Why “all dollars are temporary” and the mathematical secret to scaling SaaS, with Eliya Elon
Imagine you take out three one-dollar bills. You smear one with a red gob of paint, another with a blue gob, and a third with a yellow gob, and then you slip them into circulation in your business and watch how they travel. My interviewee today assures us that you’d probably be surprised at where they go—and how they evolve along the way.
Take this real example—one depreciates to $0.90, one stays the same, and the third sextuples to $6.65. There’s a temporal and fungible aspect to all dollars, Eliya Elon explains. Once you realize and account for it, everything about your business systems thinking will change.
My interviewee today of course joins us from IntSights where he’s the Global Head of Business Operations and Analytics. His journey into business systems started in his teens as an Israeli CrossFit entrepreneur and led him to his role today. There, he “thrills in the application of systems thinking and philosophy to real-world challenges.”
We discuss why everyone should stop saying “tech stack,” why you should draw more diagrams, and why a temporal understanding of dollars can point the way to long-term customer value.
This conversation has been lightly edited for brevity.
Salto: Can you tell us a bit about what you do?
Eliya Elon: I help B2B companies think about their revenue and data “machine(s)” holistically. I’m passionate about uniting a broad, strategic view with an operator’s mindset, and seeing where they agree and what are the unique levers of the business for scale.
What’s one thing you enjoy about it?
Honestly, being a builder, and solving complex challenges that involve people, tech, and process. I find my best solutions while peeking into other industries and domains to look for things to borrow. Just consider how B2B is only just now adopting things like cohort analysis from ecommerce and B2C. Whenever I ask leaders in B2B about their conversion rates and sales cycles, they tend to give me one answer—a broad average. But of course, this thinking, while appealing, is flawed.
Anybody in ecommerce or social media knows the truth is hidden in cohorts. It’s not hidden in one global figure. If you want to operate a business at the highest level you have to get into the details like that.
Where can readers apply that sort of thinking?
I’d say there is nowhere more fun to merge theory and practice than in compensation plans. It combines so many exciting areas: psychology, economics, finance, mathematics. Fixing the comp plan is always the CFO’s first or second goal. They’re always asking, “What’s the industry benchmark?” And here’s the thing: Your company is far too unique for you to be modeling yourself after the average of thousands of other companies who share little in common.
Instead, you want to zoom in on the jagged parts of that graph. In what areas do you outperform compared to your competition? That’s where you differentiate. There’s this saying which I think sums this up nicely: Strategy lives in the variance from the mean.
So let’s apply this idea. Let’s say you excel in one area. Maybe you have ridiculously amazing retention rates. Maybe that’s a feature of your team or product—maybe your service is really difficult to migrate off of. You know that no matter what you do, your retention rates will remain naturally high.
So where do you invest and where do you underinvest? I think this one’s a little counterintuitive, but you should not invest in selling multi-year deals. Why? Because the retention will happen as part of the system. Instead, you skew their compensation toward getting people in the door. There's a core concept in systems thinking that says, “The system you’re in is perfectly optimized to deliver the results it’s already delivering.” It’s up to us leaders and system designers to nudge systems to perform the way we need them to perform.
That’s the sort of strategic thinking you can do when you think beyond averages, and have the data to support non-trivial decisions like this.
What do you think businesses should prioritize more?
Adopt a net dollar retention (NDR) mindset, because the world is more complicated than we want to believe! NDR is a calculation of how many of (or how much of) the dollars you’ve acquired have been retained. It’s a bit more complicated and requires more data, but it’ll change how you think.
This doesn’t mean you should overcomplicate things. I still believe younger companies should focus on achieving product-market-fit and landing new customers above everything else. But this is the direction you want to go.
I believe business leaders are hired to build sustainable enterprise value, not just bring in revenue. That increasingly means looking at NDR, ARR growth, and CSAT, as opposed to just LTV/CAC, which used to be primary. These newer metrics change over time, but tend to change slowly (read: over years), and lead to better decisions. This isn’t just my opinion. I encourage you to download the Meritech Public Comps data and run a regression, where you try to explain enterprise value as a function of other KPIs. You’ll find that NDR is the strongest explanatory variable (for my fellow data nerds roughly R^2 = 0.35). For a systems thinker, it’s pretty clear why—it accounts for the whole value of the business, and it’s just as useful for executive compensation as inter-departmental alignment.
Few people are comfortable with making decisions on long time frames like NDR demands, especially when their board operates at a quarterly interval. But in a world where the cost of experimentation and software are always declining, and barriers to entry are almost a joke, the riskiest thing is not taking risks (under a framework). If you can think in NDR, you can make decisions today that consider downstream effects that’ll occur in 12 or 24 months, like a customer being a bad fit and churning.
These concepts are pretty straightforward, but it gets complicated once we add the necessary nuances. For example, if you have consumption-based revenue that is more probabilistic than temporal, or a mixture of contract lengths, you quickly get metrics pollution. It’s the job of the RevOps leader to clarify and decide what’s useful.
What are some things you can learn from looking at NDR?
That math can be brutal. It might tell you all your revenue machine does is chew up ten cents of every dollar and spit out $0.90. That’s a “bad dollar.” In other words, that is not a customer, segment, or product line you want. Again, you only get the real value here if you’re looking at cohorts.
With NDR, you can see that dollars have a temporal value. They waste away. They decay. Or ideally, they grow exponentially. You can see that some deals are going to automatically be renewed, which is good money, and some are going to churn, which is bad money. Quarter over quarter, year over year, you can look at the health of the dollars you’re bringing in and calculate their true value. (Yes, that’s net present value).
NDR is valuable no matter who you are in the business, but let’s take it back to business applications. If you’re able to build business systems that help you understand your business at an atomic level, say how a specific product was performing, or how a region was performing, based on NDR, you can come up with really lucid answers to questions. For instance, “Is this recurring? Is it not? What are the chances of them churning? Is this a good customer?” With NDR, it’s clear. Now add usage and CSAT data, and you’ve got yourself an infrastructure you can grow with.
What’s a question you’re currently trying to answer?
I’d say I’m always trying to understand customers and their journeys in every gritty detail. I say, talk to the field. Go track the entire lifecycle of a sample of 10 customers that came in this quarter. Go figure out who’s the SDR who created the lead. Talk to solutions engineering about objections they heard. Heck, talk to legal. What was the closing process like? What did they redline? What about finance? Procurement?
Now compare that with those that were closed last year. Are there anecdotal differences? Any insights about the market or your GTM machine? I bet there’s gold in those conversations. I like to say RevOps is part science but also part art. It’s investigative work, and quite similar to the work of a product manager building a product.
From that, I always want to pull out a good group and a bad group and see if I can tease out even anecdotal differences to pursue. Maybe you find that someone who brought everybody onto the demo and had lots of objections early on ends up staying a customer for a long time. It’s often not intuitive.
The beautiful thing about it is, it’s almost magical that when you solve for a KPI like NDR, you tend to solve everything else too. CSAT, NPS, capital efficiency, etc., all fall into place.
Any other recommendations for business systems teams?
Yes, I do have one semantic recommendation. I say let's stop calling it a “tech stack.” A stack is a pile of things. If we instead call them architectures, it evokes all the right ideas. It brings the goal into focus. If it’s an architecture, everyone understands it needs a foundation and a structure, and there should be no unnecessary parts.
What’s one thing every system thinker can apply today?
I would say, draw your architecture. In trying to create it, you’ll probably realize there are gaps in your understanding. When I do this, there’ll be a spot with a big, red, transparent arrow with a big question mark that says, “What is the hidden process?” That’s an area to investigate.
You might look at your architecture as a series of flows, color-coding customers and sending a red dollar, a blue dollar, and a yellow dollar through the system to see where they go—and how they change. And based on your new understanding, maybe you color-code the parts of that pipeline based on how well they function. Green is good. Red is bad. Blue is optimized. And you take a step back and look at it with your executive team, with the perspective that it’s an architecture, and ask, is this good? Is this bad? Is this a focus for next year?
Not every business has the luxury of thinking on the timescale NDR demands. But those who do tend to realize dollars are temporal and fungible. They are not all equal, and if you want to build a higher-functioning system, you have to identify the bad dollars and stop taking them.