In January 2021, we launched Alan Baby, a mobile app empowering new parents to live a positive and fulfilling parenting. And after 12 months of iterations, we decided to shut it down. We had 35,000 members and 1,200 "5⭐️" ratings, but we decided to focus on our mental wellbeing offer (more details here, in french).

Going from 0 to 1 on a consumer mobile app, extremely focused on creating health related value for our members was singular for Alan, which at the time was still mainly a B2B focused health insurance.

In this post, we share the main lessons we learned the hard way so that you can avoid doing the same mistakes we did:

Solve one important problem very well for core users, do not aim too broadly
Work on acquisition and product as one, to find product channel fit
Create an emotional connection with members on their first visit
Drive highly iterative, yet forward-thinking product work
Build a tight team with diverse skills

We hope our experience can be useful for consumer-oriented product builders. We’d love to hear your reactions ([email protected])!

Solve one important problem extremely well for core users

Alan Baby’s ambition was very broad: become a trusted health partner to help new parents overcome their challenges.

In the first few months, we built a generic offering based on chat-based support from midwives and doctors, parent support communities and content on diverse baby-related issues. It fell short of creating the special spark we were looking for, and engagement stayed disappointing despite our efforts.

Conversations with members and health professionals led us to identify baby sleep as the most visceral and acute problem mothers were encountering. We decided to focus all our energy on this well-identified** core member problem and persona. **This enabled us to build an effective and elegant feature, the “sleep program”, a day by day personalised coaching program to remove obstacles to better nights.

Solving a hard problem for them brought us members love: we started to get genuine amazing reviews like the one below, and saw engagement metrics increase.

A sample of the Alan Baby’s reviews after recentering on sleep

Once we had built enough confidence in these foundations (and only then), we started to build for the “adjacent member problem”. We did so by tackling problems that were not yet addressed by the product, and prevented parents with no deep sleep issues to experience the product fully.

In our case, to build on the core offer, food diversification was the next one because babies around diversification age often have sleep issues. Breast and bottle feeding would have followed as they can impact diversification and sleep and were open problems for our members. \

Work on acquisition and product as one, to find product channel fit

Acquisition and product are part of the same member journey towards using your product. As such they impact the member behaviour the same way and need to work in deep synergy. This has several implications.

Focus on cracking only one acquisition channel…

We initially saw performance marketing (paid social media ads) as a way to generate traffic on our app to learn about members’ behaviour, not an acquisition channel we wanted to rely on long term.

We therefore explicitly tried to devote as low attention to it as possible, while looking for the holy grail of “organic acquisition” across several channels (virality, word of mouth, influencers, SEO...).

We ended up failing to create an efficient paid acquisition engine, AND not finding significant leverage with organic distribution. It taught us that cracking any acquisition channel requires significant efforts and focus, and you should go one at a time:

Performance marketing requires complex integrations of attribution tools with the product and robust data pipelines to feed the advertisers optimisation algorithms, with strong restrictions coming from privacy protection measures such as Apple’s ATT. It also requires constant creative work, as a given campaign will create fatigue and perform less after a few weeks.
A successful organic acquisition comes from positive feedback loops that are core to the product narrative (the more members the better the product). Initiatives that are not core to the product like incentivised referrals may marginally boost acquisition, but they won’t allow you to magically crack organic channels. They made us lose time and focus.

The minute we decided to focus on a single acquisition channel, performance marketing, we were able to make a difference in acquisition performance.

… make it work with your product

Users coming through different acquisition channels behave differently in an app, so success is largely linked to bringing users with the right intent for the product you’re building.

To achieve this, all the touchpoints you have with members must tell the same story: from ads, third-party recommendation or piece of content showcasing your product, through the application stores product pages (screenshots, name,...).

We were initially reluctant to do so when focusing Alan Baby on sleep, because we thought it would undermine the scope of our product in member’s minds.

Before (top) and after (bottom) baby sleep rebranding for ads (left), appstore screenshots (center) and app landing page (right)

We finally overcame these objections and centered ads on sleep, revamped our Appstore page and screenshots, changed our main color to blue to suggest quiet nights… We saw a dramatic impact on the product usage and perceived quality by members as a result. This likely made more for our success than many “pure product” iterations we made in app: we had reached product channel fit.

To help you do this, a key element is to define your brand early, and not be afraid to adapt it when you pivot. You should also apply it consistently through product and marketing: tagline, storytelling, tone of voice, visuals, emotional relationship you want to have with the member, members personae…

Create an emotional connection with members on their first visit

Similarly to what happens between humans, the first impression your product makes is a big part of how your member will use it. Brand, product quality, tone of voice, value proposition are communicated and assimilated during this experience, and this perception is very hard to change.

We had been very successful in differentiating our insurance product with a very short and paperless onboarding, because of the low digitalisation of incumbents.

That’s what guided us in designing a first onboarding with a few screens, leading to the app.

It however failed to engage members on our key features well, mostly because they were not guided enough to do so. In the last version of Alan Baby, our onboarding was 45 screens long and lasted around 10 mins, but had a better conversion than the few screens we had originally beyond email/password input, and was an important generator of retention.

Alan Baby onboarding: before = 4 screens, D+1 retention: 20%

Alan Baby onboarding: after = 45 screens, D+1 retention: 40%

What's the trick? After onboarding, members knew:

That the experience would be personalised (onboarding survey)
That the product was empathetic (tone of voice and messaging)
That the product could really help them, and how (they had experienced the value proposition with a first program activity)
Which value they would find when coming back (more activities).

Don't rely on best practices, test for yourself instead because every product is different.

Drive highly iterative, yet forward-thinking product work

Work with a split mindset: forward thinking AND fast-paced weekly iterations

To successfully impact members behaviours on a digital service, you need to be focused, ship, learn and iterate fast. It took us a while to figure out how to do this efficiently (way of working, metrics…) because:

our true goals and success metrics were lagging indicators that take a long time to measure on a cohort (ex: monthly retention, Lifetime value). These are not actionable to conduct iterative product work as it takes too much time to know if you are moving in the right direction.
Yet, only working in short cycles iterations, prevented us at first to take a step back and take bets to trigger step changes, or pivot when needed

What unlocked efficiency for us was to define a split way of working, to drive both a high learning and iteration rate, and stay forward-thinking. We alternated:

Strategisation phase: 2 weeks to interrogate our long term success metrics, and decide to pivot (change goals, kill current initiatives, launch a new one) or continue
Building phase: a few weeks to ship the first version of an initiative (milestone), with no metrics based goal
Iteration phase: focused work to make progress on leading indicators of success on our goals (more in next paragraph), without looking at longer-term metrics.

The iteration phase: rely fully on leading indicators

We learned that it was key to define very few cohort-based leading indicators of success, and only use these as drivers for ideation and prioritisation. We relied on metrics measured after the member’s first day of usage:

usage of our main value proposition on first day (activation)
repeat usage on second day (retention)
conversion to trial on first day (when we had a paywall)

These are measurable very fast for a new version of the app - we released one per week - so looking at the cohort of new members, you can learn a few days after shipping whether the assumptions you made were correct, and define the next iteration.

Using these forced us to agressively focus on the first timer experience, and to neglect what happens for current users. We originally thought that this would prevent us from convincing members to use our product later on, at a time where it suits them best. Yet we moved away from this conception because:

First days on a product build the foundation: if members do not interact with your product a certain way then, it rarely happen afterwards, and trying to engage members on a value proposition that is not a part of their mental image of the product requires a high long term retention to modify this conception over time. Since it’s unlikely you start with a high retention, focus on first days over longer term is natural.
These short time range metrics are often correlated with your longer-term goals: if you increase usage on the first day, you’re likely to mechanically increase repeat usage a month later, and thus lifetime value of your member. We repeatedly measured this effect on usage data
They are a very efficient driver of progress, if you persevere: focusing the whole team for 3 months exclusively on activation and repeat usage is what allowed us to move these metrics, a step at a time.

Lifting D0 usage and D+1 retention lifts the whole retention curve (learn about retention here)

Generating assumptions by digging deep in data and being in close touch with your members

Finding the truth about members’ behaviours needs you to lean on 2 legs: usage data AND real life contact with members

Usage data will help you to understand how members behave better than what they and your intuition can tell you. As an example, we were proud to have a great introduction video for our sleep program with a skilled pediatrician, as we thought it would help members understand how it works… Truth was, this video was not opened much and actually caused a drop in the funnel, and we removed it.

Having so much data at hand can however lead to neglect speaking with members, thinking that they’ll tell you that they need faster horses. Keeping a constant stream of members interviews (2 per week), allows you to dig deeper than data and understand their motivations, intents and problems. When Alan Baby was free, we learned that this made members suspicious: “if it’s free you’re the product”. We added a short message to explain our reasons, which lifted the conversion of our funnel.

A positive side effect is that involving all the team in member’s interviews is very energising and a fruitful way to foster team creativity.

The secret sauce: a small tight-knit team with a varied skillset

Being a strategic bet for Alan, a company in hypergrowth, we had access to a lot of resources, and made the mistake to scale the team size too much, from 3 to 5 engineers after 3 months, thinking it could allow us to increase testing rate by parallelising experiments. It in fact defocused us, blurred learnings and put the team under pressure. Small teams are more flexible, and more easily aligned: scale the team only when you have found your product channel fit!

The high diversity of skills and how meshed the work of different team members needs to be to ensure impact was also a surprise for us: software engineers, marketing, product, design, content production, medical operations, data and health professionals all need to share workflows to ensure efficiency and velocity.

Doing so allowed us, in a month and a half to take the decision to launch a food diversification program, conceive a survey and over 70 content pieces with health professionals, integrate it in a seamless app experience, track it the right way, tune our ad creatives to integrate it in members’ mental image of our product.

Reaching such a powerful team dynamics with a high level of focus and learning ability was probably one of the most fulfilling aspect of the Alan Baby adventure, and I wish to all readers to be part of such a journey.

We hope our experience can be useful for consumer-oriented product builders. We’d love to hear your reactions ([email protected])!