Whenever I start working with a new client, there's one question I focus on before (and above) all others:

What do signups who convert do differently from signups that don't?

The logic is pretty simple: activation is all about connecting signups with early user value, so that they're more likely to produce long-term business value.

User value is subjective, and therefore different for every individual user, which means we can't directly measure it in-app.

But we know that users have to do SOMETHING after signing up in order to produce value for themselves, and if users who produce value for themselves are more likely to produce value for the company, then it makes all the sense in the world to:

• A) See what actually happens between signup and revenue, and
• B) Especially see what the revenue-producing signups are doing differently from the never-converting ones

Across basically all companies, the things that (eventual) revenue-producers do immediately after signing up skews differently from what average signups do.

That difference can be demonstrated mathematically: if, for example, the average signup converts at a rate of 10%, but signups who do Activity X convert at a rate of 20%, then you can say that the "Activity-X-doing" signups are twice as likely to convert as the average signup -- in other words, it produces a 2x multiplier on a given signup's likelihood to produce revenue.

Of course, correlation is not causation, and some of that "multiplier" effect is the snail trail left behind by signups who were bound to convert anyway. I'm not suggesting that pushing more users toward the things that high-intent users want will magically turn every signup into a high-intent user.

What I am saying, though, is that the profitable signups are showing you a viable path to mutual success, and paving that path can only help more signups -- especially the ones on the margins -- produce value for both parties.

To be blunt: If your activation efforts are focused on guiding signups toward particular early user behaviors, you want to at least know that those behaviors actually make signups more likely to succeed in the first place (and by how much).

To explore this in detail, let's imagine looking into the activation data of an imaginary company called SaaSCo: an imaginary self-serve project management tool with an imaginary 7-day trial.

(The numbers are fabricated, but they represent patterns that I see across most real engagements.)

Casing the Joint

Before pulling up PostHog (or Amplitude, or Mixpanel, etc), I like to gain as much of an orientation as possible toward that elusive, subjective experience of "user value" that the given app unlocks.

I talk with the company about what they think users are trying to accomplish, what they've ruled out, where they're headed. I read the marketing site the way a prospect would: what problems does this solve? how can I expect my life to be better? what do the testimonials speak to?

I also walk through the activation journey itself, in all its permutations, to see what the app emphasizes and facilitates. I will also skim Capterra, G2, App Stores, Reddit -- anywhere actual users describe what they actually get out of the app.

All of this informs which behaviors get checked for multiplier-ness in future analysis. Things that go on the list:

• The activities that the value prop highlights
• The actions that the onboarding flow emphasizes
• The user behaviors that show up repeatedly in user reviews, whether the product explicitly encourages them or not
• The company's own hunches as well, but they're weighed against the rest of the picture rather than treated as the starting point

The idea is to go as broad as possible, but still keep things grounded with a litmus test of "which user actions would signups HAVE to take in order to get the value that the app seems to offer?"

By the time I open PostHog/Amp/Mix/whatever, I have a working model of what the product is generally trying to get users to do, as well as what users seem to actually want to be doing.

With that in place... *cracks knuckles* it's time to dive into the charts.

---

Funnel Settings

The formula for calculating multipliers is dead simple:

• First, calculate the conversion rate of all users within a given signup cohort
• Then, calculate the conversion rate of that cohort's users who did [Activity X]
• Then, compare the conversion rate of the activity-doers vs the average signups

But in order to get clean data from that formula, there are four big decisions to make first:

1. Which signup cohort?
   We want a window of recent signups that's fresh enough to reflect current product behavior but long enough to be statistically meaningful. Too recent and you risk capturing seasonal or campaign-specific noise; too old and you're analyzing a product version that no longer exists. Let's say that in a cohort window that covers a 30-day span, SaaSCo receives roughly 82,000 signups.

2. Which signup event?
   By the time an app is getting thousands of new users per week, there are usually two or three events that plausibly represent "signup": a legacy one, a current one, a mobile variant, etc. For this reason, it's key to confirm HOW exactly a given signup event fires (e.g. immediately after account is created in the backend, vs after a signup button is clicked on the frontend, vs when "onboarding is completed", etc). For SaaSCo, let's say signup_completed fires on the backend, as soon as the account is registered and has a persistent ID.

3. Which conversion event?
   Conversion here represents the production of business value. When I say "conversion" I usually mean "first payment", but it can stand for any point in the revenue-production timeline. For example, I often find "renewed in month 2" as a better long-term revenue indicator than first payment itself, because so much churn can happen between first payment and first renewal. You can even push it out as far out as CAC payback, which is the first sign of real profit. For SaaSCo, subscription_created is clean and consistent, so we will use that as a "first payment" conversion event, and keep an eye on longer-term revenue from there.

4. Which conversion window?
   When pulling these conversion numbers, it's easy to forget that the entire signup cohort needs enough time to perform the conversion activity. This is especially relevant in longer-term conversion contexts, like when conversion is defined as "renewed in mo2", which usually happens 40-50 days after signup. Long-term revenue is where profit is made, but tighter windows provide faster feedback. If SaaSCo is anchoring on first payment and has a trial length of 7 days, it's probably safe to say that a 10-day conversion window will represent most successful signups. For that reason, we want to offset our cohort window by 10 days as well.

With those parameters in place, we can establish a conversion baseline in the form of a two-step funnel: signup_completed → subscription_created, over a 30-day cohort, offset to allow the full 10-day conversion window.

The result:

• 82,450 signups
• 1,220 converters
• 1.5% baseline conversion rate

This is the reference point -- the rate that the other, event-specific multipliers will be based off of.

With it in hand, we can now compare the conversion likelihood of different user behaviors against this baseline and see which ones correspond to meaningfully higher conversion rates.

Running the Numbers

Now we can (finally!) start plowing through calculating the conversion likelihood of each of the behaviors identified earlier.

We can take each behavior and place it as a middle step, after signup and before conversion (in SaaSCo's case, signup_completed → [behavior event] → subscription_created).

For each behavior, we're looking to calculate two simple numbers: how many signups in the cohort did the behavior, and how many of those signups went on to convert.

Let's say the raw data for SaaSCo looks like this:

Behavior	# Who Did It	# Who Converted	CR%
signup	82,450	1,220	1.5%
paywall_clicked	7,600	980	12.9%
automation_created	2,474	198	8.0%
teammate_joined	5,772	436	7.6%
template_applied	4,947	351	7.1%
task_assigned	9,070	524	5.8%
integration_connected	14,841	747	5.0%
teammate_invited	11,543	547	4.7%
dashboard_created	23,086	718	3.1%
task_created	34,629	923	2.7%
calendar_viewed	56,066	1,079	1.9%

With these data points in hand, we can begin comparing them to the baseline.

Take teammate_joined as an example:

• 5,772 signups had another teammate join their workspace
• 436 of those signups went on to convert
• That makes for a conversion rate of 7.6%
• 7.6% is roughly five times the 1.5% baseline -- a 5.1x multiplier

And while we're here, we can also pull a couple more insights from the same data:

• Percent of signups: how many of this cohort's signups did this activity?
• Percent of converters: how many of this cohort's converters did this activity?

For SaaSCo, 5,772 of the original 82,450 fired the teammate_joined event, so their percent of signups is 7%.

And in terms of their percent of converters, 436 out of the 1,220 total converters fired that event first, which represents 36%. In other words, more than one out of three of all converters had a teammate join first -- not too shabby.

But how does teammate_joined stack up compared to the other candidate behaviors?

Funny but true: there's no way to find out.

Just kidding, that's the whole next section.

Comparing Multipliers

After calculating the three numbers for each behavior, we get a data set that looks like this:

Behavior	% of Signups	Multiplier	% of Converters
signup	100%	1.0x	100%
paywall_clicked	9%	8.7x	80%
automation_created	3%	5.4x	16%
teammate_joined	7%	5.1x	36%
template_applied	6%	4.8x	29%
task_assigned	11%	3.9x	43%
integration_connected	18%	3.4x	61%
teammate_invited	14%	3.2x	45%
dashboard_created	28%	2.1x	59%
task_created	42%	1.8x	76%
calendar_viewed	68%	1.3x	88%

Looking through the list, paywall_clicked has by far the biggest multiplier, but it's also kind of doye -- of course users would need to click through a paywall before they can start paying.

Still, there's room for a couple quick observations:

• If we look at percent of signup, only 9% of all cohort members ever clicked on any paywall. That seems low; are we sure it's firing in all the places that it should?
• And since this number is relatively low, it would make a lot of sense to track views of paywalls (not just clicks), ideally broken down by the features/limits that trigger them.
• Lastly, the 80% share of converters means 20% of customers have converted without ever clicking on a paywall. Could very well be worth looking into where they're converting instead -- maybe from a standalone 'upgrade' button? email promotion? users beelining it to the billing area of the settings panel, perhaps?

Either way, paywalls aren't a proxy for user value, which makes them a poor activation lever. This data point is included here mostly to act as an internal benchmark, rather than a behavior to target unto itself.

Speaking of behaviors to target, here's what they look like when plotted out across their signup%/convert% values, with multiplier zones highlighted diagonally:

Here are my thoughts as I work my way through each:

• automation_created has the highest multiplier of non-paywall events, with 5.4x. That said, barely anyone does it: only 3% of signups fired this event. If there are straightforward ways to surface automations more within the activation journey, they might well be worth split testing around. Otherwise, this is more of an acquisition insight than an activation one -- the real leverage is probably in getting more automation-desirers to sign up in the first place, rather than trying to talk the current acquisition mix into wanting it.
• teammate_joined comes in a close second with a 5.1x multiplier, and even though the percent of signups is relatively meager (7%), the percent of converters is pretty juicy: 36%, aka more than 1 out of every 3 new customers. What kind of levers could we pull to get that 7% exposure rate up into double digits? The invitation experience probably has some low-hanging fruit (they almost always do). In my analysis notebook, I'm circling this one as an area to investigate further.
• template_applied is next-highest at 4.8x, and accounts for just under 1 out of 3 converters (29%), so the same logic as above applies: how do we get a relatively-scant 6% percent of signups higher? Are templates currently featured in the onboarding, or highlighted in empty states? Bonus question: which templates are carrying most of this load? I bet it isn't distributed equally.
• task_assigned comes in fourth at 3.9x. Unlike the behaviors above, this IS something that a sizable chunk of users are doing -- more than one out of ten signups (11%) and nearly half of all converters (43%). It's interesting to consider the 43% vs 57% who have/haven't assigned a task before converting, though: are they two completely different groups, or is it more of a thing where all customers will use this as a key feature eventually, just not always pre-conversion?
• integration_connected had a similarly-healthy 3.4x multiplier, and very healthy volume as well: 18% of signups did it, as well as 61% of converters. Because of the volume, this 3.4x multiplier feels "sturdier" than the higher-but-fringier multipliers above. If I had to bet my life on moving only ONE behavior in order to move revenue, this would have a lot of curb appeal. Definitely worth investigating further, especially re: which integrations are doing most of the work.
• teammate_invited is the 3.2x upstream companion to the 5.1x teammate_joined event above. Some things that jump out to me: almost half of signups invite a teammate before converting (45%), and I bet a fair amount of signups invite their first teammate shortly after converting, as well. Even though a relatively small minority of signups invite a teammate, a significant amount of customers do. For that reason, it makes a lot of sense to be thinking multiplayer-first as a business model. At the same time, it would be super interesting to learn more about how singletons activate and convert -- clearly there are a lot of them, which could provide a whole other growth lever (and potentially a lower-lift one as well).
• dashboard_created is a somewhat ho-hum 2.1x multiplier, sitting somewhere in between "bankable lever" and "too generic to optimize for". More than a quarter of all signups create a dashboard, so clearly there's some traction in terms of intent here. How do the 59% of dashboard-creating customers differ from the 41% of customers who DIDN'T create a dashboard first? And which kinds of dashboards are customers counting upon the most? This feels less like a lever as much as a cluster of questions worth investigating further (and then lever-izing with more granular info).
• task_created comes in at a relatively uninspiring 1.8x, despite representing some of the highest volume (42%/76%) among any of the events analyzed. In that sense, this behavior feels more like basic vocabulary than a revenue lever. Is it possible to get value out of SaaSCo without creating a task? A quarter of converters seem to think so. While we're here, which specific events have to fire before this one? In our list, the only more popular event is calendar_viewed (next), but whatever comes between signup and task_created has to be recorded somewhere -- that's a funnel waiting to be tracked.
• calendar_viewed reads to me as an event that signups are railroaded into. Within any signup cohort, a concerningly large amount of users will drop off before firing ANY meaningful user event. Seeing that 68% of signups and 88% of converters experience this makes me think that it's a step that users are required to take as part of the signup process. If that's true, this is less of a lever and more of a health metric. If anything, it might be worth looking into the 12% of converters who haven't fired this event.

With all of this in mind, my focus would largely be on investigating these five events further (with dashboard_created as an honorable mention):

The basic idea is to key in on behaviors that...

1. You can guide more signups toward doing (aka push "up"), and
2. Maintain their multiplier goodness even when more signups do them (aka "move to the right when pushed up")

The "keeps moving to the right" part is the most important. You don't want to only focus on "pushing up" (getting more signups to do something), because if there isn't a corresponding uptick to conversions, you're just creating busywork for users while diluting the correlation that produced the multiplier in the first place.

Instead, the core idea is to identify the behaviors whose multiplier will hold up to increased volume, which is only find-out-able through a battery of split tests. Testing is expensive, though, so you really want to put these behavioral targets through their paces before committing.

On that note, this is really more of a starting point for further analysis, before committing to a direction to invest testing resources toward, rather than a conclusion unto itself.

The next step is to put the most promising candidates through even more analysis:

• How different are multipliers based on quantity? For example, are there even juicier multipliers after users have applied their second or third templates, instead of just their first?
• Which combinations of events produce even higher multipliers than individual events on their own? For example, what if people create a dashboard AND connect an integration? And even more intriguingly, does order matter?
• What about timing? Is there a big difference between signups who invite a teammate within the first five minutes, vs after a few days?
• Ooh and how about segments? What do these patterns look like on mobile vs desktop, or paid acq vs organic, etc?
• And lastly (but perhaps most importantly), how well do these hold up across longer-term conversion events? What is month-2 retention like for each of these behaviors? How about full-on payback?

But that advanced shit, my friends, requires at least a whole other article.

For now, we already have a decent idea of where SaaSCo's "conversion heat" lies, and plenty of opportunities to refine and pursue.

The only question left is which to dive into first.