to Hurdle Rate in Capital Allocation - the minimum acceptable Return on deployed capital is the same mechanism applied to a different Asset. In both cases, you'd rather hold the resource idle than deploy it below threshold, because accepting low-value commitments crowds out the Expected Value of future opportunities.