There is no denying that there is an ongoing change in the startup world, where AI-based startups have challenged some of the time-tested metrics used for evaluating a business’s performance. There is a growing debate on the importance of one such metric called the Annual Recurring Revenue (ARR). Once seen as the benchmark for growth in the SaaS space, the ARR is now being questioned by investors and founders in the current climate where AI startups are growing at breakneck speed.
The Debate on ARR Heats Up
This discussion took another leap when a young AI startup called Emergent AI claimed that its ARR had crossed the $100 million mark in just eight months of operation. The rapid increase in its ARR from $15 million to $50 million in a matter of months and then again doubling in a short while has brought attention to this emerging company.
These figures have started raising a question: Is this the true reflection of their financial health or just the effect of a different kind of pricing model that does not conform to the usual SaaS standards?
Shift from Subscription to Usage-Based Revenue
Traditionally, ARR has been based on subscription-based revenues. Businesses have relied on long-term contracts, which have ranged between 12 and 36 months, providing consistent revenue streams that make the calculation of ARR an accurate measure when assessing investments in a business.
AI-native companies work differently since they do not rely on traditional subscription models for revenue generation. Instead, these companies generate revenue from usage-based models. In such cases, the users pay according to their level of usage, which is calculated in terms of tokens—the smallest unit of text that an AI system can process.
In this case, revenues become inconsistent because the revenue generated increases with the increased level of usage; at the same time, so does the cost of providing the AI services.
The Cost Conundrum: Tokens & Margins
Usage and profitability are perhaps the two key things when it comes to business models in the AI space. And usage drives token consumption, which means increased expenses as well.
For example, premium AI products tend to offer top-tier quality but at a cost of high computation expenses. This means that even when such AI companies manage to boost their revenues, they don’t actually become profitable.
And investors are very cautious about investments into startups showing great numbers on ARR but failing to explain what is happening with the other key metrics.
As one expert puts it, “Token consumption is activity, but not value.”
Why ARR May Not Be Valuable Anymore
The primary problem with ARR lies in the fact that this metric doesn’t help us to evaluate value creation and sustainability of operations.
While ARR does allow us to see annual revenue that a business manages to generate, it does not reveal anything about:
1. Profitability
2. User retention rate
3. Costs involved in operating an AI product
4. Benefits received by users
In case of classic SaaS products, high ARR implied positive fundamentals. But such is not the case with many AI companies.
New Metrics in the Making
With the realization of the inadequacies of ARR, investors have been searching for new methods of evaluating AI startups. Some of the most promising new metrics include:
1. Revenue Based On Real-Time Usage Trends
Calculating revenue through usage in real-time rather than annually.
2. Cohort Retention
Determining the level of customer commitment in terms of their continued use of the product.
3. Expansion Revenue
Calculating additional revenue from current customers.
4. Time-to-Value
determining the time required to realize the value of using the product.
5. Gross Margin
Deducting the cost of computations and infrastructure from the total margin.
The above metrics give a comprehensive assessment of the progress of a startup in a more accurate way.
The Investor’s Angle: A Balanced Perspective
However, this is not to say that ARR has totally lost its relevance in the world of finance. Indeed, many technology startups such as OpenAI and Databricks emphasize ARR as a growth metric.
But now there are attempts to adopt a holistic approach to evaluating investments. This means that investors not only take into account ARR but combine it with other indicators to get more information about the business.
As a venture capitalist once said, a highly profitable AI enterprise with a high ARR but low margins is completely different from a regular software-as-a-service company that generates good profits.
Pressure on Startups to Show Fast Growth
Another problem that arises because of the emphasis on ARR is the necessity of fast growth. To attract investments, startups strive to show high revenues in their reports.
The Emergent AI episode is a warning for those who tend to distort the reality of business processes when providing their results.
The Road Ahead: Defining Success in AI Startups
With the continued influence of AI technology, it is essential for the metrics used in evaluating startups to adapt to the changing realities. Thus, with regard to shifting away from the subscription model, startups have to consider the following aspects:
Dynamic pricing schemes
High computational cost
Value realization
These factors mean that the measure of success for AI-based businesses cannot be simply determined by the amount of revenue generated.
Conclusion
It is clear that the diminishing role of ARR will represent a paradigm shift for investors and founders alike. Although there is no doubt that ARR will continue being an important measure of success, it will no longer serve as a reliable metric in terms of AI-based businesses.
Thus, investors will need to pay attention to more sophisticated metrics, including quality of revenue and profitability. In the years to come, it can be expected that a new set of metrics will emerge, reflecting the true nature of success in artificial intelligence.
