By the Monetizely Team
In Greek mythology, Atlas was condemned to carry the sky on his shoulders after the war of the Titans—an eternal burden. Over centuries, the image has morphed into Atlas carrying the entire world, straining under a weight he can’t put down. At SaaS Metrics Palooza 2024, Dave Kellogg painted a striking parallel between ARR (Annual Recurring Revenue) and Atlas: for years, ARR has shouldered the entire world of SaaS metrics, but the strain is becoming overwhelming—especially with AI piling on new demands.
Below is our recap and interpretation of Dave’s talk, infused with a modern twist on an ancient myth, showing how AI is triggering a mini “Titanomachy” in the realm of SaaS.
1. Atlas’ Role: ARR as the Bedrock of SaaS Metrics
When SaaS first emerged, ARR was like Atlas: strong, stable, and capable of carrying the “heavens” of all key performance metrics—churn, CAC, NRR, and more. The assumption: “If the ARR figure is correct, everything else follows.”
Yet, just as the weight of the sky never lessened for the mythic Titan, the growing complexity of contracts (multi-year deals, usage-based components, expansions) has started making ARR buckle. Month-to-month fluctuations—mean ARR cannot always be the simple, stable pillar it used to be.
The Cracks in ARR’s Shoulders
- Longer, Prepaid Contracts: Multi-year deals can deliver large sums upfront—do you recognize it as “3x ARR” or keep it to a single year’s revenue?
- Usage & Overages: If usage spikes, you collect more revenue—but is that “recurring” or an unpredictable add-on? Conversely, if you have a usage lull, what is the impact on ARR?
2. The Titanomachy of AI: Why the Weight Is Getting Heavier
In ancient tales, Titanomachy was a cosmic clash—Titans versus Olympians. In modern SaaS, AI is the new force colliding with traditional subscription revenue:
- Cost Surges: Building and running AI models devours colossal resources (GPU clusters, data centers). This can slash margins, forcing companies to adjust pricing—or risk toppling under the weight of cost. Per Dave, cogs for traditional SaaS were around 20%, but the COGS for the foundation model companies are anywhere from 50-75%. This is quite insane!
- Unpredictable Usage: AI usage can soar unpredictably as customers embrace new features or scale them globally. That means monthly bills can veer wildly from forecasts, eroding the very notion of “annual” recurring.
AI’s emergence upends the old hierarchy. The once all-powerful Titan—ARR—suddenly finds itself battling a new reality of consumption, cost, and complexity.
3. New Pricing Drivers: Rewriting the Script of the Gods
Consumption-Based pricing is like forging new cosmic laws. Instead of a flat fee or per seat, AI solutions can charge by tasks completed, problems solved, or hours saved.
- From Seats to Usage or Success Metrics: Imagine an AI co-pilot for sales: the more deals it helps close, the higher the fee. Or an AI help-desk agent priced per resolved conversation. It’s a big/complex shift. Products that provide image generation can charge by the number of images generated. AI voice products can charge by the minutes of AI audio produced.
- Hybrid or Cost-Plus: Many companies still want a “base subscription” for stability (Atlas’s original role) but stack on usage or outcome-based fees—like new layers of sky that can expand or contract.
4. The Future of Metrics
Despite the upheaval, Dave Kellogg pointed out that core SaaS metrics—churn, CAC, and net retention—remain indispensable. But the way we calculate them, has to adapt:
- Trailing Spend: In a variable world, monthly (or trailing 12-month) spend can feel more “real” than a neatly annualised figure.
- Implied ARR: Public markets still like to see a big, bold “ARR” number. But it’s increasingly an approximation—a single data point pinned onto a complex constellation of usage patterns.
- Net Retention: Instead of this being calculated y-o-y annual ARR, this becomes the delta between trailing spend over two years. Snowflake calculates this to be trailing one-year spend divided by trailing year-before-that spend for customers who started on or before the first month of the year-before-that period. Dave in this blog, also mentions how Snowflake, Datadog, Twilio, and a few other companies have their own unique way of calculating NRR. Herein lies one of the problems of standardization (or non) in the new world of usage based pricing.
- Net Expansion: This becomes a comparison of the delta in spend between Q2 of last FY and current FY
Monetizely’s Take
Now, advanced pricing models are taking on a bigger share of that weight. This will put a burden across the company from engineering new meters, to sales adjusting their comp plans, to sales leaders leaning more on mathematical forecasting models to finance team crafting company specific metric and/or preparing metrics with different calculations to look at performance with different lenses. It requires care and thoughtful roll out across a company.
Conclusion: Beyond Titanomachy—A New Divine Order in SaaS?
In Greek legend, Atlas never escaped his burden. He remained forever stooped beneath the sky. In SaaS, ARR might still carry a significant share of metrics, but usage-based variability and high AI costs mean the load is changing shape. The question is whether we’ll see ARR slump under the growing cosmic weight or stand tall with the help of new pricing and billing strategies.
At Monetizely, we believe that a lot of things will need to be relooked at:
- Seat Based Is Dead: Usage based is killing seat based due to COGS. No going back on that. A new way of doing business needs to be devised.
- Metrics Will Need Consistency: Consistency in metrics is still important, eventually companies will have to converge on new definitions of NRR, Net Expansion, etc. because they will need to be valued consistently as well.
- COGS are Queen: COGS will take a new front-seat. Software historically has been mostly value-based due to low COGS, but this is no longer the case. This makes the math harder, because customers are going to be getting bigger bills and while they may complain, this time the price increases will not be about capturing more value but about covering costs.
- Uncertain Valuations: Valuations are going to have to be reset. If the new AI companies have lower margins, then will they have greater or lower multiples/valuations than older SaaS companies?
- More Plumbing: New software will support the addition of meters, billing, financial reporting – i.e. all the plumbing that makes usage based companies work.
- Put A Ring On It: Executives will realize that pure usage (just pay as you go) with no commitments is a recipe for disaster. This does not provide the predictability that they, their customers or their investors seek. Minimum commits or platform fees or 3-part tariffs will be the order of the day.
Wanna explore more? Book your free 1:1 call with us today!
.png)
.png)