SaaS Metrics: Logo Churn Breakdown

Based on a dataset of 250 Private SaaS companies

7 min readJan 11, 2018

If you are into SaaS metrics, you’ve probably read about a million articles about churn rates and leaky buckets. If you haven’t, here are some outstanding resources to get up to speed:

The purpose of this article is to present analysis on logo churn rates based on a dataset of 250 Private SaaS companies. Specifically, I will be focusing on the relationship between logo churn rates and ARPU, MRR and company age.

Source and Methodology

You can find all the companies of the dataset and the company metrics that I used for this analysis on GetLatka.com. The data was gathered by Nathan Latka, directly from 1 on 1 interviews he does with CEO’s on his podcast, The Top Entrepreneurs.

This sample contains the 214 private B2B SaaS companies of the dataset I used for my previous 2 articles, plus 36 new ones that were added since. You can check those articles here:

The main criterion I used for dividing the dataset into cohorts was that size of the sample in each one should be adequate; that is, at least 40 companies should belong in each cohort. Cohort sizes are visible in parentheses, next to the nametags on the horizontal axis of each chart.

Logo churn by ARPU

It has generally been observed that as the price point moves upwards, products and services become stickier and churn rates get lower. This observation holds true for the SaaS industry as well and is fully supported by the findings of this analysis:

The chart above presents logo churn rates by average revenue per user. ARPU, a commonly used SaaS metric, is actually the price each customer pays per month, on average.

According to the chart, the companies with ARPU of up to $25 have the hardest time retaining customers, churning 5% on median and 5.5% on average, per month. To get a better sense of the magnitude of these numbers, look at the corresponding annual values:

Assuming churn rates remain steady month over month, the annual churn rates for this group come out at 46% median and 49% on average; very high values that imply the need for aggressive (and, likely, expensive) customer acquisition strategies just to sustain a customer base, let alone expand it.

Moving up in ARPU, churn rates decrease rapidly. Median rates plateau at 1% for the last two cohorts, but the average still drops between the two by almost 1%. A steady 1% monthly churn rate translates to about 11% annual.

These findings are in line with Tunguz’s observations:

“Startups serving SMBs tend to operate with higher monthly churn, somewhere between 2.5% and 5%+, because SMBs go out of business with greater frequency and tend to be acquired and managed through less retentive channels, e.g. self-service. In the mid-market, which I’d define by average customer revenue* of between $10k and $250k loosely speaking, the churn rates I’ve seen are between 1% and 2% per month. Enterprise companies, those with customers paying more than $250k per year are typically closer to 1%. As the spend per customer grows, startups can afford to invest significantly more in retaining the customer, hence the improving rates.”

*Referring to annual revenue per customer.

Regarding logical interpretations of the churn-ARPU correlation, I would add that lower-priced products and services are generally easier to try out and possibly discard afterwards than higher-priced — think of buying a smartphone vs buying a house.

This stems from the different levels of cognitive load associated with any decision to switch to a new product/service, but also, from a practical perspective, has to do with the amount of value to be replaced by an alternate solution. Simply put, a higher-priced solution is more likely to be indispensable to a company, e.g. as a core component to operations, thus, harder to replace.

Lastly, SMBs are more agile than big corporations, can make decision and switch services way faster, with little to no switching cost.

So, does all this mean that SaaS companies offering services of higher value/price have an advantage? The short answer is probably yes. To get a better sense, let’s see how the companies in each ARPU cohort are doing in terms of MRR.

According to the second chart, as ARPU increases, MRR also increases. Most interestingly, it increases at rates comparable in magnitude to the rates of decrease in churn we observed in the first chart. So, looking at the two charts together, there is clearly a negative underlying relationship between logo churn and MRR as well, albeit a bit weaker than the churn-ARPU.

The median and average age lines where included to show that there is no underlying company age bias that could explain the gradual increases in MRR observed in this chart.

The first cohort, that is, the lowest ARPU cohort with the worst churn rates, presents clearly the lowest MRR median and average values.

The highest ARPU cohort ($3000+) presents by far the highest median and average MRR, more than double and triple respectively than its adjacent cohort ($800–3000), which is interesting, considering these cohorts have the same median churn and just a 56% difference in average churn.

This can be an indication that, in this sample, the MRR-ARPU relationship is stronger than the churn-ARPU one. But if that’s the case, how can we claim that the churn rates rather than the ARPU ranges are in fact the driving force behind the observed MRR behavior?

Let’s take a look at the “logo churn by MRR” chart to clear the air.

Logo churn by MRR

Looking at the median and average values, there is a tendency towards reduced churn rates as revenues grow. In this case, though, variance and distribution tell the rest of the story.

While the upper quartiles follow the downward slope of the average values, notice that the lower quartile thresholds are at 1% churn for every cohort, meaning that at least 25% of the companies in every cohort present churn of 1% or lower per month.

For MRR values over $200K in particular, the median churn is actually very close to that, with the maximum median being at 1.35% for the second to last MRR cohort. This means that among companies with $200K or better on MRR, more than 50% do it with less than 1.35% in monthly churn.

Moreover, the average churn values are very close for these cohorts, indicating there is not a significant difference between them. On the contrary, in the churn-ARPU chart both lower and upper quartiles follow a continuous downward slope, as do the average values.

So, what’s the conclusion? Churn rates correlate better with ARPU than MRR, but once you get to that $200–500K MRR range, there’s a very strong chance churn will be low or relatively low. Bellow that range churn will probably be higher, keep in mind, though, that MRR is not a very good indicator due to excessive variance.

Bonus “churn by company age” chart

You may have noticed that I didn’t address the age trend in the previous chart: There is nothing surprising about a positive relationship between revenues and company age — all companies want to grow and older companies have been doing it longer.

From a logo churn perspective, company age can be a factor, especially at the early years.

Younger companies churn more customers as they search for their ideal product-market fit. Once they start scaling and optimizing, many companies concentrate on their most valuable customer groups and churn (even on purpose) the less valuable. Lastly, companies either get their churn rates under control and grow old, or they die young.

Notes and Takeaways

SaaS economics are friendlier to companies offering mid and high-priced services because they have an intrinsic advantage in terms of customer retention, can spend more to retain their customers and, ultimately, retaining existing customers is cheaper than acquiring new ones.

According to statistical analysis (T-test), the adjacent cohorts defined by ARPU (first and second chart) are significantly different to each other with a level of certainty of about 90%. Not quite the standard required level of 95% to proclaim any definitive conclusions (in a scientific sense/approach), but close enough to allow for the observations and statements above to be made with confidence.

Most of the SaaS surveys I’ve seen include churn metrics only for companies above a certain run rate threshold. Be aware of that, as revenue range can have a pretty significant effect on churn rates.

Churn rates can also vary significantly by industry or core product vertical. I’ve talked with some founders and after popular request, next post I’ll break down this dataset by industry to see what comes out.

You are more than welcome to leave your feedback and thoughts or suggest topics for the posts to come. Find me on LinkedIn.

Originally published at getlatka.com.