From Growth Hacking to Product Thinking
Gousto’s approach to solving customer problems to create massive business value has radically changed in the last two years. We’ve learnt that Sean Ellis’s Growth Hacking is great in the short-term but destructive in the long-term. We’ve expanded our principles to reflect Brian Balfour’s better-defined “New Age Growth”, and adopted Product Thinking to unlock huge A/B tested conversion, retention and AOV uplifts.
The story and secrets of our evolution are grounded in the theories of great thinkers across the world. I’ll share these secrets here, with detailed examples of how we use them today.
When Boris Johnson announced the first lockdown, at Gousto, we could not service the massive demand for food delivery. In the second two weeks of March, I was up every day from dawn until midnight working out ways to prevent our service from being flooded by orders.
We blocked signups, we blocked resubscriptions, and we blocked unsubscribed customers from ordering one-off boxes. We created queues for people to wait to resubscribe and tried our hardest to prioritise NHS workers and vulnerable people. Each decision we made affected thousands.
But when the first storm subsided, and the company stabilised, I looked around at our Growth team working with a large tech squad and wondered…
“What the hell are we going to do for the next six months?”
I felt incredibly lucky to be at a company that had been positively affected by Covid-19. But in one week, my role as Head of Growth at Gousto had quickly become Head of No-Growth. We had to stop all our tests indefinitely until we could release more capacity in our factory. As lockdown looked like it would last longer and longer, I wondered if the team and I would sit around doing nothing for the rest of the year.
Life had other plans. In Growth, we pivoted to work on capacity management and other work. And one day, I got a call from my amazing ex- boss Lavi to tell me an interim role had opened up in Gousto’s Menu Tribe to be a Product Lead there. By this point, the team were getting on perfectly well without me. There was no better time to realise my dream of moving into Product.
Now, having worked in Product for the past year (I stayed!), I’ve got to look back on Growth with an entirely new perspective. The principles in Growth and Product are very similar, but reflective of the backgrounds of those that created it. So with that new and relatively unique perspective, I’ve solidified my thinking on a few big home truths that span both disciplines.
Do not drink the Kool-Aid on Growth Hacking
Disclaimer, I have a bone to pick with Sean Ellis. I think he is an exceptional thinker, but I think he took a beautiful set of principles to extremes, leading hundreds of aspiring Growth Managers to an early grave. The problem stems from his bias towards speed, ideas, and iteration, which can crowd out teams’ ability to solve actual customer problems.
By focussing primarily on idea generation, teams can fall into the trap of building solutions that don’t actually solve customer problems. Purely selecting ideas based on ICE scores also prevents the use of coherent product strategy, which typically can release more value over time.
Old Age Growth: Great in the short-term, destructive in the long-term
In the early stages of a startup, a Growth team might net valuable wins by finding things that are genuinely broken, missing or misunderstood in a flow. However, as a business finds Product-Market fit, like Gousto has, identifying these’ silver bullets’ becomes less likely. And Ellis’ loops of rapid ideation and iteration can often end in a crowded user experience optimised to a local maximum. I can only imagine that’s why his website has become so complex.
I do not doubt that you can sometimes achieve big wins with comparatively low effort, but I do not think that this kind of thinking can sustain long-term growth efforts for years and years.
When we first set up the Growth team at Gousto in early 2019, we moved in an untouched opportunity space. Many of the core lifecycle recovery journeys were missing, so we could realise very significant retention wins in a few short months. However, within 3–6 months, these opportunities dried up, and our success rate of tests dropped significantly. To drive the next phase of metric growth, we had to dive much further into genuine customer problems to solve, rather than test & learn at speed.
We learned it is essential to be opportunity-driven, not ideas-driven: an approach the Product world has used for years. And when Brian Balfour began crafting his vision of Growth after Sean, he tied in these approaches to develop a much more problem-centric definition of Growth.
Enter Product Thinking (and New Age Growth)
I believe Brian’s New Age Growth practices are a combination of two Product world frameworks that existed years before — Lean Principles and Design Thinking— with a big splash of Brian’s novel and excellent commercial frameworks and clear explanations. (Disclaimer #2, I’m a big fan of Brian’s).
How Design and Lean Thinking underpins “New Age” Growth
Design thinking is an iterative approach to identifying valuable opportunities, just like the test & learn principles within Growth. The human-centric process requires you to (1) deeply understand user needs, motivations, and problems. (2) define opportunities and assumptions and (3) ideate, prototype and test to challenge assumptions and learn what is most valuable to solve for.
My favourite model built from Design Thinking practices is the Desirability, Feasibility and Viability model. It brings together the three critical aspects of identifying an optimal solution for both customer and business.
From this model, we can draw connecting lines to Brian’s and Ries’s work. To determine viability, you can use Brian’s Growth models to identify the most impactful areas in your business to improve. To increase the feasibility of options, you can layer on Eric Ries’s principles in The Lean Startup (rooted in Steve Blank’s work) to reduce testing effort: (1) build, measure and learn and (2) move faster and faster.
The missing piece from New Age Growth
The hardest and arguably most crucial part of Growth is less covered even in Balfour’s articles: building solutions that genuinely address customer needs when signing up, onboarding, ordering or reactivating. Anyone can build models to identify that you improved a specific part of a funnel, you could increase revenue by millions. But just identifying valuable areas to improve does not mean you will be able to fix them.
The most significant challenge in my experience has been twofold:
Read on if you want a condensed view of frameworks that I’ve found most useful in the past year. Note, these approaches are not just useful for Product people, you can apply them all in the Growth world.
Note, the below assumes you have a strong product strategy in place and have identified the key objectives and metrics to move (not the subject of this blog).
1. Understanding users’ needs
Spending more time understanding and then defining an opportunity space has enormous payoffs. Some of our least successful tests attempted to solve relatively unevidenced customer needs, as shown by low percentages having the problem we were trying to solve or mixed receptions in user testing. No matter how creative or exciting our ideas-driven experiments were, they failed if they weren’t solving for genuine customer needs or opportunities to delight them.
Being more user-centric way works well in principle, but what does it mean in practice? Working out the best opportunity to work on is a three-step process:
Learning to do the above process well across Gousto has been incredibly impactful. We continue to push Growth and Product Managers to focus on problems first, then solutions. Indeed, at Asana, their Product Managers’ only role is to cultivate and curate problems for designers and teams to solve.
i. Map the opportunity space
Begin by creating mental frameworks to organise the problem/opportunity space that underpins your goal. Even if you have a limited early understanding, set out your thoughts early to guide discovery. Making your opportunity mapping clear also puts implicit views on the table early so others can input effectively.
A great model to use is Spotify’s Thoughtful Execution Framework (similar to Teresa Torres’ Opportunity Solution Tree), which enables you to visualise any space like a branching tree. Each Problem/Opportunity should be mutually exclusive and at similar altitudes (try to avoid ‘laundry lists’).
Below, you can see a hypothetical example of how we’ve applied the framework to the goal of improving early life retention.
Mental models, like the Thoughtful Execution Framework, are not one-size-fits-all. What’s important is your framework effectively supports decision-making in the problem space at hand. Sometimes, we approach opportunity spaces with a’ value-creation thesis’, which can be more useful if you’re using data science to improve customer experiences.
ii. Tighten problem/opportunity statements
Once you’ve created a framework for mapping opportunities, make sure you spend the right amount of time refining the problem and opportunity statements further. Having these ‘statements’ airtight it critical before generating hypotheses to solve them. Well-defined problems/opportunities:
When we’ve had to do mental contortions to define the user’s problem or opportunity, you likely don’t have a genuine need on your hands. When this happens, expand your perspective to find a problem that genuinely exists or question the broader goal. Most importantly, resist the forward momentum to come to a solution eventually.
You can see an example of a poor problem-statement below, which resulted in a recent unsuccessful test of ours at Gousto (that may feature later).
Customers rarely use categories on our menu, either because they don’t see them or they don’t see the value in them (especially when they begin scrolling)
Customers don’t use categories… So what?
We’ve found that spending large amounts of time defining opportunities up front pays back tenfold. Write down key questions and your working assumptions about them. Most importantly, share these with your peers and senior folk around the business to challenge and refine thinking.
iii. Use data & insights to identify the most valuable opportunities
Use insights and data to both validate and refine existing opportunities as well as adding new ones. How to identify the most impactful opportunity? Look for problems/opportunities that, if solved, create a meaningfully improved or new experience for a meaningful amount of people, that in turn generates business value. Have data-informed conversations to identify great areas — no need to use decision-making frameworks like ICE scoring.
Note, don’t get stuck at this step if you don’t feel you’ve ‘completed’ the tree with data. As with all these models, they are tools to make better decisions, rather than checklists to ensure you have every bit of information you need.
Beyond classic datapoints like past tests, behavioural data, and insights, two learnings changed the way I thought about sizing impact:
A) People already exhibiting desired behaviour in the wild
Sign you’re solving a real problem: if you can find a significant amount of people already solving for your solution using existing tools or demonstrating highly-similar behaviour in another industry.
For example, Netflix identified that offering download functionality for their shows would be valuable for international expansion when they found that 70% of Indian consumers used the same functionality for their YouTube videos! With patchy 4G and WiFi across the country, people would often use their work internet to download their favourite shows before commuting.
B) User testing, quantified at larger scales through insights surveys
Quantifying user testing learnings with surveys is an emerging best-practice research methodology at Gousto. We identify customer needs or pain points in customer interviews and then validate what % of customers have those needs in more extensive surveys. We’ve found this to be exceedingly helping in cutting through variability when interviewing 6–7 customers at a time.
For example, in user research, we learnt that many customers use ‘mental shortlists’ to choose recipes. I.e. they go through all recipes while noting recipes they like, then go back and choose their top four. We then quantified what % of customers did this using a survey that described the different ‘choosing methodologies’. We found that mental shortlists constituted only 24% of our user base, so deprioritised the opportunity. We believed it wouldn’t improve the experience for enough people to make up for the complexity it would introduce to everyone else.
2. Breaking the solutions to those problems down
Once you’ve got a clearly defined and valuable problem, you can start creating hypotheses and then solutions to solve it. Some say a well-defined problem statement is all that a great designer needs, but there are ways in which you can support them do their best work.
Identify riskiest assumptions
Riskiest Assumption Tests (RATs) validate your riskiest assumption first, so you can test your largest unknowns and understand the value of an opportunity as quickly as possible. We’ve found that identifying riskiest assumptions are incredibly helpful for clarifying our most important hypotheses and guiding where we aim to learn first.
Returning to our ‘relevance of early recipe recommendations’ opportunity from before, we can pull out assumptions that determine whether that opportunity will be valuable. Once you align on RAG assessments of the different assumptions, you can then investigate the riskiest assumption first.
When thinking about riskiest assumptions, you can use the IDEO framework to broaden thinking on what might change the value-unlock assessment of your opportunity. The assumptions above were largely in the Desirability Venn, but risks might instead be financial (e.g. cost to act on opportunity is too expensive) or that solutions are technically infeasible (e.g. hypothesis requires data science for personalisation and you have no data scientists).
And off the back of identifying that risky assumption, you can build out hypotheses to validate via tests, user research or data.
Apply lean testing methodologies
Look for the cheapest way to learn using Lean methodologies. In particular, consider whether you can test qualitatively without significant engineering effort before moving to build something. As Eric Ries puts it: create the minimal viable test required to get the insight you need to learn whether you’re creating something valuable for your customers.
Once solutions are on the table, re-check decisions back up the opportunity framework
You likely will have learnt tonnes as you moved through this process. Continually re-challenge your decisions and mental models on the most important places to invest time into as you discover more.
Again using the previous example, let’s say we discover that it’s incredibly difficult to collect valuable data to inform recipe recommendations for early customers. We might then choose to focus on a different opportunity than ‘improve the relevance of early recipe recommendations’, as a core assumption on why that opportunity valuable has been proven untrue.
Two big final learnings
I’ve found the processes and tools above invaluable to make better decisions over the past six months. But my ‘weightings’ and ‘senses’ to make the right prioritisation decisions are still in their nascent form. But after three years of experimentation in Growth and Product teams, I’ve found two themes have repeatedly emerged. And of course, Intercom described them first.
1. When increasing complexity, make sure your new experience provides profound benefits
The more complexity you introduce to an experience, the more you’ll need to unlock outsized benefit for customers. If two designs perform similarly in user testing, but one is more complicated to use, it is highly likely to fail when you release it. Be careful when building complexity into your product! These sorts of product changes are hazardous.
2. Avoid grand theories. Embrace the idea that the world is messy
Customers using features in different ways than you want them to doesn’t mean your system is suboptimal. And just because you’d like to explain your work simply doesn’t mean you should build simply. Avoid ‘grand theories’ of categorising customer behaviour, as they lead to non-customer-centric work.
Grand theory is a term coined by the American sociologist C. Wright Mills to refer to the form of highly abstract theorising in which the formal organisation and arrangement of concepts takes priority over understanding the social reality.
Intercom uses a great example with email and how Gmail works within human’s messy constraints:
We email the same groups of people over and over again, rebuilding the same list manually every single time. But this is the messy reality. The cognitive effort to add in people’s email addresses is low. Doing it over and over again makes sense.
The tip of the iceberg
These mental models have helped me navigate changing people’s lives for the better. I’ve found that the main principle that underpins them — real problem-first thinking — has massively matured my approach to building successful customer experience changing in Growth and Product.
However, these models are just the tip of the iceberg. I haven’t discussed how this links up to overarching Product/Growth strategy, which is critical when taking a long-term view of value creation. I haven’t covered second-order consequences of experiments, ranging from how you can negatively affect future product changes to how new builds might lead to ever-increasing Effort estimates (there are no small changes).
I’ve just begun scratching the surface of what it takes to develop value-creating customer experiences. But with these new and improved mental models, I feel much more equipped to solve the challenges ahead. If you have any models that positively impacted your own thinking, please share with us.
I’d like to give a special shoutout to Barry Pace, Sian King and Tereza Sustrova. Their guidance since I’ve joined Product has elevated my thinking so much. They introduced to me all of the frameworks above!
