The AI wave presents both opportunities and potential pitfalls, leaving investors urgently seeking a more reliable compass for navigation. Currently, AI is profoundly transforming the investment logic within the software industry. To filter out genuine "AI winners" from numerous companies, US capital markets are focusing on a key metric — Net Revenue Retention (NRR). This is not merely a financial figure, but is viewed as a new method for evaluating a company's organic growth momentum and customer stickiness, particularly in assessing market acceptance of AI products. However, this metric lacks unified standards, with companies employing varied calculation methods that make horizontal comparisons nearly impossible and leave room for data "beautification."

**Investors' "New Favorite": Why is NRR So Important?**

NRR, also known as Net Dollar Retention, fundamentally measures a company's ability to continuously generate revenue from existing customers. An NRR exceeding 100% intuitively demonstrates that existing customers are not only retained but are also increasing their spending. According to software equity analyst Rishi Jaluria from RBC Capital Markets, this organic growth is crucial. "If customers discover tremendous value in your product, they're willing to invest more year after year," he notes, pointing out that compared to the increasingly expensive new customer acquisition market, enhancing existing customers' "wallet share" represents a more efficient and sustainable growth model. Therefore, high NRR is often regarded as strong evidence of successful AI product strategies.

**Beware of Number Games: "Honor Student" Figma's Financial Report Magic**

Naturally, this new approach is not without flaws. Investors must maintain clarity when using NRR. Research reveals that this metric currently lacks unified industry calculation standards. For instance, IPO documents show that software company Figma disclosed an impressive 132% NRR in its IPO filing, but this calculation scope was limited to customers with annual contract values exceeding $10,000, excluding numerous smaller clients. While the company claims this approach is more meaningful, DA Davidson analyst Gil Luria directly criticized the "controversial calculation method" — the company selected only 11,000 customers with annual repurchase amounts over $10,000 (representing 2.4% of total customers) as the sample, despite their contribution of 64% of revenue. The exclusion of 87.6% of customers renders horizontal comparisons meaningless.

Other companies employ various "techniques" as well. Software development tools company GitLab initially published a precise 148% NRR during its IPO period, but subsequent quarterly reports changed to only disclosing whether NRR fell below the 130% threshold, creating information blind spots for investors. Cybersecurity company Rubrik uses the average NRR over the most recent four quarters, refusing to disclose single-quarter data; data management company Snowflake extended the NRR calculation period from the standard 12 months to two years.

**From Horizontal Comparison to Vertical Tracking: The Correct "Approach" to NRR**

Despite calculation method differences, when software sector stocks are generally under pressure due to AI disruption, NRR is still viewed as an important screening tool. Gil Luria believes investors should shift focus from comparing absolute NRR values across different companies to tracking quarterly trend changes within individual companies. "Now, if you focus on companies whose NRR this quarter is higher than last quarter, that's an excellent clue for determining who's performing better."

Gil Luria cites Snowflake and Datadog as examples, noting that these companies' NRR metrics have rebounded from recent lows, powerfully demonstrating that their AI products are receiving positive market reception. This dynamic, vertical observation method provides investors with a new tool to filter out market noise and discover genuine value.

Therefore, when applying this new NRR methodology, understanding the underlying calculation logic and combining it with other financial data for comprehensive judgment represents the wise approach.