%20scilla.studio.svg)
Your product metrics are not a scorecard of independent numbers. They are a system that moves together. Push retention up and your CAC payback period usually shortens, while your K-factor barely budges. "Fix" activation and your Day-90 retention can quietly get worse. The single most common diagnostic mistake is treating one weak metric as a thing to fix on its own, and the fix backfires because you ignored what it was attached to. Here's how the numbers interact, and how to read them as a system instead of a list.
Every headline metric is a downstream sum of the same underlying behaviour: people show up, get value (or don't), come back (or don't), tell others (or don't), and pay (or don't). Because they all draw from that one chain of events, you can't move one link without nudging the others.
A few of the load-bearing relationships:
What's the practical consequence? Any time you read a single number against a benchmark and conclude "this one is broken, let's fix it," you're one step away from a wrong move. The diagnosis has to start one level up: what is this number attached to, and which direction does the attachment pull?
Retention is the percentage of a cohort still active some number of days after their first use. More than just one metric among several, it's the input that the others are computed from.
Walk the chain. Lifetime value is, roughly, average revenue per user multiplied by how long they stay, and "how long they stay" is retention. So when retention improves, LTV rises, which improves LTV:CAC, which shortens CAC payback. One behavioural change, four metrics moved. This is exactly why our cohort growth model runs retention and growth off a single monthly retention curve: they're not separate dials, they're the same dial viewed from two angles.
For B2B SaaS, the benchmark bands look like this:
| Metric | B2B SaaS (avg) | Consumer (avg) |
|---|---|---|
| Day-1 retention | 50 to 70% | 20 to 30% |
| Day-7 retention | 40 to 60% | 8 to 15% |
| Day-14 retention | 35 to 55% | 4 to 8% |
| Day-90 retention | 25 to 35% | 1 to 4% |
Sources: Pendo Product Benchmarks, Amplitude, Mixpanel (B2B); Adjust, AppsFlyer, UXCam (Consumer).
Here's the interesting part: because retention sits upstream, a weak retention curve quietly poisons every downstream metric, and you'll spend months optimising the symptoms. If your CAC payback is too long, the instinct is to cut acquisition cost, but if the real cause is that users churn before they pay you back, cheaper acquisition just gets you more people who leave. The cause and the symptom live in different metrics. (More on that failure mode in why your CAC payback is too long.)
LTV:CAC is the ratio of what a customer is worth over their lifetime to what it cost to acquire them. The healthy band is 3:1 to 5:1 for B2B SaaS and 2:1 to 4:1 for consumer (ideal around 3:1). Below 2:1 is structurally risky; above 5:1 (and this is the part most dashboards won't tell you) often means you're underinvesting in growth.
Sources: Bessemer State of the Cloud, OpenView SaaS Benchmarks, a16z (B2B); Adjust, AppsFlyer, a16z (Consumer).
So you see 8:1 and feel great. Honestly? That's the number I worry about most. An 8:1 ratio usually means there is cheap, profitable demand you're leaving on the table because you're spending too cautiously. The metric in isolation says "efficient." The metric in context says "you could be growing twice as fast and still be healthy."
This is the cleanest example of why one number lies. LTV:CAC only means something when you read it next to two companions:
We wrote a whole piece on this exact trap: LTV:CAC looks great but you're not growing. The ratio is a constraint, not a goal.
Activation is the share of new users who reach first meaningful value (completing onboarding, hitting the "aha" action). The healthy band varies too much by product and funnel definition to pin to a single number. What matters here is not the absolute rate, but the direction it pulls the metric next to it.
The naive move is to treat low activation as a funnel to widen: remove steps, lower friction, push more people through. And activation goes up: the metric in isolation looks fixed. But activation and retention aren't always allies. If you widen the funnel by pulling in poorly-qualified users, more of them "activate" and then churn, and your Day-90 retention erodes even as your activation chart climbs. You optimised one number into a worse business.
What's a healthy activation improvement, then? One where the retained cohort grows, not just the activated one. The only way to know is to read activation and the retention curve together, which means activation is never a metric you fix alone.
K-factor is your viral coefficient: how many new users each user generates before their influence runs out. B2B virality is usually weak: 0.1 to 0.3 is the average band, and anything above 0.3 is unusually strong unless you've got real collaboration loops built in. Consumer sits higher at 0.3 to 0.7, with K above 1 being genuinely rare (Reforge, Andrew Chen).
But K-factor on its own can't tell you how fast you'll grow, because it has a hidden partner. What's cycle time? How long the viral loop takes to complete, from a user creating an output to a new signup landing. The relationship is multiplicative:
Monthly growth rate ≈ K / cycle time (in months)
Consider two products. Product A has K = 0.8 on a 60-day loop. Product B has K = 0.6 on a 10-day loop. Read the K-factors in isolation and A wins. Read them with cycle time and B grows roughly 4.5× faster, because its loop turns six times as often. Shortening your cycle from 30 days to 15 has the same effect as doubling K. We unpack this in improving K-factor without referral hacks. The point here is just that the headline number is half the story.
And notice the second-order interaction: cycle time depends on retention. If users come back daily, you get ~30 chances a month to trigger a viral output; if they barely return, you get one or two. So retention quietly feeds K-factor too. It really is all one system.
The fix isn't a new metric. It's a different habit: diagnose from the system, never from the single number. When a metric looks wrong, walk three questions before you touch anything.
This is why we built the benchmark tool around bands and relationships rather than pass/fail targets. A number that's "green" against a benchmark can still be the wrong thing to optimise, because the benchmark only knows the number, not what it's attached to in your product. (And yes, that means a metric inside the healthy band can still be telling you to do the opposite of what you think.)
The benchmark tool at benchmark.scilla.studio charts your retention, K-factor, activation and unit economics against B2B and consumer bands in a couple of minutes, side by side, so you can see how they move together instead of squinting at one number at a time. It's free, and it's built to surface the interactions, not just the scores.
Why can't I just fix my weakest metric? Because metrics are downstream of shared behaviour. Your weakest number is often a symptom of a different, upstream metric (usually retention). Fixing the symptom without the cause wastes the effort and sometimes makes the cause worse, e.g. cutting acquisition cost when the real problem is that users churn before payback.
Which metric should I look at first? Retention. LTV, CAC payback, K-factor and growth rate all inherit from how well users stick, so a weak retention curve quietly distorts everything downstream. Evaluate retention quality before CAC efficiency.
Can a good metric be a bad sign? Yes. An LTV:CAC above 5:1 often means you're underinvesting in growth rather than winning: there's cheap, profitable demand you're not capturing. The ratio in isolation reads as "efficient"; read alongside a flat growth rate, it reads as "spend more."
How are K-factor and cycle time related? Multiplicatively. Monthly growth ≈ K ÷ cycle time in months. A lower K-factor with a faster-turning loop can dramatically outgrow a higher K-factor with a slow loop. Halving your cycle time has the same effect as doubling K.
Does improving activation always help retention? No. If you widen the activation funnel by admitting poorly-qualified users, more of them activate and then churn: activation rises while Day-90 retention falls. A healthy activation gain grows the retained cohort, not just the activated one, so the two must be read together.
Sources: Bessemer State of the Cloud, OpenView SaaS Benchmarks, KeyBanc SaaS Survey, a16z, Reforge, Andrew Chen, Pendo Product Benchmarks, Amplitude, Mixpanel, Adjust, AppsFlyer, UXCam, Mobile Dev Memo.
