[5/10] How to drive user growth
Tips on digital ads, influencers, referral, and content marketing
The previous essay broke down the problem of growth into User Growth and Value Growth. It then listed the 3 different ‘growth lanes’ a business can take to pursue User Growth, along with the criteria that makes a particular lane suitable or unsuitable for a specific business.
In this essay, I will be going into the most common channels in the 3 different lanes of User Growth, their fundamental concepts, and the best practices. Let’s start with the most common one: digital advertising channels.
Digital advertising: how can your creative strategy be the real force multiplier for your campaigns?
Suppose you are working on a consumer brand with a fairly common ideal customer profile, and you have a cost structure that allows you to have a certain % of revenue as marketing costs. In this case, digital advertising seems like a good growth lane for your business to pursue. But, how do you go about it?
First, which platform are we going to advertise on? We, as consumers, spend a lot of time on content and utility apps. Many of them monetise our attention by serving us ads. A few of them are big enough to build their own ad platform to get businesses to advertise on their platform, such as Twitter, LinkedIn, Snapchat, Quora, Pinterest, etc. However, the big ad platforms on which the majority of the advertisement spends happens are just a few: Google, Meta, and Amazon. And that probably is where you would start too.
The other good news for you as is that, despite the wide variety of platforms and their respective nuances, they all typically require 3 components of information to serve an advertising campaign:
Business Objective
Customer Profile
Communication
Let’s go through them, and seek to understand what makes for a great campaign from the respective lens of each of these.
Business Objective:
To start with, we have to decide what is the objective of the campaign. It could be one of these:
Making more people aware of the product/service (top-of-funnel)
Making people who are already aware of the product/service, consider it (middle-of-funnel)
Making people who are already aware of and/or are considering it, purchase it (bottom-of-funnel)
How much are you willing to pay for every action for the given objective
Top-of-funnel: how much would you pay for reaching a new person or getting one more person to recall your ad (and your brand by proxy) i.e. cost per reach or cost per ad recall
Middle-of-funnel: how much would you pay for every person coming to your website (cost per click) or installing your app (cost per install), or for every person to complete a slightly more immersive action on the ad platform (filling a form i.e. cost per lead, or watching a complete video i.e. cost per video view)
Bottom-of-funnel: how much are you willing to pay for a purchase (cost per purchase), or a non-purchase action on your platform that is a leading indicator of high intent customers e.g. completing a loan application form on your app.
How much are you willing to pay overall for the campaign, and what should be the duration of the campaign, and details like that.
Putting it all together, the first level of information you would give to an ad platform would look something like this: spend about ₹100k over the next 30 days at a cost of ₹500 per purchase.
Where would this information come from? For the bottom of the funnel campaign, it would come from the growth model we previously built. We would know the number of new customers we are going to acquire, the cost associated with it, and the time period. But how do we decide how much to spend on the other parts of the funnel? And, wait, why should a business spend money on any campaign that’s not contributing to a bottom-of-funnel action such as purchases?
That would be because we, as consumers, do not make purchases from a new brand the very first time we see it. A brand is a shorthand for a certain set of attributes in a category, and it would take a few repetitions of the brand for our brains to remember the name and build the association between the name and the category and the differentiating attributes. Those repetitions can be done better through top-of-funnel and middle-of-funnel campaigns, so that by the time those potential customers are targeted through the bottom-of-funnel campaign, they have been primed.
So, how do we plan how much to spend on the top-of-funnel and middle-of-funnel campaigns and how much to spend on the bottom-of-funnel campaign? There are two ways to go about it. One: you can use a heuristic such as: for every ₹10 you spend on a bottom-of-funnel campaign, spend ₹1 building the top-of-funnel. The other way is to estimate the number bottom-up.
Let’s say we built the growth model, and from there we have arrived at requiring 10,000 new customers at a CAC of ₹500 in Jan’24. Once you gather the benchmarks for different ratios at different stages of the funnel, you can estimate both the bottom-of-funnel and top-of-funnel numbers:
Customer profile:
We buy products and services all the time as consumers. But we are actively searching for something only some of the time (e.g. searching for flight ticket prices on Google), while the rest of the time we are passively browsing the internet. The ad platforms will, therefore, have 2 kinds of inventory or properties to serve us ads on: pull properties or push properties.
Pull properties are where customers are actively looking for the product/service you are selling e.g. Google search, Amazon search, Apple app store search, etc. Push properties are platforms where customers are, at that moment, not looking for the product/service. For example, when we are browsing Instagram or Facebook in case of Meta Ads, watching YouTube in case of Google Ads, and watching Prime Video or browsing Amazon homepage in case of Amazon Ads.
So, when we are advertising on pull properties, we would define the intent rather than customers themselves. E.g. people who are searching for ‘red leather jackets’. Whereas, when we are advertising on push properties, we have to tell the ad platform the attributes of the ideal customer that our business is looking for.
We can define the ideal customers through their demographic and behavioral attributes e.g. show this campaign to 18-35 year old women living in the city of Mumbai. Or, we can rely on signals that the platforms have collected on their users based on their content consumption or browsing patterns e.g. serve this campaign to people who are in the market to buy a car currently; serve this campaign to people who have yoga as an interest area. The customers can also be defined by the social graph i.e. their similarity to other customers e.g. serve this campaign to people who are similar to my top 1,000 customers.
While this might seem like the most important part of digital advertising, it actually is something that is either being completely done away with by the ad platforms (Universal App Campaign by Google doesn’t have this option at all, for example), or being discouraged implicitly by them (Meta lets you know that not letting the platform go beyond the audience persona defined by you will come at a suboptimal cost). Now, this might seem very counterintuitive at first. How will you target the right customers, you might ask the ad platform, if you don’t even ask me who they are?
The idea here is that if you have fed the platform the right campaign objective details, the ad platform can figure out the ideal customer attributes by learning through real data on which customer attributes are converting for your objective at your desired cost.
The job of the advertiser becomes increasingly limited to: 1. figuring and instrumenting the right business objectives and 2. having the right communication strategy. And, this seems almost marketing blasphemy: are we as advertisers not supposed to know our customers at all? Until you realize the importance of knowing the customer when it comes to designing the communication, and the magnitude of differentiation it can bring.
Communication:
Building a piece of communication has 3 parts: designing the brand-customer relation, crafting the message itself, and making it suitable to the medium. And all the 3 parts have two dimensions: the no-so-visible core part and the visible part.
Let’s start with the core of brand-customer relation: positioning. Getting to a clear positioning requires answering the question: what about your product/service does the customer primarily care about and what truly sets it apart from the competing products. This answer can come from knowing your customer – first through qualitative research such as interviews to come up with a long list of potential options and then through quantitative research such as surveys to rank them. The outcome of the exercise should be one sharp use case that your ideal customer profile cares most about, even if there are a few secondary good-to-have use cases.
An example from the early days of Meesho’s social commerce phase: based on our primary research, we realized that the primary use case differed quite a bit by customer profile. And, more importantly, the primary benefit of our app that we had been advertising so far was not the primary use case for our ideal customer profile. A quick change to align the positioning with this discovery led to massive gains for us in the long run.
While the positioning is the core of the brand-customer relationship, the brand’s personality is the visible part. The latter often gets more focus, but the two are different. For example, one news channel can differentiate from others by positioning itself to be the fastest, another can differentiate and position itself as the most analytical. This is, however, different from the brand’s personality: friendly versus avuncular, funny versus serious, etc. Sometimes the personality itself can be a differentiator (especially in the case of media brands) but, in most cases, and especially for most physical products, you have to work both out independently and not confuse them with each other.
Similarly for the creative itself, the core is the fundamental insight – about how the customer can benefit from the product in most cases, but just about the ideal customer's archetype in some cases – that we are trying to deliver, while the visible parts are their creative manifestations: the copy and the image/video. The core can be the persuasion technique at play, the emotion we are trying to evoke, and so on, while the visible parts can be the duration, the color schema, the font size, etc. Both have to be cared about and focussed on separately, but, most importantly, they shouldn’t be confused with each other.
Coming to the medium through which the creative is getting delivered, it might seem, at first glance, that it is just one of the attributes of the creative itself. However, it is of significant importance to deserve a separate dimension altogether (hence the adage: the medium is the message). Since how we consume information depends greatly on the medium itself, the rules of the game change radically from one medium to another: what makes for a great TV ad doesn’t necessarily make for a great Instagram Reels ad; what makes for a great billboard doesn’t make for a great WhatsApp message; and so on.
Now, with the three aspects of communication design listed, coming to why it is the biggest differentiator for the success of a campaign. It comes from my thumb rule: in any funnel, the ratio you can most easily influence, in the short run, is the top most one. And, as per that rule, the biggest needle-moving ratio in any campaign is the click-through rate on the ad at top-of-funnel. And yet the rigor in testing and maximizing click-through rates is often much lower than for conversion rates and retention rates.
A simple build-measure-learn attitude, along with a basic analytical framework, would go a long way in making the entire funnel more efficient. While the positioning and personality won’t change much from one campaign to another, and keeping medium’s best principles in mind, we can keep learning on the creative aspects with a basic classification system such as the one below:
Partnerships: how a combination of segment & funnel thinking can crack this growth lane
The biggest difference between digital advertising as a growth lane and partnerships as a growth lane, is that in the latter the important aspects of the ad platform’s job falls on the advertiser. Let me explain.
You see, the ad platforms are really great at understanding the objective you have chosen, pursuing the relevant audience accordingly, as well as learning during the campaign whether or not they are meeting the criteria of cost per desired objective that the advertiser had set, and accordingly course-correct. So, the job of the advertiser is all about:
1. Selecting the right business objective on the ad platform,
2. Instrumenting the feedback loop mechanism, and
3. Deploying the right creative strategy.
However, if you are working with a partner (an individual or another business), there is no feedback loop, built from machine learning algorithms, available with the partners to learn or course-correct in case the objectives are not being met. The other important function that ad platforms perform is that they do the work of selecting the right audience for you based on your inputs and the campaign learnings. Partnerships are, on the other hand, like a buffet system. You have to either buy all their audience or none of them. So, the only real lever available to you is not segmenting the audience of partners but segmenting the partners themselves.
Influencer or Affiliate Partner Marketing
While it might seem like all bad news so far, it really isn’t. While the difficulty of this growth lane will be higher compared to running campaigns on ad platforms, you can look at the bright side: if you do make it a sustainable channel, it will be a competitive advantage for you. Also, while the learnings on other ad platforms are privy to their machine learning models, for this channel, while the models might be simpler, they will at least be transparent to you.
In the case of influencer partnerships, for example, the learnings that you build on your side are:
1. What type of partners are working out for the given business objectives and what kind are not?
2. What kind of creative strategy is working for the respective partners?
The first one is, once again, a classification problem. If you have worked with 10 influencers, for example, and can define them along 5 major attributes, you can find patterns in the attributes. These patterns can then be used to predict which among 1000 other influencers that you haven’t worked with will work out for you for your objective and cost per objective. How do we find these patterns?
The major attributes at the first level can be the platform and the content niche they operate in. For example, if all the beauty influencers you have worked with so far on Instagram have worked out for you ROI-wise so far, while all wellness influencers haven’t, if you have to work with one more influencer, you would rather go with a beauty influencer than a wellness influencer.
However, not every influencer in a given platform and content niche are the same. So, the influencer specific attributes would be their reach (volume metric being the average reach for their content and the quality metric being % of the reach in your ideal customer profile) and their engagement (the absolute metric being the average engagement for their content and the quality metric being the depth of engagement).
Once you bucket all the influencers you have worked with and put the classifying attributes as independent variables and the campaign outcome as the dependent variable, patterns will emerge.
User Referral
Just like partnering with individuals and businesses to reach your customers indirectly, you can partner with your existing customers to reach other potential customers. The problem with building a User referral program is also similar: you again have to do the job of the ad platform. That is, the task of ensuring the right objective and cost per objective is being achieved, and course-correct if it is not.
At the top-level, the problem is simple enough:
1. Design an incentives program, i.e. why should someone refer their contacts, and
2. Distribute the incentives program to the right people at the right time.
Designing the incentive program has two potential paths: monetary and non-monetary benefits. Monetary benefits, such as receiving money for every successful referral, or getting a discount on the next order can be straightforward and would apply to most cases. Non-monetary benefits, such as getting exclusive access (currently being employed for Notion AI waitlist, for example), are less common, but can be powerful.
Designing monetary benefits will have two parts:
1. Which objective are you ready to pay for, and
2. How much?
Let’s consider the above two questions, and the potential pitfalls of wrong answers.
If you give monetary benefits for a middle-of-funnel objective – such as a sign-up or a lead, in case the process of transaction is not straightforward, or app install in case of B2C referral – you are creating an incentive for the referrer as the agent that might not align with your business’s objectives as principal: what if none of the leads or installs convert, and you have to pay for all of them. So, how do we solve it?
We can create a quality metric on which payments are incumbent e.g. payment per qualified lead. Or we can decide to pay only for bottom-of-funnel action. That should solve it, right? Not really. Even moving the incentive to a bottom-of-funnel metric such as purchase, can create perverse incentives. For example, your CAC on Meta Ads might be 500 and LTV might be 1000. With these numbers in mind, you design a referral program with a CAC of 400. That is, the person you referred to gets 200 as discount on their first order, and the referrer gets 200 as discount on their next order or as cashback. However, this makes sense only if the LTV is 1000. What if the LTV of customers coming via referrals turns out to be 600. In that case, the CAC will have to be 100 to have the same net value (LTV minus CAC).
A user referral program requires continuous monitoring not just for outright fraud — which different applications can help detect — but to detect the unintended outcomes of short-term incentives being introduced, and to accordingly course-correct.
Now that the incentive structure for the User Referral program has been designed, how do we use it as a growth channel? While, the design of the incentive program drives the core conversion ratio for the program – that is, if 100 people refer using this program, how many new users will join the platform – it is the distribution of the program which can often make a difference. Thinking of this from a funnel point of view:
Awareness: How many of your customers/potential referrers are aware of your referral program
Adoption: How many of those are going to refer
Conversion: How many of the referred are going to convert
Frequency: How often will they refer
As per the thumb rule on ‘first moving the needle on the top-of-funnel ratio’, let’s first seek to maximize the awareness of the referral program. The levers can be the real estate on your platform being provided to the referral program (pull properties such as in-app banners), and campaigns being sent out to the users as part of their lifecycle journey (push properties).
While the other parts of the funnel might seem to be a function of just the incentive design, the incentive design often provides for the ‘why’ but often not the ‘why now’. Having time-bound, one-off incentives (for example, refer a friend today and get surge benefits or stand to win a prize, over and above the regular benefits) can solve the ‘why now’ problem.
Organic growth: how to approach it in a structured manner
The last growth lane is different from the previous two, in that while they depend on you proactively approaching the prospective customer, this lane depends on the customer discovering you organically. This discovery is often through some content that you have put out there which is useful for the prospective customer. The channel would then be referred to as content marketing. But it can also be through some other intellectual property, such as a useful tool. The channel is then referred to as ‘engineering as marketing’ by Gabriel Weinberg in his book ‘Traction’.
Focussing on the content marketing channel, there are 2 ways we can bucket this content:
Is the content being generated by humans? Or is it being programmatically generated?
Is the content editorial in nature but low in volume? Or is it being generated in high volume?
An example of editorial content being generated at a low volume by humans would be your typical YouTube creator channel, or a blog by an industry expert.
An example of human-generated content in high volume, which is however not high-authority or editorial, would be the User Generated Content (UGC) by communities and discussion forums such as Reddit, Quora, Stack Overflow, etc.
An example of programmatically generated content would, of course, be an article written by ChatGPT. But a more effective example would be review websites, such as TripAdvisor, Glassdoor, G2, etc., which take in primary data given to them by humans, and programmatically generate millions of unique pages to accurately match specific searches like ‘top 10 churches to visit in North Goa’ or ‘content writer salaries in Berlin’ or ‘top mobile marketing tools’.
Selecting which channel within this growth lane to pursue, therefore, first requires answering what your strength as a business is.
Do you have the authority as a creator to generate editorial content which will have unique knowledge or personality as perceived by your prospective customers?
If not, does your business have a community which will generate content in high volume and thus help your discoverability?
If not, do you have access to some primary data that you can use to programmatically generate secondary data that would be valuable to the prospective customers?
While the focus is often on the platform-native techniques (such as Google Search SEO, or YouTube SEO, Instagram growth hacks, etc.) when it comes to this lane, a business should first answer the above questions to identify their strengths as a business, identify their strategy clearly by placing themselves in one of the quadrants in the 2x2 grid above, before jumping to execution which can be a long-term effort.
Summary
In this essay, we discussed the nuances of specific growth lanes and the key channels within them (albeit heavily in favor of digital advertising and partnerships). We covered concepts such as
Designing a digital advertising campaign, its three key aspects of business objective, customer profile and communication design, and their respective specifics
Building a partnership channel such as influencer marketing, and how to build contextual learnings
Incentive design for a User Referral program and its pitfalls. And, the importance of distribution campaigns for Referral programs.
Potential paths in Content Marketing, and how does a business identify their strength as a business and therefore its relevant content marketing strategy
While this essay focussed on levers for User Growth, the next essay will focus on levers for Value Growth i.e. how to improve your Activation and Retention.
If you have any suggestions on this article, or need any clarifications, I am reachable at sudhanshu@skilletal.com