Valuation Abandonment: When Metrics Didn't Apply
Why Did Investors Stop Using Traditional Valuation Metrics?
In 1999 and early 2000, research analysts, portfolio managers, and financial commentators routinely argued that traditional valuation metrics — price-to-earnings ratios, price-to-book, discounted cash flow analysis — did not apply to internet companies. The "new economy" narrative held that the internet was so transformative, and internet companies' growth trajectories so steep, that the established tools of investment analysis were inadequate to the task of pricing these securities. History records that the opposite was true: traditional valuation metrics did apply, which is why the stocks that violated them lost 90-100% of their value.
Quick definition: Valuation abandonment refers to the systematic replacement of earnings-based valuation frameworks with metrics designed to justify existing high prices — including price-to-sales ratios in the hundreds, user count multiples, and "eyeball" metrics that valued web page views — during the dot-com mania.
Key Takeaways
- When a company has no earnings, traditional P/E analysis cannot be applied directly; but discounted cash flow analysis, which projects future earnings, can still be applied and will reveal unsustainable valuations if done honestly.
- The "new metrics" of the dot-com era — price per unique visitor, price per registered user, price per page view — were proxy metrics for user engagement that required unstated and often unjustifiable assumptions to connect to actual economic value.
- Price-to-sales ratios of 100-400x require net profit margins and terminal multiple assumptions that are impossible to satisfy simultaneously with any realistic business model.
- The narrative justification — that the internet's winner-take-all dynamics would eventually produce monopoly profits that would justify current prices — was correct for a small number of companies and wrong for hundreds of others.
- Aswath Damodaran's work on "story and numbers" in valuation provides the framework for understanding how narratives can legitimately influence valuation and when they are being used to override rather than inform financial analysis.
The Legitimate Challenge: Valuing Growth Companies
The challenge of valuing early-stage growth companies is real, and the dismissal of traditional metrics was not entirely without intellectual foundation. A company with no current earnings but genuinely enormous future earnings power will be undervalued by any metric that relies on current earnings. Standard P/E analysis of Amazon in 1997, applied literally, would produce either a division by zero (no earnings) or a negative P/E (losses). The correct response to this is not to abandon valuation but to use discounted cash flow analysis, which projects future earnings, discounts them at an appropriate rate, and derives a present value.
The honest application of DCF analysis to most dot-com companies in 1999 would have required assumptions about future revenue growth, operating margins, capital intensity, and terminal value that, when specified explicitly, would have produced present values far below trading prices. The problem was not that DCF was unavailable — it was that honest application produced inconvenient answers, and the financial ecosystem had strong incentives to find frameworks that produced convenient ones.
The substitution of "new metrics" for traditional analysis was the mechanism through which this occurred.
The New Metrics
Several alternative valuation frameworks circulated during the mania, each sharing the characteristic of measuring something real but requiring a further leap — often an unstated one — to connect to economic value.
Price per unique visitor measured the cost in market capitalization terms of each person who visited a website each month. If a site had 10 million unique monthly visitors and a market cap of $1 billion, the price per visitor was $100. Advocates argued that users who regularly visited a site were building a "relationship" with the brand that would eventually be monetized. The unstated assumption — that each visitor had an expected lifetime value of more than the price per visitor — required specific assumptions about conversion rates, advertising rates, and customer retention that were rarely specified.
Price per registered user applied a similar logic to users who had created accounts, reasoning that account holders had stronger relationships than anonymous visitors. The metric had some validity in contexts where registered users had genuine switching costs — email, for example. It was less valid for services without switching costs, where users would simply register with whichever service was currently preferred.
"Eyeballs" — referring to page views — measured the total number of times users viewed pages on a site. The advertising revenue potential of page views was real: display advertising did generate revenue proportional to views. But the multiple applied to page view counts required specific assumptions about advertising rates per thousand views (CPMs) that were unsustainable in a market that was adding capacity faster than advertising budgets.
Mary Meeker, the Morgan Stanley internet analyst whose reports were widely influential during the bubble, developed models that applied multiples to user metrics derived from comparisons to other internet companies — valuations that were circular when the reference companies were themselves overvalued.
The Winner-Take-All Narrative
The intellectual framework that provided the most defensible justification for extreme valuations was the "winner-take-all" network effects argument. This argument had genuine merit for specific types of businesses.
Network effects — the phenomenon by which a product becomes more valuable as more people use it — are real in businesses like telephone networks, social networks, and marketplaces. A telephone network becomes more valuable as more people connect; a marketplace becomes more valuable as more buyers and sellers participate. In a market with strong network effects, first-mover advantage can be permanent: once one company achieves sufficient scale, the advantage compounds and rivals cannot catch up.
The argument was valid for some internet businesses. eBay's marketplace had genuine network effects; sellers went where buyers were and buyers went where sellers were, and the resulting concentration was durable. Amazon's customer base, order history data, and review ecosystem created real switching costs that compounded over time.
But the winner-take-all argument was applied indiscriminately to businesses that lacked genuine network effects. Pet supply companies did not have network effects — a customer who ordered from Pets.com derived no benefit from the fact that other customers also ordered from Pets.com. Grocery delivery companies had limited network effects within a geographic area but not across geographies. Many business-to-business software companies had legitimate switching costs but were not winner-take-all markets.
The systematic overapplication of the network effects narrative to businesses that did not have network effects was one of the key mechanisms through which the valuation errors of the mania propagated.
Illustrative Valuation Arithmetic
The arithmetic of dot-com era valuations was not hidden. It was simply not performed. The following table illustrates the implicit assumptions embedded in various valuation multiples, using realistic assumptions about long-term net profit margins and terminal P/E ratios.
Price-to-Sales Multiple | Required Net Margin | Required Terminal P/E | Implied Revenue Growth
10x P/S | 10% | 20x P/E | Moderate
50x P/S | 10% | 25x P/E | Extremely high
100x P/S | 10% | 25x P/E | Implausible
400x P/S | 10% | 25x P/E | Mathematically impossible
At a 100x price-to-sales ratio, a company would need to generate 10% net margins and trade at 25x earnings to produce a price-to-earnings ratio of 250x — which itself would require further substantial growth to be justified. The arithmetic was not complicated; it was simply not performed.
The Valuation Progression
Common Mistakes When Analyzing Valuation Abandonment
Concluding that DCF is not applicable to growth companies. DCF with explicit growth assumptions is the correct tool for growth company valuation. The problem was not the tool — it was the unwillingness to make honest assumptions that produced inconvenient results.
Treating all P/S ratios as unjustifiable. High-quality software businesses with durable competitive advantages can legitimately trade at elevated price-to-sales multiples if their path to profitability is clear and the terminal margins are high. The problem was applying these multiples to businesses without those characteristics.
Assuming the new metrics were entirely worthless. User metrics and engagement data are real leading indicators of future revenue potential in businesses with genuine network effects and monetization paths. The problem was the lack of honest connection between the metric and the implicit valuation assumptions.
Ignoring the role of benchmarking to other overvalued stocks. A significant portion of dot-com valuation analysis consisted of comparing one overvalued company to other overvalued companies and concluding that the comparison company was cheap. When the entire reference set is overvalued, relative analysis produces results as unreliable as absolute analysis.
Frequently Asked Questions
Are there legitimate situations where traditional P/E analysis doesn't apply? Yes — early-stage companies, cyclical companies in cyclical troughs, and companies making accounting-required write-offs that distort reported earnings. In all these cases, the appropriate response is to use alternative valuation methods, not to abandon valuation entirely.
Was Aswath Damodaran right that every company has a value? Yes — Damodaran's insight is that every security has a fundamental value derived from its future cash flows and risk, even if that value is highly uncertain and difficult to estimate. The uncertainty of the estimate is not a reason to pay any price; it is a reason to apply a larger margin of safety.
Which investors avoided the valuation trap? Value-oriented investors — those with explicit earnings or cash flow requirements for investment — largely avoided the worst dot-com excess. Warren Buffett famously declined to invest in technology companies he couldn't value, and Berkshire Hathaway significantly underperformed technology-heavy benchmarks in 1999. Buffett was widely criticized at the time; he was vindicated in 2001 and beyond.
How quickly did investors revert to traditional metrics after the crash? Very quickly. By 2001, analyst reports on technology companies that lacked earnings were discussing revenue trajectories toward profitability, not user metrics. The institutional memory of the metrics' failure was rapid, even if the structural incentives that had produced them were only partially addressed by the 2003 settlement.
Did any companies justify their dot-com era valuations in retrospect? A small number. Amazon's peak 2000 market capitalization was approximately $30 billion; the company subsequently grew to be worth more than $1 trillion. Google's pre-IPO value was arguably underestimated. The survivors' eventual performance demonstrated that winner-take-all dynamics did materialize for a very small set of companies — just not for the hundreds to whom the narrative was applied.
Related Concepts
Summary
The systematic rejection of traditional valuation frameworks during the dot-com mania was not primarily a failure of investor intelligence — it was a response to incentive structures that made honest valuation inconvenient and alternative metrics financially rewarding. Price-per-user, price-per-eyeball, and price-to-sales multiples in the hundreds were not intrinsically unreliable metrics; they were metrics applied without the honest specification of the assumptions that connected them to underlying economic value. The winner-take-all narrative that justified extreme multiples had genuine validity for a small subset of internet businesses and was catastrophically misapplied to the majority. When the crash came, the reversion was to earnings-based reality — which had never actually stopped being the relevant anchor for equity value. The lesson is not that non-traditional metrics are always wrong but that any metric applied without honest connection to terminal earnings capacity becomes a vehicle for rationalizing whatever price the market is currently producing.