« Fall in love with the process, and the results will come. » — Eric Thomas
Hard truth: Growth Hacking is not somehow magical. So, if you imagine that just tweaking the color of your current piggy pink call-to-action for a fantastic barbie pink call-to-action will bring your company a crazy 30% uplift conversion rate, you better think twice about it. Did you follow a strict process to research, prioritize, design, test and then analyze this experiment accurately? If not, I’m sorry to announce you that this experiment is not that likely to attract new customers magically.
Be reassured. Whether it is Airbnb, Uber, Booking or most high-growth companies, they didn’t find out their remarkable growth hack thanks to some random idea. Discovering “silver bullets” was the result of hard work and a strict process to get something worthwhile to focus on.
A “Magic Growth Hack” working for any business doesn’t exist outside of sales letter hype and over-the-top sales pitch.
Besides, if I agree that all the trendy “hacky” techniques that we heard about -scraping, black hat marketing, automation, viral loops, etc. – can be super powerful to push a company’s growth, I feel that we may have missed the most crucial point. Beyond the buzzword, Growth Hacking (GH) is a step-by-step process.
Enough small talk, in this article, I’ll dig into the Growth Hacking Methodology I’ve been trained at during the 3-month in-house program of the Growth Tribe Academy in Amsterdam. Note that I will make a particular focus on the Rapid Experimentation process that a lot of successful tech startups have used to scale their business. Hope that knowledge can help some of you!
PS1: Please note that the following paragraphs are my interpretation of the Growth Tribe learnings and my own experience as an entrepreneur.
PS2: If you are expecting a fancy version of a To-Do list, you won’t find it here.
STEP 1: Before applying the GH Methodology blindly, make sure that you reach the Product Market Fit
“Startups require a solid foundation of product/market fit before progressing up the pyramid and scaling the business.”
1 — A fast driven growth process is useless if you haven’t achieved the product market fit (PMF).
If your product or service doesn’t have PMF, a growth team won’t be of much help… It doesn’t make sense to try scaling a product if you’ve have not found that your users like your product and keep using it. Period. At that stage, you should obsessively focus on delivering the right product to the right market. Adopt the lean startup mindset and methodology and figure out who your (possible) customers are, understand their needs and build a fantastic product that will transform them into your first ambassadors. Yes, it can take some time and a lot of adjustments, but you probably agree that trying to scale a shitty product is even worse for your time management.
2 — But how do know you have reached this holy grail PMF ?
- The Sean Ellis “Very Disappointed” Method
1. You send a questionnaire mostly to your active users
2. You ask them: “how would you feel if you could no longer use our product tomorrow?” and they can choose between “very disappointed” “somewhat disappointed” “not disappointed” or “I don’t use this anymore”
3. After segmenting your responses afterward, if at least 40% of the respondents are “very disappointed”, you reach the threshold for PMF.
Nevertheless, for some products, this could be more difficult to apply, so that’s why some Growth Hackers prefer using the Brian Balfour approach to state that you have PMF.
- The Brian Balfour Trifecta Approach
1. Your product has some form of strong organic top-line growth → products with PMF usually get recommended and benefit from organic growth.
2. Your product has good retention → you retain your users, and your churn (if applicable) is low. For instance, Snapchat used to have 50% of daily active users.
3. Your users are using your product correctly → you have to be honest with yourself about what is a meaningful use of your app or product. If you have an online job board for startups and each of the startups posts one job every three months, that’s probably ok! But, for a tool like Slack, it’s perhaps sending a message twenty times per day.
3 — Is Growth Hacking useless before PMF then?
Well, if a growth team experimenting established a fast-paced process doesn’t make sense before the PMF, we can still identify an overlap (cf. slide below) between the Problem/Solution Fit phase and the PMF phase where it makes sense to use Growth Hacking (GH). During that time, GH should be used to identify growth opportunities in pre-product market fit products by looking at data and usage patterns. The Bullseye Template developed in the book Traction can be used at that moment. Note that it will also be helpful to optimize the growth of products that already have product/market fit at a later stage.
4 — Find the sweet spot where it makes sense to apply the Growth Process Methodology
For Brian Balfour, « a startup goes through three phases of growing.”.
As you can see, focusing on the right strategy at the right moment could help your startup reach the next level. If you want more insights, take a look at the article here. Obviously, every business will have a slightly different moment when it makes sense to try getting to the next phase. Please note that since we’re in the “Transition Phase” with my Startup Les Sherpas, I could be a little bit biased in the rest of the article.
PS: Les Sherpas is an ed-tech startup which has joined the first batch of the Founders Program by Station F, a giant French Incubator with more than 1000 startups directly on-site. Besides, we were followed by the Station F documentary series during three months (four startups selected out of 1000). One year after the first release of our first Minimum Viable Product, our online tutoring service for high school and universities has just reached the 10 000 hours given on the platform. We now want to grow faster and reach the 30 000 hours given for next scholar year!
STEP 2: The five prerequisites to implement the growth hacking process in your organization
PS: I highly recommend this youtube video from Growth Tribe with David Arnoux where I use most of the knowledge process to write the following.
Once you have your product-market fit, it’s only the beginning of a long path before implementing an efficient Growth Process and achieve your goals. Indeed, it will require yourself to understand and communicate correctly within your team what are the critical components of your business. Why? You’ll need everyone involved in this growth process to make it a success.
To do so, define clearly the first five prerequisites before going in the rapid experimentation G.R.O.W.S phase → the Business Model Canvas, the Value Proposition Canvas, the Personas, the Pirate Funnels and, the OMTM.
Don’t you know what these acronyms are? Let’s break them down together!
1 — The Business Model Canvas
To begin with, let’s take the Wikipedia definition, “the Business Model Canvas is a strategic management and lean startup template for developing new or documenting existing business models. It is a visual chart with elements describing a firm’s or product’s value proposition, infrastructure, customers, and finances.”
I think you got it. It describes visually on one page what your company is doing. Pretty cool to have a global overview, onboard new people and get everyone with a basic knowledge of what the f*** you’ve been doing for the past few years (months?). It’s such a typical Business School stuff now that I think pretty much everyone knows it, so I will not bore you with all the steps to create it. But, if you still haven’t done it, take a pen and begin filling in the model below.
2 — The Value Proposition Canvas
Integrated into the Business Model Canvas, it is interesting to dig deeper into this element because this is where the heart and value of your product lie. Thanks to this canvas, you can picture who your customer is and how your value proposition is shaped to make them loved your product. If you’re one of the founders, it may be evident to you but are you sure that all the rest of your team get it? Anyway, it’s always better to make sure that everyone is on the same line.
Besides, as Phil Morle explained it here, the next step once you have filled in the document is “to figure out if this is all just a dream you have for the perfect value proposition, or whether it is anchored on how your customers behave.” (the second solution is the one you’re looking for ;))
Image Credits: Strategyzer
3 — The Personas
A persona is a fictional profile that represents a brand audience. Developing a persona helps you to bring the audience to life and target it more accurately.
Why define your personas?
The objective is to succeed in putting yourself in your customers’ shoes, try to make a mental representation of their way of thinking and behaving and finally find a solution to their problem. Once done, it will be much easier to sell your products because you’ll know what they’re looking for.
As an example, your strategy on social networks will not be the same if you talk to Alicia, freshly employed as a human resources junior in a tech startup or Philippe, the sales manager of an established company.
How to build your personas? A few rules you should follow.
- Limit yourself to the three most important fictional personas. I know that you want to target the most considerable number of people with your product, but if you laser-target your personas with a custom strategy, it will be much more efficient. Once you have reached the threshold you were looking for, you can expand gradually.
- Follow a rigorous process of precise questions to define your typical personas better. Beside the classic name, age, job role and some demographics, you should also focus on more thorough questions. Which aspects of their personality are essential? (You can use the psychology Big Five Model or MBTI Model). Where do they hang out? What is their behavior on the internet? What does their day look like? What are their needs and goals? How are they trying to solve their issues? What are they using instead of your product? Etc.
- Validate your assumptions on your personas by testing how they respond to messaging design and persuasion. If it’s not conclusive, try to refine your persona type and validate your assumptions until you get conclusive results.
Images Credits: Xtensio
4 — The Pirate Funnels
The Pirate Funnels is a helpful customer-lifecycle framework invented by Dave McClure that you can use to map out and optimize your marketing funnel. The main idea is that for every product, users go through a similar journey (AAARRR). If you’re not familiar with it, let me present yourself these six steps and then map it out for your own business.
Three things to keep in mind when you’re designing your pirates funnel:
- You should define precisely the actions that customers take in part of their journey. And next to it, you should indicate which metrics are showing the customer interest (soft conversions) or intent (hard conversions). As an example, Awareness could be defined by the traffic to your website, but the metrics you should focus on to measure it is the monthly unique visitors. Or if you’re on the Retention part, you could define it by the number of people who invite friends and measure it by the viral coefficient.
- The funnel is flexible. The funnel can be in a different order depending on the type of business you are in. The one I’ve put above is more “e-commerce stylish”. But for a SaaS business or any subscription-based model, you would be more likely to have this kind of flow: awareness → acquisition → activation → retention → referral → revenue. Anyway, take a deep breathe and make the funnel which is corresponding to your business.
- Don’t forget to segment the funnels by customer segments. It’s important to note that your product or service is likely to have multiple funnels. As an example, a company like Hotjar has different definitions of each step whether it is a Small startup or an Agency. Or, if we take Airbnb (marketplace model), the hosts (supply) and the guests (demand) don’t have the same customer journey.
5— The OMTM
At any given time, there’s one metric you should care about above all else. – Lean Analytics, Croll & Yoskovitz
The One Metric That Matters (OMTM) is the one metric you’re completely focused on above everything else for your current developing stage. It makes it easier to design and have meaningful experiments since you know what to focus on.
Image Credits: Growth Tribe
The long-term focus is your North Star Metric (NSM). This is the direction you go towards, and all decisions and OMTM’s along the road should help you get to that NSM. Imagine being on the road towards that north star, but a mountain suddenly stands in your way. You have to come up with a different short-term goal of overcoming that mountain. This could be your OMTM for a month for example.
Two things to keep in mind regarding this topic:
- You don’t focus on the OMTM exclusively forever. See it as a focus for a predefined period. This will change once you get to the target metric/date. Then you decide whether it’s still the most important metric to focus on. As an example, at Les Sherpas, my startup, our current OMTM is the conversion rate after the free trial, and our NSM is the number of hours given on the platform per year. But our OMTM used to be retention when we were focusing on achieving product-market fit.
- You’ll always track and review multiple numbers. These are your key performance indicators (KPIs), which you’ll track and report every day to get more data-driven decisions even though the most important one is the OMTM. Other metrics will be stored away for future use, such as when it’s time to tell the company history to an investor or to make an infographic. Setting up and managing instrumentation is relatively easy these days with tools like Geckoboard, Mixpanel, Kissmetrics, etc. Anyway, don’t let your ability to track so many things distract you. So, yes, why not capturing everything, but above all, focus on what’s important.
STEP 3: Implement the Rapid Experimentation Process (G.R.O.W.S)
Now that we have the five prerequisites done (STEP 2), let me take you through a step by step guide on how to implement the G.R.O.W.S loop within your organization. It doesn’t matter whether you apply the Sean Ellis or Growth Tribe approach (G.R.O.W.S), the Growth Marketing Process will ensure you can experiment rapidly and find huge wins for your business.
1 — Gather Ideas (G of G.R.O.W.S)
During that phase, your team should generate hypotheses for growth experiments that are likely to improve your OMTM.
How to come up with impactful ideas?
- Observe your Market. Best Practices, Ebooks, Trends, Articles, Competitor Piggybacking, etc.
- Do some Research (Quantitative and Qualitative). If Quantitative Data (Analytics, Heatmaps…) tells you what, where and how much, Qualitative Data (Usability Testing, Surveys, Customer Feedback…) often shows you the ‘why’ your customers are behaving like this or this. Qualitative will enable you to peek inside the mind of the people you’re trying to sell to.
How to insert new ideas in the pipeline?
“The Growth Lead has to set up a project management system to coordinate easily the submission and management of ideas, as well as the tracking and reporting of results. The more ideas that go into your pipeline, the better your chances of finding winners that spur growth.” – Sean Ellis, Hacking Growth
First, if you don’t know how to set up a project management system (CRO tool) for your team, you can use the Sean Ellis tool, Effective Experiments, or even some Airtable Free Templates (I’ve designed my own freely on this tool).
Once this is done, focus on the formatting part of each idea which should be the same to rank them after easily. You can follow this flow:
– Idea Name: Breve and Clear → 50 characters max
– Idea Description: Who, What, Where, When, Why, How of the Idea.
– Hypothesis: Make your assumption simple, unambiguous and testable. It’s a simple proposition of expected cause and effect. One sentence I really like to make the hypothesis is the one from the Dutch agency Online Dialogue: If we “apply this change”, these “metrics change” for “this group of users”, because of this “behavioral reason”.
– Metrics to be measured: Put the Hard and Soft Data that is going to be measured and don’t forget to put the minimum success criteria for this experiment
2 — Rank Ideas (R of G.R.O.W.S)
So, right now, you should have an idea backlog full of amazing ideas.
Nevertheless, before an idea is considered by the team during the “sprint meeting” (the day you are reviewing the ideas and choose the experiments you gonna launch), the person submitting the idea should always score it in the project management software. Even though it’s quite harsh at the beginning, you’ll see that it gets easier with practice since you get used to the industry benchmark and previous experiment findings.
You’re spoiled for choice since a lot of models already turn out quite efficient. In my startup, we usually use the ICE model (Sean Ellis approach), but you could also use the BRASS and PIES model of Growth Tribe (cf. this Youtube Video). Your call!
Be aware that very impactful ideas (high Impact Score) often require more work, and smooth implementation is essential. So it’s always good to have some easier “low-hanging fruits” tests to run without having weeks to prepare for launch. Additionally, don’t spend too much time overthinking the scoring since the Growth lead can always adjust it if he sees some issues that the submitter didn’t notice. The goal here is to keep the team moving.
Even after going through the scoring process and narrowing down a list of experiments you know you want to try, you’ll likely still end up with more ideas than you can test during the coming week. So, the trick here is:
- Ask every person to choose only two or three ideas as candidates for this week experiments.
- These “most promising” ideas will be presented and reviewed at the growth sprint meeting (every week or every two weeks usually), where the experiments to be run will be chosen.
- The growth team will score the ideas collaboratively according to its assumptions and select the experiments to be run with the highest ICE score. As an example, we often take two of them in my startup since we still are limited regarding resources, but it can vary from one week to another.
3— Outline Experiments (O of G.R.O.W.S)
Let’s assume that your Growth Meeting is done, and you have two experiments you’re going to launch this week. How should you organize your team to design this experiment?
“Done is better than perfect.” – Sheryl Sandberg
- Make it quick and dirty. Please keep in mind that at that stage, you should focus on designing “quick and dirty” experiments more than perfect ones. The goal here is to learn if the experiment could be a success in a short amount of time (the velocity is critical in this growth process). If it appears to be promising after the experiment just ended, you’ll able to double down on it and improve the results all along. If not, you won’t have spent too much time overthinking a d**** solution.
- Get your team to collaborate and execute fast. As you grow bigger (more than one person in the Growth Team, woot-woot), you could organize yourself as the following. A Growth Lead that is process driven, a Developer that builds, breaks and fixes fast all the time, a UX/UI Designer that is ok with “not that pretty” design and a Data Analyst that can process your data and make conclusions out of it. Of course, as you grow, the Growth Team can even welcome more people (marketers, other developers).
4 — Work, Work, Work (W of G.R.O.W.S)
“Sometimes you gotta run before you can walk”
— Tony stark, alias Iron Man
During that phase, the design of your experiments is finished. So… let’s get the shit done. Make sure that you are doing the 20% of work that produces 80% of the results! You’ll have time to adjust this experiment and double down on it if it’s promising. As the quote says, “Work Smarter, Not Harder”.
I decide not to focus too much on this part since I feel that every company will have a slightly different way to organize itself to get the things run. I would only recommend focussing on applying the principles of Agile Scrum for Marketing. If you’re not familiar with it, I highly recommend this article from ConversionXL.
5 — Study Data (S of G.R.O.W.S)
“Sometimes, things may not go your way, but the effort should be there every single night.” – Michael Jordan
Let’s be crystal clear. This phase is one of the most difficult ones for two main reasons.
- The chances to get no (or even worse) results are quite high. Don’t get depressed, it’s normal. Before finding the sweet spot that gonna fuel your growth, you’ll probably end up launching many failed experiments.
- The analysis part of an experiment is not easy. This part should be done carefully by the Growth Lead with the Data Scientist (if there’s one) since you could have false positives or wrong assumptions.
Knowing that, you usually go through three phases when it comes to analyze the results of an experiment:
- Your experiment is over (after one to three weeks usually to avoid seasonality effect) and is awaiting for evaluation.
- The Growth Lead analyzes the results following a template he has created to evaluate each experiment. (Cf. left image)
- The Growth Lead puts the experiment in one of the following cases for the next steps: Inconclusive, Failure not valuable to Tweak, Failure valuable to Tweak, Success & continue exploring, Success & scale.
Well, once you have finished this phase, you can go back into the G.R.O.W.S loop and make experiments again and again. Don’t forget to use this analyze phase to end up with some actionable learnings for your next experiments. To do so, you can think about segmenting your audience (by country, by mobile vs. desktop, by number of products in the cart, etc.). Even if the chances to have a false positive are higher, it’s still interesting to have this data, notably for launching new experiments.
- Across all variations, the feature module below the fold increase conv rate for new users.
- The presence of the new theme produced a 1,4% uplift in CTR from the homepage for desktop users.
Some thoughts about A/B Testing
You’ll be able to do it only if you have enough traffic and conversions. Ideally, you should always have 1000 conversions on the thing you want to increase with your A/B test. Before reaching this “optimization phase”, you’re usually in the lambs of early-stage startups trying to get their first customers, so it’s all about taking risks. Don’t focus too much on small fine-tuned A/B tests and go for big changes. If you’re in that case, do big tests (a lot of changes at the same time) with big impact such as multivariate tests.
How to be sure that your A/B Test is correctly analyzed?
A hard question since I’m not a specialist but here are some learnings I discovered. First of all, you have to decide which statistical approach you’re going to embrace to declare your variation as a winner. “There’s a philosophical statistics debate in the optimization world: Bayesian vs Frequentist.” (ConversionXL)
When you’re going to select one, you should have three things in mind:
– The risk you’re willing to take
- Frequentist Approach
– Risk: In the research domain and an optimal world, I would say that this method is more likely to bring you nearly “perfect” statistical results. In this one, you will declare the variation (vs. the controlled group) as a winner only if it reaches the statistical significance level you were looking for (95% or 99% usually).
– Velocity: Nevertheless, the rapid experimentation process can slow down a little bit if you’re using this approach since you need to reach these high thresholds before implementing an experiment declared as a winner.
– Communication: Statistical significance, p-value, power, beta… Hum, all these “nice” words don’t help you when it comes to explaining your experiments to the rest of your team.
- Bayesian Approach
– Risk: We’re in a business world with a lot of variables to take into account that could also impact your experiment (special promotion, holidays, weather, etc.). Plus, if you’re still in the “transition phase” and not really on a perfect fine-tuning optimization model, reaching 99% significance may not be your primary focus. In that case, you could define thresholds such as: 80% probability → the variation is declared as a winner. 75% and below → Retest. 85% → Implement it and use for the next experiments. It’s all about the risks you’re willing to take to make your business thrive. While your company is growing, you can make these thresholds evolve.
– Velocity: The risk is higher but the velocity is usually also higher. Since you don’t need super high thresholds significance, you can design even more quick and dirty solutions and double down on it if it’s a success.
– Communication: It’s much more transparent e.g. “The chance of variation B outperforming controlled group A is 84,4%”.
Additional step: the Playbook
You should always have a knowledge base where you are storing all the experiments you did in the past with the assumptions and the results you had. It will avoid launching the same failed experiment at multiples times and will facilitate the onboarding of new employees in the Growth Team, because they can see the process you’re doing. Plus, it can be a nice to have for your stakeholders to follow.
Besides, many highly successful teams send out regular communication within the company to keep employees to speed on the growth process. So, consider creating a channel on Slack dedicated to the sharing of test results and discussion about them.
You got the Rapid Experimentation Process? Let’s get to work then!
Hope you enjoyed this article. Thanks for reading! 🙂