Discover more from Growth Croissant
Driving Growth with Paid Ads
Putting our CLV model into action
Today we’re putting our customer lifetime value (“CLV”) model into action and exploring a framework for how to invest in paid ads to drive growth.
Before diving in, it’s important to note that paid ads are often not the best way to drive growth early on. Like many other types of investments, investing in paid ads and marketing typically work best to accelerate growth. If it feels like you’re pushing a boulder up the hill, it’s probably best to avoid paid ads. But if it feels like the wind is at your back — your product is in a good spot, you’ve found your audience, and you’re experiencing solid organic growth — paid ads can be a great way to amplify growth.
Even if you’ve reached that point, learning to use any ad platform (e.g., Facebook, Google, TikTok, Twitter) takes time and patience. It also takes quite a bit of effort to develop ad creative, track results, and optimize campaigns. Make sure you have the right expectations before jumping in.
We’re mostly using investing in paid ads as a way to show how we can use our CLV model to make decisions — specifically, to set a goal for how much we can invest in adding new subscribers.
With those caveats out of the way, I’m excited to welcome our partner for this post: Tony Mecia, the founder of The Charlotte Ledger (“TCL”), which provides original reporting and journalism to Charlotte-area residents. As newspapers across the country continue to deteriorate, Tony and the team have shown what’s possible for a digital-first local news publication: The Charlotte Ledger now has thousands of paid subscribers, more than $250,000 in annual revenue, and a growing team of talented reporters.
I grew up in Charlotte and work at Substack (which is honored to support The Charlotte Ledger), so obviously I’m a massive fan of Tony. We initially met while I was working on Yem, where Tony was gracious enough to be the guinea pig for our crazy ideas, including partnering on paid ads experiments.
Let’s dive into the framework we used to set goals and measure performance for paid ads.
If you want to show that you value this type of work, please consider a free or paid subscription to Growth Croissant. 🚀 🥐
What’s our CLV?
One of the biggest challenges we see folks encounter with running ads is knowing whether it’s a good use of time, energy, and capital. Like any investment, we want to avoid lighting money on fire. A well-informed goal can help ensure we’re making sound investments.
Our first step is to put together a customer lifetime value model. We’ll then set a target for how much we’re willing to pay to bring on a new subscriber (i.e., "customer acquisition cost" or "CAC").
So far, our CLV models have only been two years, but we’re going to extend that to three years for The Charlotte Ledger. TCL turned on paid subscriptions in February 2020 (a year after launching), so we’re lucky to have quite a bit of retention data. Also, since TCL has made considerable progress with subscriber growth, we can feel more comfortable extending our forecast. When we view CLV below, that’s the cumulative earnings we expect to make over three years.
From our post on subscriber economics, we know the first step in determining CLV is to figure out how much cash we earn each time a subscriber pays us. The Charlotte Ledger offers paid memberships for $9 per month or $99 per year. Here’s a summary of TCL’s subscriber economics:
In our last post, we covered how to get our Invoice data from Stripe and update our decay curve with actual retention data. Let’s start by focusing on TCL’s monthly subscribers — here are a few key retention points on the decay curve:
88% make it past the 1st payment (i.e., do not cancel in the 1st month)
67% make it past the 6th payment
61% make it past the 12th payment (i.e., do not cancel in the 1st year)
Next, we use TCL’s retention data to forecast longer-term retention during years 2 and 3.We then combine the decay curve with our subscriber economics to produce an estimate for customer lifetime value, shown in the charts below:
Now let’s check out retention for TCL’s annual subscribers:
87% make it past the 1st payment (i.e., do not cancel in the 1st year)
81% make it past the 2nd payment (i.e., do not cancel in the 2nd year)
To stay conservative, let’s assume a slightly lower decay curve for our CLV forecast: 83% of paid subs make it past the first year, and 76% make it past the second year, as shown below:
As a quick aside, these are some of the best retention rates I’ve seen. In the Yem days, most publications we worked with would retain 60% to 70% of annual subs after the first year (vs. TCL’s 87% retained) and 40% to 50% of monthly subs after the first year (vs. TCL’s 61% retained). Kudos to Tony and the team for delivering tremendous value to their audience and doing an excellent job keeping folks retained.
What’s a good CAC goal?
Now that we know how much money we expect to earn from each subscriber over time (CLV), we can set goals for how much we would be willing to pay up-front to bring on a new subscriber (CAC).
The extremes are obvious. If our CAC is higher than our CLV, we’ll lose money on each subscriber and won’t be in business for too long. But if our CAC is too low relative to CLV, we may be under-investing in growth and leaving money on the table, or even opening the door for competitors to swoop in and capture our potential audience. We can use a few efficiency metrics to guide us toward the right level of investment:
Earn-back period — how long it will take to get our initial investment back; usually best to keep this under 12 months.
CLV to CAC ratio — how much higher (or lower) is CLV relative to CAC; usually want your CLV to be at least 3x higher than your CAC
It can help to think of this like putting a puzzle together, where the result is a range of CAC targets you can feel comfortable operating in. Once you have a reasonable range of CAC targets, you can shift toward the upper end when the wind is at your back. Here are a few more popular reasons to elevate your CAC target:
You’re experiencing favorable seasonality trends (e.g., holiday shopping, NFL season returning)
Your product is markedly better (e.g., launching a new feature, new episodes of Game of Thrones)
Growth is a company priority (vs. profitability), or there’s a generous macro-environment (e.g., zero-interest-rate policy)
Of course, the opposite is true: when you’re facing headwinds or any favorable trends start to reverse, it’s usually prudent to shift toward the lower end of your CAC target range.
Using TCL’s CLV from above — $162 for monthly subs and $222 for annual subs — we can come up with an initial range for The Charlotte Ledger’s two subscription plans: $45 to $75 CAC for monthly paid subs and $60 to $100 for annual paid subs.
In actuality, we would likely have an even higher CAC target for annual subscribers, potentially up to the point where the CAC equals the cash we earn immediately (e.g., $85 for TCL). But it’s also probably going to be hard to add a new sub through paid ads and convince them to immediately buy an annual subscription. It’s probably better to set our CAC target based on the earn-back period and CAC to CLV ratio for monthly subscribers.
Further, we may focus most of our paid ads efforts on bringing new free subscribers, hoping to convert them to paying subscribers once they’ve had a chance to experience our product. Also, most ad campaigns will perform better if there’s a sufficient volume of conversions to optimize spending. Free subscriptions can be a happy medium between paid subscriptions (perhaps too low volume) and clicks or impressions (likely to drive lower-quality prospects).
To back into a CAC goal for free subscriptions, we can combine our CAC target for paid subscribers ($60) with an assumption for how many free subs we’ll be able to convert to paid.
TCL usually converts ~20% of their free subscribers to paid subs over time — like their retention rates, TCL’s free-to-paid conversion rate is much higher than the usual 5% to 10%, another robust health metric for TCL. We expect a lower free-to-paid conversion rate for subs coming in through paid ads, so we assume roughly 5% will convert to paid, giving us a CAC target of $3 for every new free subscriber.
Equipped with our CAC target, Tony and the team can use it to set goals and guide their decision-making. If TCL runs a Facebook ad campaign for a few weeks and the CAC for new free subscribers is $10, they would turn that campaign off. If TCL spends $1,000 to print a bunch of flyers to hand out at a Charlotte FC game, to hit their $3 CAC target, they will want to get 333 new free subscribers (or roughly 17 new paid subs at a $60 CAC). We can also determine our break-even point for our $1,000 investment: based on TCL’s $222 CLV for annual subs, at a minimum, we would need about five new annual subs to make sure we don’t lose money on the flyers ($222 CLV x 5 paid subs = $1,110 in total earnings over 3 years > $1,000 investment).
With any marketing efforts, it’s helpful to consider the concept of incrementality, especially as your marketing efforts expand to new channels or scale. For any subscriber we bring in through paid ads, we want to know how likely they would have been to subscribe anyway, so we can gauge how much incremental value we’re getting.
From a modeling perspective, it’s easy to make incrementality overly complex, but it’s an important concept. Different marketing channels can have varying degrees of incrementality, and it’s helpful to adjust our CAC goal accordingly.
The most straightforward example is to compare branded search (e.g., buying phrases and keywords for "The Charlotte Ledger") against more prospective ad buying (e.g., ads on TikTok, Facebook, or Twitter). Folks coming in via search have much higher intent, yielding much lower incremental value (i.e., much more likely to have signed up anyway). Below is an example of how we could adjust CAC goals based on incrementality — given the lower incrementality, our CAC target for branded search ads ($5) would be much lower than our CAC target for prospective ads ($43).
Again, doing the actual modeling around incrementality is much less important than just being aware of incrementality as a concept and allowing it to work into your intuition.
It’s also worth noting that in nearly every case, the subscribers we add through paid ads will have a lower customer lifetime value than subs coming in via more organic channels. We’re intentionally trying to expand from our core audience to more tangential audiences — folks that may not understand or need our product as much. Expanding our audience almost always puts downward pressure on health metrics (e.g., free-to-paid conversion, retention rates, customer lifetime value). We should be hyper-conservative when making assumptions to set our CAC target, especially before we have any performance data from paid ads.
That’s all we got for this week — next week, we’ll dive into part 2, covering how to measure performance and optimize campaigns.
Let us know what you think in the comments. If you’ve experimented with paid ads for your product, what have you learned? Do you run paid ads as one-off, ad-hoc efforts, or do you have campaigns that are always on? If ad-hoc, what’s holding you back from running paid ads on an ongoing basis (e.g., too hard to understand the impact, too much overhead, investment doesn’t seem to be ROI-positive)?
Another round of applause for Tony 👏👏 — thank you for partnering with us on this post.
Thank you for reading,
We have some retention data for year 2, but the data volume may be too limited to be reliable enough to use in our CLV model. The year 2 retention data can be helpful by acting as guardrails as we shape our forecast. For example, TCL’s retention rates in year 2 are slightly better than our forecast, so we can feel good about our assumption being conservative.
This was a recurring topic at Hulu (where I tried to keep the peace b/w Finance and Marketing & Distribution), Crunchyroll (especially early on, we got recurring questions around how much marketing spend was truly lifting our subscriber ramp), and HBO Max (if I had a nickel for every time one of our friends from McKinsey said “incremental”…). The topic of incrementality is intertwined with attribution and can get complex fairly quickly. Potentially a fun post for the future.