How to structure Product-Led Growth efforts by understanding revenue drivers
The first post of 2024 is focused on a structured approach to defining opportunities within the Product-Led Growth area and how to select what to work on and how.
Product-led growth is about growing the business through the product without overreliance on sales or marketing strategies. Understanding specific factors that drive revenue is critical to structuring product-led efforts.
Simple math relationships define a product’s revenue
The revenue is ultimately defined by a set of simple direct mathematical drivers, each of which may have other direct sub-drivers or indirect correlational factors. At the most basic level, revenue equals units sold multiplied by the average price. This is one of the first things that you learn in Economics 101. The same approach applies to much more complicated products, and we’ll go through a sample revenue tree in this post. You should build your PLG strategy around opportunities that directly or indirectly impact these revenue drivers.
We will start by looking at an example of a generic B2B SaaS company with seat-based pricing and a few add-ons. At the highest level, the revenue of this product can be disaggregated into new, retained, churned, expansion, contraction, and reactivation revenue.
It’s pretty hard to jump from the scheme above to defining opportunities in the product. So, let’s take a closer look at each of these.
New revenue
Reminder: we are going through an example B2B SaaS company with per-seat pricing, a few add-ons in the mix, and a free trial as the acquisition business model. The exact structure won’t apply to another business with different input parameters. However, building a revenue driver tree for any product isn’t hard. Minor complications are expected with marketplace models due to their multisided nature and a potential non-linear relationship between demand and supply.
About the trees…
The scheme above explains the new revenue driver tree for our fictitious company. There are a few minor things in its visualisation that I want you to keep in mind:
Purple lines indicate a direct mathematical relationship between different levels. For example, number of new users equals traffic multiplied by conversion to signup.
Orange lines indicate an indirect, correlational relationship between different levels. For example, the activation rate impacts conversion to paid; however, that’s a correlational relationship, and we can’t say with certainty that the activated account will convert.
Minor yet critical omissions were made in the scheme (and further schemes) that make the most difference for your product-led growth strategy. For example, a two-step signup will mean that you can disaggregate conversion to signup into more direct factors, such as the conversion of the first step and the second step. As another example, a set of ‘indirect’ drivers exists for each metric; however, these would vary too much and take up too much space on the scheme. Finding these drivers is the core job behind driving growth.
Segmentation plays a critical role in understanding the behaviour of each specific metric. For example, while conversion to signup can be averaged across different traffic types, looking at how different traffic types/marketing campaigns/landing pages perform is critical. That, though, is a different (and hopefully well-known) topic.
The boxes on the scheme indicate examples of product-led growth initiatives that you can build that will impact the driver. However, these suggestions are not exhaustive or will work for everyone.
Back to new revenue
OK. Look it up once again before reading. To save time and space, I didn’t disaggregate many drivers into additional factors. Ideally, it should outline all direct relationships and identify as many indirect relationships (or factors) as possible, and you should update the tree with more factors with time.
Identifying and prioritising specific opportunities
To put things into practice, you’ll need to identify the current level of each metric and assess how sensitive each of them is to changes. This is where your regular product management skills come into effect.
For example, if user research indicates that most new users (including account owners) do not know that you have add-ons, you can map it to “Avg. # of add-ons” and its factor of “Discoverability”. This will lead you to create hypotheses related to improving discoverability through building an addon store, adding hooks for addons within the core experience or adding addon-related tasks to the post-activation onboarding experience.
Assessing opportunities is your job
As I mentioned, the list of suggestions on the scheme is non-exhaustive (and mostly mentions product-related experiences rather than other types, such as discounts, pricing, and marketing). It is not meant to be used as a laundry list of experiences you need to build. There is no silver bullet; you must identify the best opportunities yourself.
You can take inspiration from this list, competitors or other products. Still, you must always map ideas to a specific driver/factor on this scheme and assess whether building this experience will impact it meaningfully.
Optimisation vs. Building new stuff
Most Product Growth teams spend time optimising existing experiences instead of building new ones. However, the scheme makes it clear that some drivers depend on the company introducing new use cases, creating new experiences, and building add-ons. While I don’t expect Growth teams to start delivering new functionality suddenly, any Growth team needs to interact with other teams to see which new developments should be packaged into existing plans and which should become an add-on. Growth teams can help analyse which use cases / new products will help drive revenue growth.
Structural changes
Global changes, such as changing the acquisition model from a free trial to freemium or adding freemium in addition to a free trial, are bound to disrupt your tree and change the metrics and the structure itself. This will have significant implications if you are contemplating or experimenting with global changes. The key is to identify the post-change structure and forecast baseline metrics.
With a forecast (or multiple scenarios) at hand, you’ll be able to assess whether the change will be helpful to your growth strategy or not. You can start from a different point by building a structure relevant to a global change and working backwards from it to identify what level of metrics you need to achieve with a global change to justify it.
Retained / Churned revenue
Retained and Churned revenue are two sides of the same coin.
Retained revenue describes how much revenue of the previous period was transferred to the next period. Churned revenue represents how much revenue we will not get because of the accounts that decided to stop paying for our solution. Other positive or negative changes to the existing account's base will be described later within the expansion and contraction revenue parts.
The same four rules described in the ‘New revenue’ section generally apply to Retained / Churned revenue and other types. You start by identifying sub-drivers relevant to your product, moving to finding opportunities and assessing/prioritising them.
Two perspectives – same results
In practical terms, revenue retention and revenue churn are the same (if you consider expansion and contraction separate). The reason for you to think about both goes back to the mentality with which you are approaching growth work.
Retention can help you adopt a mindset where you proactively work on driving continuous value discovery of the product’s features/experiences by the end users and helping them use more of a solution (adoption) at more frequent intervals (engagement).
Churn can help you adopt a mindset where you start thinking about all the possible reasons your customers decide to cancel their subscription with you and address these reasons.
While not necessarily groundbreaking, you can identify more growth opportunities by thinking about both.
The most challenging growth area?
Establishing indirect drivers of retention at the lowest level, identifying opportunities and launching experiments in our example would take a lot of work. High-impact low-level drivers (the farthest level from a starting point on the scheme) are rare and are rarely linearly connected to retention. Quite often, it is a combination of factors that includes external factors, such as competition, economic environment or change of strategy on the buyer’s end.
Some drivers of retention are hard to evaluate numerically. Typically deemed one of the strongest retention drivers, lock-in factors, can be assessed through proxy metrics (i.e., the number of use cases adopted per account). Still, its connection to revenue retention/churn will be hard to establish as it will be confounded with multiple other factors.
Experimenting with low-level drivers won’t be hard. It’s easy to understand whether discoverability, adoption, engagement or other measures have improved. However, given how lagging churn is, especially in B2B, the experiment's impact on churn would be hard to analyse.
Working on retention is the main avenue that connects back to the initial values behind PLG (such as building the product with the user in mind, which many years ago wasn’t the norm for B2B software products). Retention work in a B2B context means that you’ll have to dig much deeper into user-level metrics (as opposed to account-level metrics) and understand different roles and segments much deeper than before.
Approaching churn/retention work
Churn/Retention product work as a strategic direction
Given the abundance of low-level drivers, mature companies with enough resources may dedicate themselves to working on these drivers as a strategic priority. In that case, positive improvements of low-level drivers (engagement, adoption, satisfaction, lock-in factors, etc.) will be considered a success, even if there is no apparent change in churn/retention.
Lately, most companies tried to become more data-driven, which has resulted in cultures that reject initiatives with unclear impact on high-level measures. Given the additional pressure created by the challenging economic environment, many companies prioritise initiatives that stakeholders can defend as having a direct, measurable impact. Contrary to that, dedicating some of your resources towards initiatives that can’t be easily measured regarding top-level metrics should be cherished as long-term, strategic work.
First-month retention/churn
Another approach is to consider improvements to first-month retention/churn as a main priority. The metric, while still being a lagging indicator with a long time window, allows you to launch experiments rapidly, hold the experience of cohorts, and wait for the results.
The approach depends on whether you can get enough new accounts for sample size to be collected quickly. First-month retention won’t solve the issue of retention/churn being a complex metric that depends on many factors, so single experiments are unlikely to move the needle.
Cross-departmental projects
Product-led efforts could synergise with processes or initiatives the customer success department (or similar other role) is running. Identifying these opportunities and building a holistic strategy can bring more value than minor in-product optimisations.
Expansion revenue
Revenue is attributed to ‘expansion’ when an existing account starts paying more in the period under consideration. The difference between the expected revenue from the account (retained revenue) and additional revenue from the account in the period would be expansion revenue unless some additional component (such as contraction) was present during the same period. Essentially, it means a customer started paying you more than previously.
Products with a usage-based model naturally tend to have a significant component of revenue coming from expansion as the usage grows with the growth of their customers. In our example, a company with a per-seat model and addons with flat pricing get no benefits if usage (for example, meetings scheduled for scheduling software) increases. Instead, it would be derived from more seats purchased, plan upgrades, or both. Add-on sales to existing accounts would also result in expansion.
Expansion is vital for mature products
For our example company, expansion revenue is a critical pathway to growth. Incremental monetisation of existing customer base makes the company more valuable. The author of The SaaS CFO blog considers >110% Net Dollar Retention great. ProfitWell’s (now Paddle) analysis indicates that top companies have 20% to 40% of revenue from expansion.
Companies with solid expansion (and retention) can invest more in marketing because each paid customer's value grows with time. Top companies like HubSpot, Intercom or Salesforce have introduced more products and add-ons to drive further audience monetisation. They have a sizeable audience already trusting the brand. New products can help capture more revenue by making customers switch from competitors to their solutions while getting synergies from having their software stack managed by one vendor. As discussed previously, customers who adopt more use cases in your solution tend to get locked in more for the added benefit.
Optimisation vs. new things
Most Growth teams work on optimisation: building better discovery mechanisms, introducing a self-service marketplace, adding entry points to add-ons from within core experiences, and building onboarding for add-ons. Yet, expansion strongly depends on what the customer can expand to.
In our example, there are three main pathways: more seats, higher tier or more add-ons.
More seats will depend on whether the product is collaborative or can be made collaborative and whether each user needs a separate login.
Higher tier depends on packaging (whether the value of a tier is aligned with the needs and willingness-to-pay of the segment) and optimising towards customers being able to realise the value of a higher tier.
And add-ons need to at least exist for customers to purchase them, even though optimisation work towards making the value more apparent is important for products with multiple add-ons.
Our scheme didn’t outline new products as potential drivers; we only outlined addons. This is done for simplification, but that should be a part of the scheme if you build one.
Contraction revenue
Contraction happens when an existing account starts spending less on you. In our example, it could happen through decreasing the number of paid seats, downgrading to a lower plan, or removing paid add-ons.
Contraction growth work is similar to improving churn; most initiatives should overlap. Hypothetically, contraction could be an early signal of a potential churn. To avoid repetition, I will avoid writing too much time on contraction. If you work enough on churn/retention, your contraction should also be pretty healthy. For most companies, contraction is a much smaller category of revenue loss than churn anyway.
Seat utilisation rate & contraction
One particularly interesting thing about contraction is the seat utilisation rate. In the growing economic climate, many companies may not be closely looking at whether their paid seats are being utilised. At other times, however, teams struggle to defend their expenditures and will scrutinise each seat's necessity (or usage, for usage-based models).
Even B2C companies are engaging users about additional subscription benefits, with Amazon sending reminders of different prime benefits. Even though that might lead to contraction by itself (through reminders about unused subscriptions), this is a much more customer-oriented strategy in the long term. The same is best applied in the B2B context.
If seats are not being utilised, something is wrong: under-adoption, weak product-market fit, or a sales deal that oversold the product. Regardless of the reason, you will overestimate the actual performance of your product with inflated metrics, and it will be hard to work with a product that has strong new revenue metrics but fails to retain it. Don’t do that.
Reactivation revenue
Reactivation happens when a churned account becomes a paid account again. Most products that have existed for a long time have a significant base of churned accounts.
While there are indeed a lot of churned accounts in your database, most of them are unrecoverable. So the ‘pile of gold’ you are sitting on is more like gilt on a pile of ash. To better understand the real potential of reactivation, you need to identify which accounts are recoverable and what makes them so.
Beyond account recovery flows, I wouldn’t bother for most companies to invest significant resources into reactivation solely because the actual percentage of recoverable accounts is low. It's much lower than you’d expect it to be.
How to build your revenue driver tree
Building a tree like that, especially if we start by identifying only direct drivers (which will hold up mathematically), is not hard. It might be novel for some, but the concept should be easy to grasp. If it is novel for you and my explanations weren’t helpful, that’s on me, not on the revenue trees.
Regardless of the type of product you lead, your industry, and your monetisation model, you can always build a revenue tree. Follow these four steps:
Identify types of revenue
Decompose each driver into more drivers
Stop driver decomposition when you hit the lowest level where direct mathematical relationships can be established
Decompose each low-level driver into a set of correlational/indirect factors with
Data analysis (prediction models, correlational analysis, patterns)
User research
Identify types of revenue
There aren’t many ways a product can generate or lose revenue in a single period, so that part should be pretty easy. Should it always be disaggregated into new/retained/etc typisation? I don’t think so. It is helpful for B2B SaaS, but an ads-only B2C product may have a great tree with no such disaggregation. Instead, it may start right away with “ads impressions” x “cost per impression”, and “ads impressions” somewhere down the line could be disaggregated into different types of users (new, existing, churned…). Interestingly, for a product like that, engagement (pages per visit, for example) would be a direct driver, not an indirect one as it is in our example.
Decompose into more direct drivers
If two or more variables lead to the exact figure of another variable, we can say that the latter variable is decomposable further. That can happen via
Multiplication (new users = conversion to signup * signup visits)
Summation (all traffic = organic traffic + paid traffic + …)
Subtraction (retained users = userst-1 – churned userst)
Division (trial conversion to paid = new paid accounts / total new accounts, in a specified time window).
How far should you go? All the way. It doesn’t mean you’ll need to have everything on the tree but to be sure that you understand the picture wholly, strive to decompose into the lowest possible level. Later, you’ll be able to prune nodes that aren’t impactful.
Decompose into indirect/correlational factors
While there are certain truisms to follow from the get-go (for example, it’s unlikely that an account will convert if it has not gone through an activation moment), you’ll have to uncover most of these factors by utilising data analysis and user research.
While this work will take quite some time, you’ll at least know where to dig. Many teams wonder aimlessly through troves of data, not realising that the questions they ask will only result in getting “42” as an answer. Revenue trees focus your attention on what’s important.
Side note on building the tree
There is more than one way to structure that tree. Ideally, we want to utilise the MECE principle, which is brought up in any discussion involving consultants and clarity of communications. MECE means “mutually exclusive, collective exhaustive”. In some instances, it might be more fruitful to be redundant (not exclusive) for the clarity of analysis.
For example, instead of drawing new accounts as a function of traffic and conversion, you could visualise new accounts as a sum of accounts from different channels and, further, for each channel, describe that it depends on its traffic and conversion to sign up (or deeper visit conversion to signup page first). While this approach is mostly redundant and clutters the analysis, it could be helpful if these schemes are used in presentations or communications, as senior leadership may interpret the ‘non-redundant’ version as saying that scaling any type of traffic (say, viral) will have the same conversion to signup as the current average.
Structuring growth work/team
With a full understanding of all revenue drivers, there will be an unfortunate roadblock of not having enough resources to tackle all the directions that can lead to growth. Some work can be allocated outside the growth area, such as new product development or optimising engagement, adoption and satisfaction.
Still, the challenge remains: how does one structure the growth work?
Small team
For a small team that can efficiently tackle just one priority at a time, there are two options: focus on key underperforming and sensitive factor as a milestone or focus on key low-hanging opportunities.
Focusing on underperforming and sensitive factor
In this approach, the team will focus on searching for and continuously improving on a single factor. Ideally, this scenario means that the team has enough time to analyse their options and has high confidence that a factor can be improved significantly, and it is sensitive enough to the changes the team can make.
Working that way, the team has to decide when to stop tackling the factor. It could either be based on a specific milestone with pre-defined experiments and launches (with some leeway given for experiment iterations) or until the factor level hits a predefined level (or it goes for too long).
Both options are OK in my view, but the first one would be my preference as it empowers the team to set and agree on their deadlines (milestones are still time-bound) and encourages deeper thinking (as the team will need to present the logic of why the milestone should improve the factor).
Focusing on key low-hanging opportunities
The growth team may also identify opportunities that map to different parts of the revenue tree and cross-prioritise initiatives regardless of what they impact. This approach is beneficial for products with no prior growth work, as there are often many opportunities.
The team won’t be able to continuously deepen their knowledge and adapt based on new data, as their analysis will stay superficial. Constant switching is rarely beneficial for proper understanding, so this approach should not be used when the time comes to scale the team.
Big team
Big product growth teams have the luxury of spending enough time on getting deep insights both from quantitative and qualitative perspectives. They can typically stay on the same factor for an extended time and accumulate enough knowledge to understand what goes on for different segments. When you get to a big product growth team, the product itself is often complex and complicated. It’s not bad, but it also means much more complexity from a data and insights perspective.
The two approaches to structuring work in big teams would be by revenue type and by metrics cluster.
Focusing on revenue type
The approach may work for B2B SaaS companies as their revenue tree suggests that growth efforts may not translate easily between different types of revenue. As an added benefit, the company also gets someone from the growth team who ‘owns’ the type of revenue.
In our example, that would mean that there will be product teams responsible for New revenue, Retained/Churned revenue, and Expansion revenue. Potentially also for contraction and reactivation, but given their minor impact on the top line, I would expect it to be bundled in one of the core three.
However, not all revenue trees will look the same, not all revenue types will have the same priority, and not all companies can afford that amount of teams.
Focusing on metrics cluster
A metrics cluster is another way to structure the growth teams/work. This approach is helpful to minimise the overlap between teams’ activities. In the previous approach, the overlap between factors will mean that teams will have to agree on who does what and in what way. If your culture is prone to decisions by committee, the previous approach might not be the best.
Examples of focusing on metrics cluster:
By Customer Journey Map steps
By metrics closeness
The first one means that the teams own particular experiences. For example, the acquisition team works on every experience up to signup success, and the activation team works on improving the activation rate, onboarding and value discovery. The same goes for other steps.
The second one, in practice, is going to be highly similar. The closer the metrics are to each other (in terms of experiences or components), the more likely the same team will own them.
Applicability for sub-products or feature areas
Many product managers do not work directly on revenue, have responsibility for an area that has no immediate revenue impact (for example, focused on audience growth) or are responsible for a subproduct (for example, add-on).
Can we apply the same approach to drive growth work in these cases? Yes. Let’s go through a few possible scenarios.
Addon ownership
Working on an addon that modifies existing experience can be the same as working on a separate product. If you monetise the add-on, it must utilise similar drivers to achieve revenue targets. The list of drivers may be shorter, or priorities might differ, but the approach will be the same. That stays true only if there is a product-market fit for the add-on. Growth work can’t precede PMF achievement, as there is nothing to realistically scale before that. The tree will help you identify revenue gaps but won’t solve the lack of need for a solution.
New Product Development / Incubation products
The same thoughts apply to cases where you are working on introducing new products or building new solutions. Regardless of whether they target the same audience, a different segment, or a completely different audience, the first thing is to iterate until the product-market fit is achieved.
A new product won’t be able to utilise existing metrics as a baseline, and you may not have enough sample size for experimentation specifically. Still, as time passes, the revenue tree will help you build your growth strategy.
Non-revenue goals
B2C products with no immediate plans for monetisation (for example, focused on audience growth) or areas considered of strategic importance can still apply the same methodology. A revenue tree is just a type of structure you can build when analysing any business problem. It so happens that it is the easiest one to explore, but there is no principal objection to making the same structure for audience growth as a primary target.
Ending notes
All models are gross oversimplifications. Life is never as easy as any model makes it to be. Telling PMs that if ‘we just double our ARPA, we’d be golden’ is not helpful. The tree helps you break down high-level goals into workable chunks or connect specific initiatives to a growth lever.
There are more facets to growth that I could write in a blog post. I intentionally avoided any mentions of Sales-led growth, pricing and packaging (they have significant growth implications), and many other topics critical for anyone working on Growth. The post suggests how to look at how to structurally work on what experiences you build and optimisations in the product, but doesn’t go beyond that.
P.S. Lenny published a post on a very similar topic. Make sure not to miss it.