Bootstrapped CPC rule of thumb: ARPU/25


In the first year of business, you have literally no data for making decisions and predictions.

Even after the first hundred customers, half of those were serendipitous one-offs, not representative of repeatable, controllable customer acquisition, and the scale of the data isn’t statistically significant.

One of the root questions you have at the start, which is supposed to be data-driven (but you don’t have data) is: What’s the maximum I should bid for CPC (cost-per-click) campaigns like Google AdWords?

The answer for a funded startup is “Bid as much as possible, to get as many customers — and data! — as you can, as quickly as you can, then rapidly iterate from there in the presence of that data.”

Easy for them to say, but what about a bootstrapped, profit-driven business? Here you don’t have the budget to “spend as much as possible,” and you’re keen on getting a reasonable return on investment reasonably quickly, and you can’t just “spend to acquire the data.”

Here’s my way.

(Tune the exact numbers below if you disagree with my assumptions, but the process should be valid for anyone interested in chasing profits with a small budget.)

LTV = ARPU x 20

ARPU (Average Revenue Per User) is the amount you charge the average customer every month, which is typically a mixture of different quantities of customers at different tiers, special add-ons, etc..

LTV (Life-Time Value) is the total amount of money you expect to collect from a customer over their entire tenure with your company.  In general you compute this as simply “MRR x [expected months]” meaning the average number of months a customer sticks with you.  Some customers cancel in one month, some cancel in a year, some in five years, and some never cancel!  So it can be difficult to compute LTV accurately for small companies, and impossible to know for young companies (where e.g. five years hasn’t elapsed yet to see exactly how many customer stuck it out that long).

If you do have data, here’s my deep dive on cancellation rates and LTV.

But since you don’t, in my experience (and in a non-scientific survey of some of the 100 startups currently officing at the fabulously Capital Factory co-working space in Austin), a good pre-data rule of thumb is 20 months.

If you have an average customer lifetime smaller than 20 months, that’s a dangerously high cancellation rate for almost any SaaS business, and you need to focus on addressing the business issues before focussing on acquiring more unsatisfied customers. Use surveys and one-on-ones to try to understand whether it’s technical failings, lack of features, missed expectations, bad service, doesn’t hit pain points, or what.

A healthy SaaS company will have a higher number of expected months, but at the start you also will have lots of mis-steps with early-adopters where your product is at its worst — least features, least quality, etc — so it’s good to assume a low LTV instead of inflating it to where it might be in future.

CAC = LTV / 5

CAC (Cost to Acquire a Customer) is your average total cost to get a new customer, which includes (as accountants say) direct costs (AdWords spend, affiliate payouts, the fees your affiliate system charges to process them) and indirect costs (consultants and your own time). So to compute CAC, take your total costs to acquire new customers and divide by the number of customers you acquired.

In general of course CAC needs to be less than LTV, otherwise it costs so much to get the customer that you can never make money. A surprising number of startups have CAC > LTV.  Many justify this either by not correctly computing CAC (e.g. ignoring indirect costs) or saying they’ll “fix that later” by raising prices, finding other channels of revenue.  Others justify by saying they’re doing a “land-grab” for customers, and just having a customer at all has intrinsic value.

Profit-seeking bootstrapped companies cannot afford those delusions. Also you need something far stronger than CAC = LTV, because you need to pay for other business expenses and still produce a profit.  So how big can CAC be before it’s “too big?”

Growing, funded SaaS companies who treat CAC with respect often commonly target CAC = LTV / 3.

Back at my second startup IT WatchDogs, my co-founder Gerry Cullen used to say “A third to built it, a third to get rid of it, and a third to keep,” meaning a third of revenue goes to pay for hardware/inventory/shipping costs of the sale, a third goes to what I’m calling “CAC” here, and a third for the overhead costs, development costs, and profit.

That’s a good model, and I think a bootstrapped company can copy it, but I urge profit-seekers to instead adopt an even more strict model of CAC = LTV / 5.  The reason is that at the start you should be able to find a few efficient ways of acquiring customers, even if those get saturated over time, because it’s exactly what you need early on when you have least-money, least-branding, and least-insight into marketing messages.

CAC = ARPU x 4

If you combine the previous two results, you see that the cost to acquire a customer should be no more than four months of revenue.

Another good way to think about it is: “The payback-period for my cost to acquire a customer is four months.” Also, ideally you’re getting the first month of revenue back immediately, so it’s really three months of cash-float.

Companies with large budgets to deploy at scale will often be happy with 12 month payback periods; some very high volume businesses like shared hosting will accept 24 or 36 months even! But a bootstrapped company’s cash-flow won’t allow it, even if the math would work in the long run.

Conversion Rate = 1%

Conversion Rate is the percentage of visitors to your website who convert to a paying customer.

This is another step which in practice should be completely data-driven, segmented by customer type and marketing channel, segmented by landing page, A/B tested and iterated, blah blah blah. But since you don’t have data, and you don’t have enough visitors to have real ratios, you have to take a swag at this number.

In that same informal survey I ran, and bolstered by other formal surveys, a huge number of bootstrapped SaaS companies report a 1% conversion rate.

Another way of saying the same thing is “You need 100 visitors to make one sale.”

And since you need to incur no more than CAC dollars in the making of that sale, you need to incur no more than CAC/100 dollars in the making of each of those visitors.

And if you’re running a CPC campaign, that means you can pay up to CAC/100 dollars per click.

And since CAC is ARPU x 4, we can substitute and get the end result:

CPC = ARPU / 25

So for example if your average customer generates $50/mo, you can spend $2/click.

Indeed, this is a great way to prove one of my main arguments for all bootstrapped companies, which is that you should charge a lot more than you think, in part because it enables you to pay quite a lot per click, which enables a wide number of marketing channels, and out-bidding parsimonious competitors whose paltry LTVs preclude them from competitive marketing spend.

“But my numbers are different!” OK, but now you have a formula you can plug them into, to arrive at the right answer.

29 responses to “Bootstrapped CPC rule of thumb: ARPU/25”

  1. This is a great post, Jason. I’ve worked with a lot of start-ups and one of the hardest things for them to do is predict any rational financial projections. As you pointed out, there just is no data. Every once in a while we get lucky and the company we are modeling on happens to release a bunch of early company data that helps, but generally no. These are some great benchmark numbers.

  2. Good point. This is exactly my problem with books like Lean Analytics. You need lots of data _first_ before you can do something useful with it. If you have 30 daily visitors to your site, you really can’t measure things like conversion rate.

  3. Hi Jason,

    I really like posts like this where there’s lots of equations to help SaaS businesses. A SaaS publisher can apply that to their business to create a workable model and help them work out a plan.


  4. My startup definitely can’t “spend as much as possible” but I do feel like there’s some room in that first year to “spend to acquire the data.” What are your thoughts on spending a good amount on AdWords for a very limited amount of time (maybe 3 to 6 months) just to get some good SEO / conversion / target audience data?

    • I think if you’re not spending $5k/mo you’re probably not getting that much data, in fact. False-positives will overwhlem real results.

      Unless you’ve raised money for the purpose of identifying new marketing channels as fast as possible, I think you should focus on getting real revenue from real signups rather than “just lean” in the generic Lean Startup manner.

  5. The obvious-though-unstated corollary is that “consumer” level apps can’t use advertising to grow, since $5/mo allows for only $0.20CPC of which there ain’t none.

    • In fact, you could go the other way and say: “if your niche has a price of x/click, you need to charge 25x/mo for your service”.

      Are you finding $2/click anywhere?

      So if you have to pay $5/click for traffic, you need to charge $125/mo!

      • Another great point! Yes I like thinking about it the other direction.

        Yes we can find $2/click (or equivalent) in lots of places.

        Yes if you’re unable to find something for less than $5/click, then you need something expensive -OR- you need to have a reason why you can relax one of the assumptions.

        Example: Raise money because acquiring customers at all is more important than profitability. Then you’re fine.

        Example: Prove that you can convert at 5% instead of 1% because you’re a badass at the landing page.

        Example: Prove that you can convert at 5% instead of 1% because the lead-quality is 5x better than average.

        Example: Have a pricing model where people naturally pay more over time, so you know you’ll make it back eventually.

        Example: You have a viral model so that one customer actually have a much higher LTV.

        This is just the start of what is possible, but yes, unless you can demonstrate that one of the assumptions is wrong in your case, then $5 CPC requires big-time revenue on the other side to be profitable.

        • Can you share some of the places (that actually generate clicks) where you can find $2/click? In the niche in which we operate CPC in adwords has reached obscene prices.

          • Of course it varies by niche and marketing channel, so that’s not really possible to answer in a vacuum. If you have low MRR and you’re in an expensive, competitive space, that sounds like a business that’s very hard to make work financially, and you’ll probably need to get creative about how to get customers — out-think instead of out-spend — and of course it’s unlikely to find such a method ever.

    • Yes I completely agree. In fact I just made this exact point in my Microconf keynote just hours ago. :-) This another reason why I don’t think self-funded startups should sell a consumer product. B2B for the win.

  6. Incredibly easy-to-read article and very useful thanks. We’ve got a fairly mature (3+ years) SaaS product with around 400 customers. Our average customer lifespan is 377 days, and MRR is around $20.

    Do you think we should focus on improving the product and raising that MRR, or growing?

    • Good question. That cancellation rate is dangerously high — you won’t be able to grow the business to an interesting size because you’re losing so many customers. Of course increasing MRR is always wise, and since it’s so low right now, it might be relatively easy to get it up to $30 or $40. That’s wise to do as well, but you can’t ignore the fact that half your customers aren’t getting value after a year.

  7. Great post, Jason. I heard about it in Twitter: Good content always helps for buzz!

    As other entrepreneurs, I´m not sure how to apply this bootstrapping model in my specific case. We´re working in a license revenue model, not SaaS.

    How can we approach to the MRR formula? Is LFV=Price? How to define period of revenue like your “x20” proposal?

    Best from Spain

    • It’s actually even easier and more accurate when you’re on a licensed model, because you don’t have to guess at LTV. LTV is simply the price you charge. So if you need CAC = LTV / 3 or LTV / 5, then you just take it from there.

      • Thanks Jason. That´s exactly what I was thinking… The pity is: I can´t define final price without considering CAC and CPC in a SEM channel, and price can be a barrier for some early adopters.

        But, who said bootstrapping was easy?

  8. I have often done these sort of back of the envelope calculations for myself and others. The only number I wasn’t sure about was what percentage of LTV to spend to get the sale. I tend to aim for somewhere around LTV/4. Intuitively it feels about right. So I was interested to see you came up with a similar ratio.

  9. Thanks for sharing such an interesting post, couple of questions:

    1) let’s say they pay in advaance for 24 months could we say:

    LTV = MRR x 24

    2) Is CAC basically the Ad cost, or is the total cost. For instance: let’s say I buy form a vendor a domain a 1$, I resell it at 5$, and doing Ads to get this order costed me 2$. Is CAC just these 2$ or it’s 3$ (2$ + 1$ of the cost of the domain I resell)

    • 1) If it’s impossible for them to get a refund, for example a pro-rated return if they cancel in 3 months, then yes you know LTV = MRR x 24, or even better if some of them stay beyond that point. Indeed, companies like WP Engine and Rackspace have more like MRR x 48 or MRR x 60 because of low cancellation rates due to service and the nature of the customers. Of course in the presence of data you should certainly adjust that calculation.

      2) In a recurring-revenue company, CAC is the sales cost only, which means sales, marketing, adverts — the costs incurred one time in order to get the sale. What you’re referring to is what’s known as “cost of revenue,” meaning costs associated with actually serving the revenue. In the case of my company WP Engine, that’s all server and support costs. Those are also VERY important, and are the cost components of GPM.

      In the way you described it, you’re talking about a non-recurring-revenue business. In that case, all costs of revenue and costs of sale should be bundled together as “COGS” (cost of goods sold). That math is actually much easier than the recurring-revenue model, because you know the totality of the revenue and of those marginal costs, so you can target e.g. COGS = LTV / 3 like we did at ITW or (hopefully) something stronger than that.

  10. This is PERFECT timing for me so thankyou Jason. I’m doing some customer validation experiments at the moment with facebook advertising.
    This gives me a great framework for verifying my back of fag (cigarette) packet calculations.
    Perhaps more importantly, I can compare each of the steps for my scenario and adjust the final costs.
    My concept is community professional coaching so much of the costs (in theory) are borne by the community members making contributions and sharing expertise.
    So. .. I’m considering whether I can justify increasing the ratio. For example, CAC =LTV/3.
    What do you think?

    As further clarification , although currently boot strapped, I have another funding source which allows me to run small non profit generating experiments.

    • LTV/3 probably does make sense. Really, you need to model your business yourself and see what ratio results in a business with the kind of cash-flow you want at a reasonably small scale.

    • Sure, and I said the same thing in my article.

      It depends completely on the trade-off you can make between higher growth or higher cash spend. Skok is writing for *funded* startups, whereas the guidelines in this article is for *bootstrapped* startups.

      Funded startups should trade higher short-term cash-burn for higher growth-rate, because they aspire to be huge, and you can’t get huge (and stave off competition) without high growth (nowadays, and in tech).

      • Thanks, Jason! I think your perspective on these issues is so…well-balanced. That’s the only word (phrase) I can think of. You understand the mind of the self-funded startup. But you don’t look upon funding as a kiss of death either. IMHO, both communities (funded and self-funded) could do a better job of learning from one another. This is slightly tangential but….what can be done to highlight the “middle ground”?

        • I don’t think there’s a middle ground. They have very different goals, so they often should make very different decisions.

          Averaging the two doesn’t automatically make sense. Rather, decide what your personal goal is, exactly, then make decisions which are consistent with maximizing the chance you can be successful with those goals.

Sign up to receive 1-2 articles per month: