Here’s the advice nobody gives, because it sounds like a lack of ambition: do not perfect your first avatar. Make it, ship it, and start the next one.
I’ll make the case, because the instinct running the other way is strong and it costs people months.
You cannot predict which one lands
You are about to be wrong about your own work, repeatedly, and there is no amount of taste that fixes this.
The avatar you agonise over — the one with the backstory you’re proud of — will get polite traffic. The one you made in twenty minutes on a whim will be the one strangers keep opening. This is not a myOrbit phenomenon; it’s what happens in every catalogue business, from music to publishing to apps. The creator’s confidence and the audience’s response are only loosely correlated, and the only reliable way to find out is to put the thing in front of people.
One avatar is one bet. Ten avatars is a portfolio, and a portfolio has a property a single bet does not: it only has to be right once to work, and being wrong nine times costs you almost nothing.
The economics are on your side
Making an avatar is not expensive here, and that changes the correct strategy.
Creating one earns you Nuggets rather than costing them — the plans page publishes the figure and the cap. Discover lists it next to everything else, rated by the people who actually used it, with no placement to buy. So the marginal cost of one more attempt is close to zero and the marginal upside is uncapped.
When an experiment is nearly free and the payoff is unbounded, running one experiment is the mistake. Run ten.
What “ten” actually means
Not ten variations of the same character. Ten genuinely different bets, ideally across different types — the nine types exist because different people want different things at different hours.
A companion for the quiet ones. A debate partner who actually pushes back. A venting avatar that just listens. A character from a world you invented. A coach who doesn’t accept excuses. A gameplay companion. Something strange you can’t quite justify.
Spread across types, you learn something the variations would never teach you: which need you’re good at serving. That is the finding worth having, and it arrives in weeks rather than never.
Kill your darlings, quietly
Some avatars won’t work. Most, honestly.
This is the part that stings, so here’s the reframe that makes it survivable: an avatar nobody talks to is not a failure, it’s a result. It cost you very little and it told you something true about what people want from you. The creators who burn out are the ones who treat each attempt as a referendum on themselves. The ones who last treat it as a catalogue.
You can leave the quiet ones up — they cost nothing to exist and occasionally one wakes up months later when the right person finds it. Or you can delete them, and everything you made goes with them, which is the ownership commitment on the creators page doing its job. Your call, either way.
Then double down on what lives
Now the part that makes the whole strategy pay.
When one avatar starts pulling — better ratings, people coming back, conversations that don’t end after two messages — that’s your signal, and it’s worth more than any plan you had. Give it depth. Sharpen the personality. Give it a voice. Build the room around that one, because now you know what people came for.
This is the loop from part one in practice: many avatars for reach, then an orbit for the one that earned it. You didn’t guess which one deserved the room. You found out.
The rhythm
Week one, make three. Week two, make three more and look at what happened to the first three. Week three, make three more and give real attention to whichever one is alive.
Don’t wait for inspiration to strike ten times — most of these should take under an hour. The point isn’t ten masterpieces. It’s ten shots, cheaply taken, so the one that connects can be found at all.
Next: your first avatar — how to make the individual attempt good, now that you know how many to make.