When 11x raised $50 million at a $350 million valuation from venture capital firms Andreessen Horowitz and Benchmark, the US-based AI startup proudly touted nearly $10 million in annual recurring revenue (ARR). 

But months later, investigations revealed that the actual figure was closer to $3 million. The company, critics alleged, had been counting short-term trial customers of three months as if they were locked into full-year contracts. 

While 11x disputed the findings of the investigation, this fiasco brought to light a deeper problem plaguing startup finance. 

ARR, once a straightforward measure of contractual commitments, has morphed into an ambiguous and at times manipulated metric in venture capital. 

As AI companies race to demonstrate explosive growth, the gap between reported ARR and business reality has never been wider. 

In this scenario, founders often lean on extrapolation, the villain in this story, taking a single month’s revenue and multiplying it by 12. 

True ARR represents contractual certainty—customers who have committed to paying for a whole year or more. Annualised revenue or annual run rate, by contrast, is mere mathematical wishful thinking, assuming that one month’s performance will repeat itself 12 times. 

When ARR is Meaningful and When It Isn’t 

Consider the mathematics: a startup that earns $100,000 in July can legitimately claim $1.2 million in annualised revenue. 

“As an investor, when we see someone say, ‘I have annualized $1.2 million of revenue’, versus a founder that says ‘I have $1.2 million in ARR’, I know immediately that the founder with $1.2 million annualised revenue does not have a 12-month contract,” Jay Krishnan, head of investments of India at Beyond Next Ventures, said in an interview with AIM. 

“The likelihood that an early founder has an ARR is reasonably slim,” he added. 

The confusion is not limited to terminology. Joseph Johnson, founder of LedgerUp, told AIM, “Folks often include pilots or one-time fees in ARR, not because they mean to mislead investors, but just because they’re confused on what ARR is. It’s also challenging if you have a heavy usage-based business to share ARR, and you typically go with run rate.” 

Besides, even when used correctly, ARR may fail to paint the whole picture. A startup could boast seemingly high recurring revenue, yet still operate at a supremely high net loss if acquiring a single customer incurs a high cost.

Read More: AI Startups Depend on Costly APIs of Companies Burning Billions

Complicating matters further, founders often interpret ARR in their own way. “Founders want to calculate ARR in lots of different ways, and everyone treats it slightly uniquely,” Johnson said. Founders, especially in the software-as-a-service (SaaS) sector, face a fair bit of mathematical challenge while calculating ARR. 

In the traditional mathematical sense, monthly recurring revenue (MRR) is simply multiplied by 12 to obtain ARR. 

“Let’s say you are a client of mine. I signed up with you in August. Tomorrow, in September, I will sign up with another customer. In October, I signed up with another customer. Each of these contracts has an MRR value, but they are staggered over time. Therefore, calculating ARR for a particular company becomes difficult. You can’t just sum up all three clients. That’s one challenge,” Krishnan explained. 

Besides, when customers finish their monthly contractual obligation and do not resume further, MRR simply multiplied by 12 cannot lead to ARR. 

Moreover, AI startups are using tokens, compute time, number of queries and more ways to bill customers. “And these tend to go up or down based on utilisation, because even the clients are trying to figure out how to best use AI, “ he added. 

Over the last few months, ARR has taken the stage in some of the biggest startup growth stories. Yet, when calculated correctly, ARR is a good way to see the predictability and legacy of both customers and business, Krishnan argues. This is probably why it remains one of the most widely used metrics. 

Besides, for a startup that is only two months old and speaks of an annualised revenue, say 1.2 million, it still matters. “Because in two months, he or she is able to crack 100k of revenue, and therefore, they’ve just used the word annualised to project forward. So it has to be taken in context,” he further said. 

If Not ARR, Then What? 

For investors seeking genuine signals amid the ARR noise, Krishnan advocates for a more comprehensive approach to evaluating startup health. Rather than fixating on revenue metrics alone, he emphasises looking at multiple characteristics that reveal the true trajectory of a business. 

He recommends examining the relationship between ARR and MRR. “ARR divided by MRR gives you an indication of growth,” he said. Yet, tracking the progression from bookings to billing to revenue is equally important. “If startups have a business development team that has done bookings, the finance team has done collection, and then there’s revenue. If that demonstrates a funnel, that means there’s growth.”

Customer retention, however, may be the single most telling metric. “Net revenue retention is a great metric because it shows upsell opportunity, which means that the customer is using the product and wants to use it more,” Johnson emphasised.  He warns against superficial enterprise deals: “It’s dangerous when someone signs a large enterprise that has an AI budget but does not actually use the product enough.”

Krishnan points to gross revenue retention and net revenue retention as essential metrics, but reserves special attention for churn, which refers to the rate at which customers stop doing business with an entity.

“The golden metric for SaaS models for retention is churn, measured in percentages…The lower the churn, the better.” This focus on keeping customers reveals far more about product-market fit than inflated acquisition numbers.

For measuring customer engagement, he suggests examining daily-to-monthly active user ratios, feature adherence rates, and time spent in the product. “The more you launch features and the more the customer adheres to it, as opposed to launching features but nobody using them, that’s a good way to measure engagement.”

Besides, the efficiency metrics that matter most revolve around unit economics. Krishnan highlights the “golden ratio” of lifetime value to customer acquisition cost at 3:1.

“That gives you an indication of how much money you need to raise or at least a percentage of the money that you raise goes towards demonstrating that your efficiency gets to a point where LTV to cap will flip from one or two to eventually get to three,” he added. 

This measurement problem requires discipline from founders and investors, including clear definitions, honest reporting and a focus on business fundamentals that ARR signifies. Only then can it accurately measure genuine, sustainable growth instead of hiding accounting tricks.

The post The ARR Mirage appeared first on Analytics India Magazine.