Client Lifetime Value Projection & Accuracy – Why Is It Critical For Online Growth

November 18, 2019

Client lifetime value, also known as LTV, CLV or ARPU is a projected figure intended to predict the overall monetary value of a new client throughout their “lifetime” – the period of time they are estimated to remain paying customers. It is usually based on a certain projection model.

Such a projection model uses past data from similar customers including spending and churn patterns, coupled with certain statistical extrapolations – to evaluate the “overall net value” of a client – how much will the client be worth to the company in total, over time.

This is not a theoretical question. Indeed, it’s a critical question for marketing and growth teams, as this figure will determine how much can be spent for online acquisition of a single customer, Aka target CPA (cost per acquisition), or target CAC (customer acquisition cost). A solid CLV model will take into account different client segments (country, age, product used, initial purchase amount and possibly other dimensions). By applying a certain logical breakdown to segments, a model can achieve more precision. But, It’s important not to go “over the bar” with granular breakdown, as too granular segments would not yield enough data for a statistically reliable projection.

Client life time value projection model

For example, looking into each country in the world as a client segment of its own might be too granular. If the number of acquisitions in each EU country is too small, its better to cluster them into larger buckets, for example, large countries like the UK, Germany, France separately, and then a bucket for the rest of west EU, and a bucket for east EU. depending on acquisition numbers, sometimes even this might be too granular and its best to start with just West Eu / East EU breakdown. over time, more data is accumulated & it makes sense to further break out more countries into their own segment.

The challenges of online businesses

Established large consumer businesses such as telcos, banks, insurers, …etc. have tons of historic data on their clients, thus, can predict quite accurately what is their expected CLV within different client segments. Furthermore, they usually operate within 1 or few core locations and know them very well.

In contrast, online-centric businesses & tech startups face a few challenges in building their own CLV projection model:

  • They lack significant historic client activity data.
  • They usually target many geo locations, possibly with very different CLV patterns.
  • They exist in a dynamic market, more susceptible to shifts, changes and surprises.

If you build it – will they come?

As explained in Peter Thiel’s iconic book “zero to one”, the answer is no. Important as they may be, relying on organic efforts, earned media and so forth is not a viable strategy for online companies in 2019. Companies must have a strong paid acquisition arm, i.e. online advertising, partner program, media buying and so forth.

In order to operate paid acquisition strategies effectively, companies must have at their baseline a projected CLV. Otherwise – how will they be able to decide how much to spend on acquiring new customers? estimate too high, and you might be losing money on each new client you acquire. Estimate too low, and you’ll be under-spending, leaving the arena to competitors. This will likely result in a small market share and low growth.

Segments of customers

In order to build a viable projection model, a company must do proper segmentation. There are 2 decisions here:

  • On which dimension to break: different demographic, behavioral and other attributes may be more or less relevant for each company.
  • After the breakdown into dimensions – the granular breakdown within the dimension, such as the example above in relation to geo location level of granularity.

Common pitfalls to avoid

Overly defensive approach – if you’re unsure about your CLV, you might take a defensive approach. For example, you might evaluate a certain segment’s CLV to be 6 month’s worth of income from a subscription customer, even though you already know that most of your customers stay onboard for more than a year. Surely, you don’t know enough yet about this particular segment’s customer behavior. But, unless you have solid reason to believe it a lower CLV segment – assume by default it will be your average customer value.

Next: frequent changes in your CLV projection – models have to sometimes be tweaked and adjusted to reflect the most updated data. Just make sure that you don’t change your estimates too often – creating distrust in the figure and resorting again to a defensive mode which will cost you in slow growth and low market share.

Time or Money?

Overall, in paid online acquisition there is a certain tradeoff between time & money. If you spend wildly, you’ll learn quickly your CLV by acquiring many customers in a short time, and learning their behavior fast. In that case, you might be losing in the short term for each unit. The other extreme, taking it slowly with an overly defensive approach will probably get you profitable on the basis of net unit economics, but will result in slow growth. Considering a company also has fixed costs to bear, in most cases this approach is even worse than overspending.

Finding the optimal ROI point isn’t easy, but that’s where you should aim and constantly work for.

Interested to learn more? Contact us