Retention is the only
growth metric that compounds

Every founder can recite their signup number. Fewer can tell you what those users are doing four weeks later — and that second number is the one that decides whether a product becomes a business. At Odyssey, retention is the first thing we look at and the last thing we compromise on. Here's why.

Acquisition is a cost. Retention is an asset.

Paid acquisition buys you a user once. If that user churns, you've rented attention and given it back. Retention is different: a user who stays generates value every cycle — more revenue, more word of mouth, more data to improve the product — without you paying to acquire them again. One is an expense line; the other quietly compounds.

That's why we refuse to pour money into a funnel until the retention curve has flattened. Spending on acquisition before you retain is like turning on a tap over a bucket with no bottom: the faster you pour, the more you waste.

If the curve doesn't flatten, you don't have a growth problem. You have a product problem wearing a growth costume.

Read the curve, not the average

A single "retention rate" hides more than it reveals. What we actually study is the shape of the curve over time:

• Does it flatten? A curve that keeps falling means you're on a treadmill — every new user eventually leaves. A curve that flattens means you've found a group of people for whom the product genuinely sticks.
• Where does it flatten? The height of the plateau is your real addressable core. Flattening at 40% is a very different business than flattening at 8%.
• Does it smile? The best products show a curve that dips, then rises as dormant users return. That upward bend is the strongest signal a market can send you.

Retention is a feature of the problem, not just the product

Some problems recur naturally — sleep, learning, money, health. Others are one-and-done. No amount of clever design turns a once-a-year problem into a weekly habit. That's why our venture selection starts with the question, "does this problem come back?" We'd rather build for a recurring pain in a smaller market than a one-off need in a huge one.

When we evaluate a new idea, we're really asking: is there a natural reason a person would open this again next week? If the honest answer is no, we don't build it — however large the initial demand looks.

What we do about it

Because retention decides everything downstream, we treat it as a first-class design constraint, not a metric to optimize later:

• We define the core loop — the smallest cycle of value a user comes back for — before we design anything else.
• We instrument that loop from day one, so we can see the curve the moment real users arrive.
• We hold acquisition spend flat until the curve tells us the product deserves fuel.
• We revisit the curve every week. It's the closest thing a consumer product has to a heartbeat.

None of this is glamorous. It's slower than buying a spike of installs and calling it traction. But it's the difference between a number that goes up because you paid for it and a number that goes up because people came back on their own. Only one of those compounds — and it's the only one we're interested in.