[7/10] How to pick the right growth problems to solve
The relevant growth problems for the stage of a startup, and common pitfalls
So far in this series, we have gone through different frameworks for tackling different growth problems. In the first part (first 3 essays), we went through frameworks to:
transform a business idea into a growth formula and growth model,
build its roadmap, and
set up its measurement system.
And, in the second part (next 3 essays), we have gone through the frameworks to:
identify the right channels,
identify levers to drive user growth, and
identify levers to drive value growth.
This essay wraps up the second part of the series, and is a meta-framework post: the objective of this essay will be to help you, as a growth lead for a business project, to figure which framework to use at what point of time.
But why is this important? While the series has been written in a sequential manner, you will often land right in the middle of a problem statement. Such as, fix this metric, or build a new distribution channel. For such scenarios, this essay will share heuristics, as well as point to the relevant frameworks from the previous essays, to answer the following questions:
Is this the right problem space for the current stage of the startup?
Is this the most optimal problem statement to work on, within the problem space?
Is this the best solution for the specific problem statement?
What’s the right objective at this stage?
The job of a growth lead (or any managerial job, for that matter) is primarily not about having the right answers, but about making sure that the right questions are being answered. This might seem too abstract a statement, so let’s make it more specific: What is the right growth objective at this stage of the startup? What if the problem statements that we are exploring are not even relevant to the growth stage?
The biggest determinant of this ‘problem space’ for growth leads is the stage of the startup itself. That is to say, depending on whether you are an early stage startup or a growth stage startup or a late stage startup, your objectives and therefore your problem spaces will be very different.
(Note: the stage might not be the same for everyone working in a company. If you are leading growth in a ‘new initiative’ division in a large public company, you might have more resources at your disposal and might have organizational problems that differ from a regular startup, but for the sake of selecting growth objectives you should consider yourself an early stage startup.)
If the stage label is a little amorphous for your liking, you can answer the following questions:
Are we yet to achieve product-market fit?
Have we achieved product-market fit, but yet to get a distribution channel going?
Have we achieved product-market-channel fit, but just for one product/category and one persona?
Have we achieved product-market-channel fit in multiple categories but for the same persona?
Have we achieved product-market-channel fit for multiple customer persona?
When you are in the pre-product-market-fit stage, the objective should solely be to get to a working growth formula and model. That is to say, for the chosen customer persona and with the current state of the product, you have to get enough data flowing in that it helps you build a growth model. This topic has been covered in detail in the first essay. But let’s quickly look at the problem statements we should be working on and, inversely, the problem statements we should not be working on at this stage.
Validating the product-market fit requires instrumentation to collect data. Specifically, collecting retention data by customer cohorts and persona. If that is not getting collected, that becomes the only growth problem to solve.
If the data is getting collected and it seems that the product-market fit is not there. Either because leading indicators of Net Promoter Score are low, or because leading indicators of retention and engagement are lower than what your growth model requires for a positive net value per user (or are below the industry benchmarks). The only problem in this case then obviously becomes getting to product-market fit. The focus of problem-solving in such a scenario would heuristically be first experimenting with positioning, then with the product, and then with the market/persona itself.
What you should not worry about at this stage is user growth. Or about scalability of the channel, or other possible user growth channels. In case you are onboarding the intended customer persona in a handcrafted, personalized manner, yes, it is not going to scale. However, at least it removes the possibility of channel problems corrupting the product-market fit data. Also, if they are not being engaged or retained, despite such specific selection, changing the channel of acquisition is not going to solve the problem.
Once the product-market-fit stage has been crossed, and you have a working growth formula and model, then comes the search for a product-market-channel fit. That is: to find a distribution channel that’s scalable, repeatable, and sustainable. We have covered the frameworks to identify the right growth lane and channel in essay-4 of this series. At this stage too, the problem statement should be very simple: collect data for establishing product-market-channel fit, or the lack of it. (Apart from the instrumentation from the previous stage, here an additional instrumentation to attribute the visit to the specific campaign needs to be done.)
If product-market-channel fit is lacking, the overall objective remains very simple: iterate with the available levers to turn net value per user for the given channel to positive (covered in essay-5 and essay-6), or iterate with a different channel or channel type altogether.
The detracting problem at this stage often is trying to get the first campaign on a channel just perfect. However, at the stage where product-market-channel fit for the specific channel has not been established, the only objective is to collect data. And therefore, the campaigns need to be treated as experiments. There need to be guardrails to ensure that the positioning is right, channel-specific rules are being met, and so on, but creative excellence at this stage is table stakes and does not need to be over-optimised for.
Post the product-market-channel-fit stage, the business has one product (or category of products), working for one set of customers, being reached through one channel. And, this is where the prioritization problem really begins. In a way, it’s harder to solve but easier to focus when product-market fit is yet to be achieved or when product-market-channel fit is yet to be reached, since nothing else really matters.
However, now you have a problem of having multiple directions you can go into:
You can try to scale the same channel by making your funnel, and thereby your cost structure, more efficient.
Or you can figure out new distribution channels.
Or you can go for new categories or products for the same persona.
Or you can decide to pick a new customer persona to serve with the same or similar product.
How do you structurally think about these trade-offs?
How to prioritize a growth problem among competing problem statements?
These trade-offs exist, of course, since resources (human and capital) and time are limited. If we could focus and execute on all the problem statements, the exercise of prioritizing would be redundant. With this implicit limitation, let’s consider the objective at growth stage is to rapidly increase market share (in case of an existing category) or increase audience penetration (in case of a new category). The objective of the prioritization between different problem statements then is to pick the ones that maximize the chances of market/audience share increase.
The scientific way to come up with problem spaces for this objective would be to keep all other components of the product-market-channel-scale fit constant and tinker with just one component at a time. If we do that, it would lead to following options for problem spaces:
Keep other parts constant, increase scale
Keep product and market constant, build a new channel
Keep market constant, build new products/categories
Keep product constant, build a new market/customer persona
One key heuristic is that the options above are ordered as per their likelihood of success: the first one with the highest probability of success and the last one in the list with the lowest probability of success. Let’s understand it with an example.
Let’s say you work at a brand that sells collectible sneakers (category/product) to Gen-Z men in India (market/customer persona) and so far you have been using Instagram ads (channel) to acquire 1000 new customers per month (scale). As per the above heuristic, the highest prioritized problem statement should be: how do we increase the volume of users coming through Instagram ads while keeping value per user constant, at least, or perhaps even increasing it. If the first problem statement has either been sufficiently explored and we are in a zone of diminishing returns, or there are surplus resources after allocating sufficient resources to the first problem statement, we can then prioritize the next problem statement of identifying a new channel within the same growth lane i.e. another digital advertising platform (Google Ads, or Snapchat Ads, for example). And subsequently to identify a new distribution channel in a new growth lane i.e. partnerships or content marketing channels.
If the above options have been sufficiently explored, or there are resources over and above required for exploring the above problem spaces, the next option would be to build new categories and products (other collectibles, for example), and subsequently to explore new markets. And when it comes to expanding into a new market, the difficulty level would be lower for selling to the same customer persona in a new geography (Gen-Z men in South-east Asia, for example), but the most difficult expansion would be to keep other parts constant and move to a new persona (millennial men in India, for example).
Now, you might disagree with the problem space options: you might think there are more/different available options than just these. Or, you might disagree with the order of difficulty or likelihood of the problem space options. However, the key idea is to have a heuristic that is shared with the team and is used for evaluating different directions a growth team can take when the problem of plenty arises.
Another thing to note is that different options can have different gestation periods. For example, scaling a channel by optimizing the funnel might have a turnaround time ranging from a few days to a couple of weeks. Building a new channel might have a gestation period of a few more weeks, considering the instrumentation required (often operational) to get the first experiment live and then for the subsequent iterations. Building a new category or product might have a gestation period of months – a similar timeline might be required for a geography or a customer persona expansion – considering the research time and upfront costs involved.
The idea is to be aware of the differing gestation periods for different growth directions and to plan the growth roadmap accordingly.
What’s the right problem? What’s the right solution?
Once you have thought about the right objective for the stage, and have zeroed in on the problem space too, you still have to select the most optimal problem statement and the right solution. For example, you might have gone from the objective of increasing market share to the problem space of scaling the existing distribution channel, and then to the specific problem statement of increasing conversion rate. Or, you might have just been handed over this problem statement with no upward visibility of how this specific problem statement was chosen. Nevertheless, your narrow goal, especially in growth or late stages of a startup, is to often pick the most optimal problem statement from one problem space or pick the most optimal solution from a solution space.
The key to finding the most optimal problem statement is the one which has the biggest upside potential. For example, if solving problem A has an upside of ₹10 Cr over the next year, and solving problem B has an upside potential of ₹100 Cr over the next year, then, other things remaining the same, solving problem B is the most optimal choice for that horizon. But, you must be wondering at this point, how does one know the upside potential of a problem statement before even solving it?
Let’s say the problem space is increasing conversion rate. Within this space, you can choose any of the following problem statements:
improving checkout completion rate
improving add to cart rate
improving % of traffic who visit at least one product page
How do you pick the one with the most upside potential, without actually solving them one by one? By estimating the upside potential with a best case scenario.
But that still leaves the question of how do you find the best case scenario? It has to come from industry benchmarks (preferably the closest segment you can get the benchmark for), or even your business’ previous best in some cases. For example, if the product page view ratio for your brand is 40% while the same ratio for brands selling in similar categories of products to similar customer persona is 50%, you clearly have an upside of 1000 basis points. This when multiplied by 10 million impressions, other steps’ baseline numbers (see the table above), and an average order value of ₹1,000, gives an upside of ₹12,00,000.
The final heuristic that I would use while picking a problem statement is to avoid diminishing returns. From the previous example, the metric was 40% against a benchmark of 50%. However, the scenario changes if additional context is given that the same metric earlier used to be 30% and has reached 40% only after three most likely solutions have been implemented in the previous 3-6 months. In that case, unless the first three solutions were picked or implemented sub-optimally, it is unlikely that a fourth solution will lead to the desired upside of 1000 basis points. In fact, what the law of diminishing returns would suggest is that the solution will probably lead to an upside of less than 300 basis points.
So, putting the three heuristics together, you should first estimate the upside of solving different problem statements, and then pick the ones with the most upside. If the channel-funnel combination is in early stages of optimisation, higher upside opportunities will lie towards the top of the funnel. If the channel-funnel combination is in late changes of optimisation, more opportunities will lie towards the ones which have not already been worked on.
Coming to picking a solution, so far in the series we have covered levers to drive user growth in essay-5 and levers to drive value growth in essay-6. Additionally, a sharp root cause identification leads to a specific solution, and 3 different methods for it (funnel thinking, segment thinking, cohort thinking) were covered in essay-3.
Structurally, solution space to a growth problem is often about one of two things:
making the fundamental business driver better (e.g. making selection better to drive long-term retention rate), or
reducing the friction of that particular business driver’s discovery for the user (e.g. making discovery of selection better with better off-app communication or better on-app navigation).
While evaluating multiple solutions to the same problem statement we should see if it is a structural solution — falls is one of the above two buckets, preferably first — or a hacky solution. The problem with hacky solutions is that:
While they might show impact on that particular funnel step, the uplift doesn’t show up in the rest of the funnel.
Even if the impact might show up in the short run, the metric might regress to the mean in the medium to long run, precisely because structurally nothing had changed.
What not to do: have solutions beforehand
While the meta-frameworks above talk about common problem statements specific to the growth stage of the startup and point to the relevant frameworks, as well as offer additional heuristics, a word about a common anti-pattern: having solutions beforehand.
This happens in two scenarios. One, when the growth leads have their own bias towards certain tools. Such as, you consider product-led growth as the one true growth solution and scoff at digital advertising as a solution. Or, if you have decided investment in retention is better than investment in acquisition. While you might end up being right at times, having one tool to solve all problems without the business context will just lead to you running out of options to solve different types of growth problems in the long run. So, don’t be that person with just a hammer.
The other scenario, and which is more common, is that people suggest solutions to you without knowing the context of growth problems, and you pass them on as narrowly defined tasks to the team. Without seeing if it serves the right objective for the given stage. Or if it has potential upside to make a difference. And if, within the solution space, it is the most optimal solution out there. Again, don’t be that person.
Summary
In this essay, we answered the following questions:
What is the right objective and problem space given the stage of the startup?
How to estimate the upside to identify the most optimal problem statement to work on, within a problem space?
Which frameworks from the previous essays map to which problem statements and solutions?
With this, the first section of the series on planning growth and the second section on executing growth are now complete. The third section on managing growth will go live on 7th, 14th, and 21st March.
The next post on 1st March will make an exciting announcement about Skilletal. Stay tuned!
If you have any suggestions on this article, or need any clarifications, I am reachable at sudhanshu@skilletal.com
Could you please elaborate upon the calculation of upside (in the table above)