If that title sounds familiar, yeah, I cribbed it from the recent TAM piece. Today, we’ll explore how CAC (cost of acquiring customers) is misused.
Traditionally, the cost of acquiring a customer for an insurance company was pretty easy to measure. You paid a commission to the producing agent at the time of sale and subsequent commissions upon renewal. There might also be some marketing support to help the agent find clients, but, for the most part, CAC was a variable, predictable cost.
That changed when some insurers began selling direct. CAC was now a function of call volume and call center staffing and utilization costs which were more fixed in nature and presented the prospect of leverage through growth.
Of course, once direct insurers realized the benefit of this leverage, there was an incentive to ramp growth to achieve scale which led to…those ever present auto insurance commercials!
Crossing The Moat
One might even say there was an advertising flywheel! More ad spend brought more growth which lowered the acquisition ratio which allowed lower prices which encouraged more ad spend and more share, etc. Eventually, the ad spend became a moat keeping smaller players from being able to replicate the direct model efficiently.
However, that hasn’t stopped new insurance startups from trying to build direct brands that can eventually achieve profit at scale. While some have had some impressive growth, there remain unanswered questions about whether the cost of trying to get past the moat is economical.
Root and Metromile appear to be paying roughly $300/customer which is about 20% of premium. That’s an affordable number if the retention was industry level (though it’s not there yet). Lemonade suggests they are much lower in $, but also has a far lower average premium so higher in %.
Now, there are some questions about whether these self disclosed CAC estimates suffer from the same self-graded exam pressures that influence established companies’ reserves which was covered well by Inside P&C.
However, there are bigger issues than that. First, it is easy to manipulate CAC by writing bad business. Let’s do an example…
Lying With Numbers
Say I offered to pay anyone who bought a product from me $100. If I put it on twitter, that’s $0 of advertising so my CAC is $100/customer. It’s a one time only offer, so there will be no repeat customers.
This is great! I can brag about my low, low CAC of $100, but I have no ongoing business, so I am kind of up a creek, especially if my gross profit is < $100. I can never make back my losses.
Remember, the whole point of distributing direct, at least for a product that is a “subscription service” like insurance, is to get recurring revenue. That means CAC is meaningless without understanding its impact on retention.
If I am spending to get high churn customers, my ongoing acquisition spend will always be stubbornly high. If I can get sticky customers, I can justify a higher nominal CAC, because the long term average acquisition cost is lower.
I know, rocket science, right? Not exactly, but it’s something no startup has talked about…how much of their low retention is a function of having too wide a funnel to try to demonstrate “hyper growth” and to make their CAC look lower?
More Mathematical Manipulation
Let’s do another thought experiment. If I spend $1000 to acquire 10 customers who each produce $50 of revenue, my CAC is $100 and my revenue is $500. But if a year from now, 9 have left and only 1 remains, my effective CAC was really $1000 and my revenue has fallen to $50.
Assume that one customer who stuck around stayed for six years, so there would be $750 of total revenue vs. the $1000 CAC.
Now, let’s go the other way. Say 9 of the 10 customers hang around for 6 years. That leads to $2750 of total revenue. Do you mind if you paid $1800 to acquire them? Your CAC went up to $180 but your recurring revenue is $450 and the effective CAC is $200.
| Which is better? | Low Retention | High Retention |
| Acquisition Spend | $1000 | $1800 |
| Disclosed CAC (10 initial customers) | $100 | $180 |
| Effective CAC (Spend/remaining customers) | $1000 | $200 |
| Recurring Revenue | $50 | $450 |
| Lifetime Revenue | $750 | $2750 |
If you really want to measure the effectiveness of your acquisition spend, CAC should be weighted by retention. Sure, I know there are lifetime value (LTV) calculations that account for retention, but those can be manipulated as well (maybe that will be part III of this series?).
The point is, while writing a bunch of high churn customers will lower LTV, it does not impact CAC and more companies are reporting CAC than LTV.
Subsidizing Business Is CAC, Too
There’s one other big flaw with CAC. It ignores loss ratio – which is kinda insane. If I were a restaurant and gave you a coupon for 20% off your meal, you would call that an acquisition cost. If I sell a product online, and give out a code for $20 off if you mention the ad you saw that brought you to my site, we call that an acquisition cost.
When I underprice my insurance to generate growth, we say that’s part of the loss ratio – and I pass 80% of it on to my reinsurance panel! Wuh???
If an insurer is writing business at a loss to promote growth – and all indications are every high profile startup is doing so – they don’t consider any of their loss ratio as acquisition cost.
Now, we can debate how much to put in and that’s fine. Like if I’m targeting a 70 and I put up a 75, is that normal volatility or acquisition spend? I think it’s safe to call that volatility.
But if I’m writing at over 100, that’s pretty clearly intentional underpricing. So when you look at all those self-graded CAC estimates, you might want to go and take a big chunk of the loss ratio, multiply it by the premium, and add that to the company’s CAC estimate.
And yes, a good LTV model will capture this through lower expected profit (note: most LTV models are BS because they only account for revenue, not COGS), but the point is the company is still fooling itself if it is evaluating its LTV “success” relative to a depressed, artificial CAC.
Choose Your Own Adventure
Which brings us to our final topic, the ridiculous “adjusted gross margin” concept. I talked about this in the PTAM concept recently, but many of these startups measure their financial success based on premiums less losses less overhead, which ignores CAC completely.
Perversely, if these companies reported a truer estimate of CAC, they could boost their contribution margin!!! Let’s see which CFO rushes out to be the first one to redefine CAC higher now!
But it also exposes what nonsense adjusted gross margin is as a concept. As I suggested with PTAM, the better metric is technical ratio which would reflect the acquisition costs but exclude overhead.
In this approach, whether spending is classified as loss ratio or acquisition is irrelevant, which is a good indicator that this is the best measurement of financial results.
Learnings
Put all this together and what have we learned?
1. CAC is an interesting metric, but it is subject to manipulation and investors should ask for more disclosure on its calculation.
2. CAC is only relevant within the broader context of company profitability. If your CAC is low but your retention stinks, you are likely lighting money on fire.
3. If your CAC is low but your gross margins are poor, you are also likely lighting money on fire.
4. If you have strong unit economics, then a high CAC is only of modest concern. If it is high because you can’t scale, then it would be troublesome.
5. In fact, if you are highly profitable with a low CAC, you probably should be spending more to grow even if your CAC metric gets worse.
CAC should be a byproduct of the strength of your business, not a creator of it. If you have a strong value proposition, then you need to make sure you are good at acquiring customers efficiently to profit off your model. If you don’t have a profitable core business, then no expertise at acquiring customers will ever make you successful.
This will likely be the last post of the year, so happy holidays to all and thanks for reading! Feel free to leave suggestions in the comments if there are topics you’d like me to cover more or less in the new year.