How Data Driven Cultures Tackle CAC

Posted on Posted in Analytics Strategy, Business, Data Driven Culture, Marketing Science, SaaS

Assume that you’re a founder of a tech startup. Assume that you’ve achieved product-market-solution fit. You’ve nailed it.

Time to scale.

Many founders are great at sales. But not all founders are great at marketing.

And that’s a bit of a problem because of three letters: CAC.

The Customer Acquisition Cost

CAC is the ratio between dollars spent on marketing, and new customers acquired. And it is related to valuation in a very important way.

Let me explain.

Take a look at the chart below. This is an output from a standard model of SaaS market penetration. Market size is 333,333 customers, the product will approach saturation at 51% of that target, with a monthly churn rate of 0.20% held constant to simplify everything that follows. (I reduced churn to a rounding error on purpose, it’s not typically that low!)



The top line is cumulative paying customers, the bottom line is new customers acquired each month. The shapes of both curves are orthodox. The rate at which new customers are added is very slow in the beginning. That rate increases slowly over time until product-market-solution fit occurs, and then hypergrowth begins, a point when growth accelerates dramatically. Growth takes off until an inflection point, and it gradually diminishes. That curve, in the shape of an s, generally occurs in nature. It describes the way a virus spreads through a population, or how air conditioning units are acquired by a market segment, these curves have been known in the marketing science for close to 60 years.

It is heavily driven by marketing spend (See: Bass, 1963, for the highly technical explanation).

Each customer pays money. This is recorded as the Average Revenue Per User (ARPU).

So, why 333,333 customers in a given market? It makes this bit of math very nice! Those 333,333 customers willing to pay an average revenue per user (ARPU) of $5 per month, which equates to a market worth 20 million dollars. Capturing around 51% of that results in a $10,000,000 Annual Recurring Revenue (ARR) or a Monthly Recurring Revenue (MRR) of $833,333. Having the conviction that there is a $20MM market is a table stake for doing a round, and, generally, powering an Innovation Driven Enterprise (IDE).

These are all very convenient numbers for the discussion that follows.

If it cost you roughly $5 to acquire a customer willing paying you $5/month, that Customer Acquisition Cost is x1. If it cost you roughly $10, then CAC is 2 times ARPU (2 ARPU).

This is a plot of the monthly marketing spend required to achieve that base growth curve at 1x, 2x, and 3x of Average Revenue Per User (ARPU).



At month 26, the month of peak growth, even with idealized CAC running at 1x ARPU, you’d have spent just shy of $50,000 on marketing. Yes, it sucks. But that’s the cost.

Instead of looking at these idealized curves, let’s try something more realistic:



CAC generally runs at 8x, 9x, or 10x ARPU. You want it to be far lower, but reality actively interferes with the efficiency in spend. At the peak, month 28, the monthly spend would $393,720 in the 8x ARPU scenario, and $490,900 in the 10x ARPU scenario.

Stated another way, at 8x APRU, the firm spends about $6.76 million over four years to acquire 169,000 cumulative customers. At 10x ARPU, the firm spends $8.45 million over four years to acquire the same number.

Achieving $10MM in ARR takes capital!

Gross Margin in SaaS tends to be very high, so often the difference between Gross and Revenue are blurred, within 20%, for shorthand (In other sectors, like retail, such a shorthand does not work!). The figure, months to recover CAC, plotted below, assumes a 0.0020 monthly churn rate. Lower CAC results in lower Months to Recover CAC. Linear. Easy.



In the instance of this chart, lower is better, and linear functions are linear. Lower months to recover CAC means lower raises and more equity retained by the founder. Higher CAC results in more equity dilution.

Customer Life Time Value (LTV) metrics are not a function of CAC and we’ll set that aside. In the models used for this analysis, churn is held constant and very low.

The Data Driven Culture’s Approach to CAC

A data driven culture would recognize the sensitivity of CAC on the outcome of becoming a business.

Implicit in the analysis above are some of the biases and realities of how markets are considered and thought about.

For one, the penetration curve is treated as inevitable. One way or another, the team is going to power their way through to a 50%+1 penetration against target, and there is absolutely no consideration to elongating the takeover time in an Innovation Driven Enterprise (IDE). At least in this model, the founders are planning for an exit at month 48, just as their winding up their early-majority customer acquisition strategy and making their Profit and Loss look way more attractive to a prospective buyer.

For two, exploding CAC’s in the face of an incorrect market size estimator is predicted up front. This point takes a bit of unpacking.

The founders assume that 333,333 customers are willing to pay $5/month for a solution. This represents a market worth $20 million dollars. Achieving a 51% penetration against that third of a million customers represents the easiest market sub-segments – the innovators, the early adopters, and the early majority. The resistance of the the market to conversion is supposed to greatly diminish (As suggested by Moore, 1993, though this is unresolved in my mind) after 51% through to supermajority (68% or so) and then suddenly becomes very, very hard again, because converting laggards is tough.

If it turns out there are far, far, fewer than 333,333 customers another set of opportunities emerges rapidly. The first indicator is that CAC dips, and then rapidly accelerates once again. New customer growth stagnates. If churn increases rapidly the startup may go into severe decline. In other words, the late majority customers are encountered much sooner than anticipated, network effects theoretically take off (a majority of word of mouth endorsements signals that there’s safety in the solution, as imitators are gonna imitate) and when the laggards are encountered then CAC soars very rapidly. For market segments, in B2C markets, it may be possible to burn through a small segment before realizing anything is amiss.

If it turns out that there are far more than 333,333 customers, a few good things happen to the firm. It can raise far more money to continue acquiring customers at 8x to 10x of ARPU until CAC rapidly accelerates. This can be viewed as bad news if the founding team did not pace out their equity positions well, but, in general, more sales solves most problems.

As a result, the signal of an rapidly increasing CAC is treated as just one signal of market size hypothesis testing.

For three, an explosion in CAC may not be treated as a failure in marketing creative. That is to say, if CAC takes off, from 8x to 20x in a very short period of time, generalized across all experiments, it is much more likely than not that penetration has been achieved. The disposition of a data driven culture, the desire to continuously improve marketing output and marketing results, may also encourage a consistent approach to content and experience optimization, as opposed to wild, across-the-board, brand redesigns, and a parade of frustrated creatives.

Four, brand and organic word of mouth effects are more likely to be considered and factored as a future discount against CAC. Stated for marketers, deliberate choices about spend against paid, owned, and earned media are likely to be made on the basis of weighted alternatives, not purely against anchor-and-adjust mechanisms. The entire marketing budget, not just the paid budget, would be assessed in CAC, with the effect of brand and word of mouth resulting in a lower overall CAC. Concretely, if direct attribution paid media is running at 11x ARPU, but there are a large number of unattributed conversion, the total CAC may run at 8x ARPU. This causes owned and earned efforts to be appropriately evaluated.

Management Responses To Soaring CAC

A common response to a decrease new customer growth is to lower the price to gain penetration.

“Don’t worry! We’ll make it up on volume!” is an inevitable statement made around the boardroom table. That, of course, is ridiculous. Decreasing ARPU just as you’re encountering market saturation causes Months To Recover to increase, AND, destroys unearned Customer Life Time Value (LTV) at exactly the wrong time. Indeed there’s very good management science supporting the notion that startups that start in high margins, and keep high margins, disproportionately survive against those that start on high margins and end on low margins. Note, also, that the price point was demonstrated to be viable at the nail stage, and modification of it at the scale phase is inopportune.

Reducing ARPU in response to soaring CAC is seldom the right response.

A better response may be to cause ARPU to increase.

Assume the initial forecast of 333,333 customers was optimistic. Let’s say that there’s only 40,000 total customers, a fraction of what was estimated. The firm has 35,000 paying customers, all acquired at CAC’s ranging between 8 and 12, but the CAC’s from the most recent cohort has spiked. At $175,000 monthly recurring revenue (MRR), there really isn’t an excuse for going all the way. The startup is small, and it isn’t quite a business yet. But it has a shot.

This calls for pivot, especially given the burn rate and the planning that was assumed for a $20MM market valuation.

The first question is if the market definition was too narrow, and if there are additional beachhead markets readily available to the firm.

Let’s say the firm specialized in software for the US and Canada. The initial assumption of 333,333 customers was too optimistic. Similar markets in the anglosphere would be Australia, New Zealand, and Great Britain. One step removed would include India, Israel, and South Africa. And two steps removed would include Mexico, Argentina, Peru, and Spain. Expanding export markets, going wide, sooner, is one approach.

Another is create features for similar industries for the United States and Canada.

Both efforts to discover 220,000 to 400,000 extra prospective customers, at a willingness to pay $5 ARPU, takes time.

The runway must be extended, the valuation must be defended, and the specter of churn must be held at bay.

One way is to increase ARPU by way of cross-selling to the existing customer base. This is an important sales channel because the acquisition cost of an incremental feature is far lower than acquiring a new customer. Indeed, it may be possible to discover incremental features of high value to this narrow population to increase ARPU, and thereby increasing the overall valuation of the market. That Willingness To Pay (WTP) would need to be mapped, and only the specific features that generate the most value, at the lowest estimated R&D cost, would be built.

Choices among a suite weighted alternatives

In the face of market size that’s much too small, there are a few choices.

The one that requires the least thought is to decrease the price while boosting the marketing spend.

Another one is to consider multiple beachheads.

Another one is to increase cross-sales earlier.

R&D resources are finite during this period, and must be tightly aligned with adjustment in marketing strategy.

(Competitor effects are typically not extensively modeled).

Recommendations for Data Driven Founders with respect to Customer Acquisition Cost and Market Segment Size Estimation

For one, marketing should be an integral part of the Build-Measure-Learn cycle. You are paying a significant portion of equity during the scaling period, in many cases, towards the peak, as much on research and development. Act as you would on the product management side.

For two, market size estimation is one of the riskiest assumptions in the startup plan, discuss the scenarios extensively with your management team and plan your minds accordingly. Planning is preparation of the mind. The founder should turn off his or her’s reality distortion field during the offsite planning and address the fact that initial market estimations will inevitability be incorrect, and plan. Working through the scenario during the planning and nailing phases should be dealt with. What if market size is 1/10th of the estimate? What then? What if the market size it 10 times estimate? What then? Knowing the choices up front will go a very long way to keeping the broader team calm when a bump occurs. What experiments can you be doing right now to validate the total addressable market? Are there changes in the total addressable market occurring right now?

For three, end exploding CAC before exploding CAC ends you. High CAC isn’t likely to be tolerated by your investors, and really shouldn’t be tolerated by you. In the event that you really are the next Zynga, there is no harm in making an equity-preserving move. In the much more likely event that you’re not Zynga, you can behave in manner that maximizes the odds of a transition from being a startup to being a business.

For four, if you’re not going to run for the exit at a particular point, you need to make that clear.