This post was written by Fareed Mosavat and Brian Balfour with key insights provided by Elena Verna and Bangaly Kaba.

Are you tired of reading about COVID-related things? We hear you. It's exhausting. It's likely that you, your team, and your company have worked hard to make some immediate changes in the changing environment. Maybe it feels like the pressure is easing. But we are here to say that the real work is just beginning.

We've been singing from the rooftops for years that retention and engagement determine category leaders. Retention is also the silent killer of products and companies. Those things haven't changed, but the underlying and surrounding factors that drive retention and engagement for your product have. You may have completed work to react to the initial environment changes, but whether you are experiencing headwinds or tailwinds, there is a lot of work ahead to continue to adapt.

This all leads to the question, how do I determine my retention and engagement strategy going forward? Over the past few months, Reforge EIRs have been talking about this in places like the Growth In Turbulent Times series. We wanted to aggregate our thoughts here.

In Times Of Uncertainty, Don't Lose The Fundamentals

In times of uncertainty, it is easy for teams to lose sight of the fundamentals. But the fundamentals are the stable foundation you build on. As you'll see below, we keep revisiting fundamentals to inform our strategy in uncertain times. In the Reforge Retention + Engagement program, three fundamentals we go deep on are:

Retention is about building and deepening habits.
To create great retention, you need to build and deepen user habits over time. Habit building is a science. Users have motivations, cognitive and physical friction, rewards, and more that all play a role in habit building.
Retention is about usage, not revenue.
It's likely the executive team is seeing revenue retention drop and that's setting off warning bells. But revenue retention is an output of usage. It's a lagging indicator, and if you are focused there you are looking in the wrong place.
Retention is an output. To move retention, you need to focus on an input.
If your team is sitting around discussing ideas about how to improve retention, you are asking the wrong question. Retention is an output. To improve it, you need to move one of its core inputs: Activation, Engagement, or Resurrection.

These three fundamentals correspond to three ultimate questions you should be asking not just during COVID but all the time:

How have the habits of my customer base changed? How might they change going forward?
The core elements (motivations, actions, friction, rewards) of your current and potential customers have probably changed. Have the habits been broken? Weakened? Strengthened? Accelerated? Why? It is likely this has happened in the past couple of months. But the answers to these questions are going to start changing.
How has usage of my customer base changed or how is it changing?
As a result of those changing habits, the usage behaviors of your current and potential customer base have probably changed, too, and will continue to change. Looking at usage metrics isn't enough — you need to understand why, which is the point of the first question. This means that things like churn prediction models, engagement notification models, live experiments, and more at a minimum need to be re-evaluated and possibly need to be thrown out the window. This doesn't just apply to retention, but your entire growth model.
As a result of #1 and #2, which input (Activation, Engagement, Resurrection) should I invest resources to improve?
Based on the first two questions, you should start to get a sense for which input you will need to align resources toward in order to build a proactive strategy. Each area has different strategies to improve it and requires different amounts of effort to drive it.

The very best companies are constantly monitoring not just quantitative measures of retention but also the underlying motivations and habits driving retention. With the arrival of COVID and global shelter-in-place orders, motivations changed and habits shifted faster than any time that we can remember. Customer behavior literally changed overnight. As a result, many businesses saw wild swings in their quantitative retention measures, which were the outputs from changes in psychological and logistical inputs. These motivation shifts occurred along a spectrum, depending on how customers think about your category in relation to the COVID-driven environment.

The Headwind/Tailwind Spectrum

There are products that are experiencing extreme headwinds or tailwinds, and others experiencing something in between. Your retention strategy will vary depending where your company sits on this spectrum.

Extreme Headwinds

The majority of your customers' habits with your core problem have been completely broken.
Examples: Yelp, Uber, Kayak, Eventbrite, OpenTable, Away, Opendoor, Airbnb, and more.

Extreme Tailwinds

The majority of your target customers' habits have been accelerated and strengthened. You are also seeing unintended audiences start to build new habits.
Examples: Instacart, Zoom, Peloton, Netflix, Disney+, Loom, and more.

In-Between

Some of your customers' habits have been broken, some weakened, some strengthened, etc. The net effect of this and how it might change is likely not 100% clear.
Examples: Everyone else. HubSpot, Gong, Figma, Letgo, Venmo, Spotify, and more.

Tailwind Strategies

Core Question: Who are you going to keep?

““If Covid ended tomorrow, how many of your users would sustain vs churn?””

— Bangaly Kaba, Former Head of Growth at Instagram

If you are in the extreme tailwind category, you have likely experienced accelerated growth in a few categories:

Pre-COVID Core Audience: Your core target audience pre-COVID. COVID has accelerated habit creation with this audience. You are probably experiencing "pull forward" from the market.
- Instacart Example: Residents in urban areas that don't own cars. Existing customers are likely increasing their frequency or their order sizes since they are probably consuming more meals at home.
- Zoom Example: Core business users with lots of remote meetings. Without the possibility of in-person meetings, usage for core customers has skyrocketed. 7 or 8 hours a day on Zoom is the new normal for some people.
Marginal Audience: The audience sitting on the periphery of your pre-COVID core audience. These are people for whom your solution was kind of a fit before, but now their habits are being accelerated because they are feeling the problem amplified and that's motivating them to get over friction that was previously too much.
- Instacart Example: Urban or suburban residents who are car owners. Pre-COVID, these were infrequent users who perhaps only used Instacart for special circumstances(small work party or a weekend BBQ). During COVID, these users may now be weekly purchasers for home deliveries.
- Zoom Example: Employees in companies with physical offices that are now fully or partially remote. These are people who might have previously used Zoom once in awhile, but are now likely using it daily.
Unintended Audience: Audiences you had not built the product for pre-COVID. They are using your product now essentially because they have to.
- Instacart Example: People who love and enjoy the shopping experience but are now being forced to order through Instacart because of the health risk of going to grocery stores.
- Zoom Example: Teachers/Professors who are now forced to teach remotely. Yoga instructors who now teach virtual classes because they can't teach in-person.

With limited resources, the core question then is where do you align those resources? Who do you try to keep vs. who do you ignore?

If you try to keep everyone, you will lose.

Here is the key: If you try to keep everyone, you will lose. Trying to keep everyone will lead to extremely sub-optimal results as we continue to see a behavior shift. Your core focus should be on identifying and retaining the marginal audience. Why?

If your product had decent retention with your pre-COVID core audience, people in that audience that have been "pulled forward" are very likely to retain with the existing product and efforts. Your unintended audience is unlikely to retain unless you pivot or significantly invest in enabling a new product Use Case. But by definition, these are people that are a couple "hops" away from your existing audience and core use case. As a result, your marginal audience will likely have the highest return per investment. Which raises the question, how do you identify the marginal audience and lock in their behavior?

““Is the marginal audience experiencing real value or only temporal value? To understand this you need to look at the underlying behavior that is driving engagement. If they are engaging only because of current events, but not experiencing the broader value prop, then those users will not retain.” ”

— Bangaly Kaba, Former Head of Growth at Instagram

Identifying your marginal audience

Your marginal audience includes those who:

Weren't regularly using the product before, but are now...
For a Use Case that is adjacent to your core use case...
And have moderate to high satisfaction.

Some of you might start looking at engagement data. But current engagement might be a mirage. Let's take Zoom as an example. I might be a college professor who is now teaching via Zoom because I have to. I have a class every day, so my engagement makes me look like I'm a daily active host. But Zoom was not built for this audience or use case initially, so I'm pretty dissatisfied with Zoom. Zoom was built more for the business audience. But the teacher's engagement data is going to look very similar to that of a remote worker. So here, engagement data alone tells us nothing about whether or not they are the marginal audience.

As a result, I need to segment my customers based on a few questions:

What Use Case are they using it for?
Is that Use Case adjacent to my existing Use Case? Or a couple of steps removed?
What is their satisfaction with my product for that Use Case? "How likely are they to continue using my product once shelter-in-place restrictions ease and motivation shifts again?"

For those that are on adjacent Use Cases with moderate to high satisfaction, then the question becomes how do we lock in their behavior? That might involve building a couple of new features to fill in gaps for the Use Case. Or, building out specific marketing and product communications to highlight that Use Case.

Headwind Strategies

Core Question: How do I restart the habit engine and with whom?

If your product is on the Extreme Headwinds end of the spectrum, then the habits have been broken with the majority of your customers. Most teams are likely thinking about when these customers will come back. But habits take time and reinforcement to establish. They aren't likely to just magically re-appear.

Products in this category are playing one giant resurrection game. Resurrection is about re-establishing the habit with users that lost it. Resurrection campaigns typically re-activate users at much lower rates than initial activation rates. As a result, if you are in this category you are likely to only re-activate a portion of your pre-COVID customers.

Similar to tailwind strategies, trying to re-activate everyone is going to lead to sub-optimal results. Your strategy should prioritize segmentation and order of operations. Essentially, the core question for products in this category is how do I restart the habit engine and with whom?

Prior Engagement Will Be Misleading

When it comes to the question of who to restart the habit engine with, a common first answer might be those who had the deepest engagement and usage pre-COVID. But this could be misleading. Let’s look at a couple of examples:

If we are Eventbrite, looking at prior usage data might lead us to a segment of customers like concert promoters. Pre-COVID, they had high frequency and high value. So what if we focus there first? It's unlikely that this segment will be the first to come back to the platform. So who is likely to come back first? Probably those hosting smaller outdoor events.

If we are Airbnb, looking at prior usage data might lead us to a segment of customers like business travelers. They have high frequency, high value and are probably traveling longer distances. Or it might lead us to world travelers who also have high frequency, but are probably staying in rooms vs. entire apartments. Both segments are very unlikely to be the first to come back. It's more likely that local travel within driving distance for whole apartments/houses will be the first.

Who do we focus on and how?

The questions you need to ask to figure out your strategy are:

What are your Use Cases?
Hopefully you have this already. But if not, list out the Use Cases for your product.
Which Use Cases do you think will come back first?
Evaluate each Use Case. Which ones do you think will return the soonest and why? It's probably useful to do a group stack rank exercise to force discussion on the why.
For each Use Case, what is the strength of the alternative they've established?
This is incredibly important. Just because a Use Case might come back first doesn't mean you focus there. Recognize that you are now competing with a new set of alternatives. In a lot of cases your customers have established new habits around alternative solutions. Those alternatives could be weak or strong for a customer.
For example, if I was a ClassPass subscriber and regularly went to fitness classes pre-COVID, but COVID pushed me to buy a Peloton to work out from home, am I likely to go back as a ClassPass subscriber at the same level? It's extremely unlikely. I've just invested in a $2K bike and gotten into a new workout routine.
It's the reverse of the tailwind groups. You will have customers who:
- Hate the alternative, and therefore are more likely to come back.
- Are indifferent to the alternative.
- Like the alternative, and therefore will be very hard to get back.
How do I align my strategies?
Let’s go a little deeper on this question. We can plot #1 - 3 on a 2X2 where we end up with four possible places for our Use Cases to land.

The quadrant where your Use Case falls determines the path you might take for that Use Case.

Quick Return, Weak Alternative
Prioritize this group with re-activation campaigns. You probably won't have to build in deep incentives with this group due to the weak alternative.
Quick Return, Strong Alternative
This group is likely a lost cause. You probably won't be able to invest enough in getting them to break their habit with the strong alternative in that short of a time period. Deprioritize this group.
Long Return, Weak Alternative
Right now, the strategy should be to "stay top of mind." They are likely to return when they can due to the weak alternative. But you want to make sure your product is top of mind when the timing is right. Creating and delivering value-add content, virtual sessions, and more is the likely strategy.
Long Return, Strong Alternative
Reactivation campaigns or just staying top of mind are not going to bring this group back. To bring them back you likely need to invest in building a new Use Case that addresses the strong alternative. This essentially means going from zero to one on a new product initiative.
For example, if we are the company Mindbody, which sells software to yoga and other fitness studios, it will likely be a long time before normal studio activity returns. Also, it is likely a lot of our customers will go out of business. So what is a new Use Case we can invest in that is adjacent to our core? Maybe we build software for individual instructors to build a virtual business.

One side note: Some of you reading this are working on companies with different types of network effects. It is possible that the network effect is now broken, i.e. no liquidity. ****This means you can't just fire up the same engine; you need to roll the engine back in time. Return to the unscalable tactics that got initial liquidity going.

In-Between Strategies

Core Question: How do I be proactive vs reactive?

Products in the extreme headwinds or extreme tailwinds groups are experiencing first order effects of the environment. Their user habits and problems are closest to what has changed. As a result, the impact on metrics was immediate, significant, and clear.

But those in the in-between part of the spectrum are there because they are experiencing second or third order impacts of the environment. This is likely because rather than a majority of your customers' habits being broken or accelerated, your customers sit across the spectrum. You have some experiencing extreme headwinds, some extreme tailwinds, and some in-between.

Compared to extreme headwind and extreme tailwind products, the impact on your metrics is not immediate, may not be significant, and may not be clear. There are a few things to keep in mind:

Your Effect Is The Net of Your Customers' Effects
The end effect (positive or negative) you are seeing on your business is the net effect of what your customer base is experiencing. In other words, you might have a portion that are experiencing headwinds and therefore churning and some that are experiencing tailwinds and therefore accelerating acquisition or expansion. Which one is bigger depends on where you fall on the spectrum.
That Net Effect May Change
You might have a handle on the current net effect, but that's unlikely to be predictive of the future. In other words, in the past couple months maybe you saw higher churn rates for those customers that were immediately impacted. But you also saw an acceleration of expansion and some new acquisition and therefore you are slightly net positive. But what happens in the mid-term and long term? That net effect can flip if the economy continues to stay down, layoffs continue to happen, small business continue to go out of business, and a lot more.
Top Line Metrics Will Hide What Is Going On Beneath The Surface

Top line metrics like revenue, churn, new acquisition and more will hide what is actually going on beneath the surface. They once again are lagging metrics. If you are guiding your company based on the effect of these metrics, you will be driving a reactive strategy and not a proactive one.

Bottoms Up Categorization Of Your Users

Rather than looking at top down metrics, in order to truly understand what is going and how it might change you need to do a bottoms up categorization of your users. You can categorize them initially based on where you think they are on the headwind/tailwind spectrum. Usage is a good guide here. Those experiencing extreme headwinds will decline their usage, those with extreme tailwinds will increase in usage, and those in-between will see smaller positive or negative effects.

Understanding why each segment is experiencing those usage changes is a first step. Ask:

Has the problem you are solving for them changed? If so how?
Has the action or way of solving that problem changed? If so how?
Has the motivation around that problem changed? If so how?

Do A Pre-Mortem On Each Of The Categories

Once you assess where each customer is, and why, you can then start thinking about the future. The key here is to not be caught off guard.

“An exercise I like to do is called a “pre-mortem.” What are the extreme scenarios that could happen? What signals would tell us that this scenario is unfolding? What is the impact on the business? What would we do to mitigate? I want the team making decisions when the stakes aren’t high, so that if one of the scenarios happens we already have a flow chart of decisions spelled out.”

— Fareed Mosavat, Former Director of Product at Slack

The main thing I think is important here is that your ability to identify possible issues and also to come up with clear solutions is dramatically improved when you aren't already in a stressful situation. Humans are biased substantially to optimism - you need to explicitly give them permission to imagine the scenario that everything goes wrong.

Dan Hockenmaier, one of the leads on the Monetization + Pricing program, also has a good post on how to think about scenario planning: "Build your scenarios around the potential shapes of economic recovery. We are likely to see one of three patterns:

V-shaped recovery: A shock displaces demand but does not impact it in the long run. Historically, most epidemics have followed this pattern, but this is highly unlikely given that COVID-19 has cascaded into broader economic impact.
U-shaped recovery: A shock breaks the growth trend. Growth resumes on the same path, but later then planned. Financial recessions typically follow this pattern.
L-shaped recovery: A shock impacts a fundamental component of growth and results in a lower growth rate even after recovery. Japan’s “lost decade” is the canonical example."

We Are In The Middle Innings

With COVID, customer behavior changed overnight. We all had to react to those changes. But those that pull out of this in the lead will realize we are entering the middle innings of this game. Customer behavior is going to continue to change and we need to continue to adapt our strategies quickly.

In the words of Elena Verna, Reforge EIR and former SVP of Growth at SurveyMonkey, "It's of utmost importance to move fast and iterate now. A day right now could save you a month in the future. Waiting for things to go back to normal is not an option. Focus the organization on what matters most, build your grit, and come out stronger on the other side."

——

To receive high-quality content like this emailed right to your inbox, subscribe here. Or apply to the upcoming Reforge cohort to go deep on areas like Retention + Engagement, Product Strategy, Experimentation, and Monetization alongside top leaders like Fareed Mosavat, Elena Verna, Brian Balfour and more.