August 4, 2026
Release notes are a list of things that went in. They are easy to write and they flatter you. The list nobody publishes is the one that says what got cut, and in a category where every competitor is racing to add more, that list is closer to what the product actually is.
So here is ours.
The statistics pack
There was going to be a bundle: variance, standard deviation, distribution charts, results by day of week, results by time of day. It reads well on a comparison table.
We took it apart. Variance and standard deviation went into Bankroll Advisor, where they’re ingredients in an answer rather than numbers on a wall. Day of week went into the free stats, because it’s table stakes in this category and charging for it would be embarrassing.
The reason we wrote down at the time: a pack of statistics isn’t a result, it’s just a lot of numbers. The apps that already win on sheer statistical depth do it very well. Their gap is that they hand you a wall of figures and no interpretation. Competing on depth would mean losing to them at their own game while ignoring the thing they’re bad at.
Two slicing axes
You can now recalculate your stats for one stake, one room, one weekday. Two candidate axes didn’t make it.
Time of day overlaps almost entirely with day of week, and once you cut a live player's sessions into time buckets the sample in each is thin enough that noise starts looking like insight. That’s worse than not having it.
Session length is worse still, because the causation runs backwards. Long sessions don’t lose money. Losing makes people sit longer. Show that as a chart and you’ve built a machine for confirming a bad habit.
The note reads: it isn’t the number of axes that makes this worth anything, it’s that they’re honest.
Ten columns of the export
The CSV export started at twenty-three columns and shipped at thirteen. Out went timestamps, time zones, the breakdown of re-entries and reloads, prize and bounty split out separately, starting stack, field size, exchange rate.
The question that settled it: what is this file for? It’s your own copy of your own results, to open in a spreadsheet and cut up however you like. It’s not a migration package for moving to a competitor. Thirteen columns do the first job completely.
Duration stayed even though the timestamps went, because without it there’s no hourly rate, and an export of a poker record with no time axis isn’t much of a record.
The upsell banner
There’s no banner on the home screen. No badges on locked rows, no teasers, no little arrows hinting that something better exists one tier up.
You find out there’s a paid tier at the moment you try to do a paid thing, and not before. Everywhere else the app stays quiet.
This is partly taste. It’s mostly the observation that the apps we admire do exactly this, and the ones covered in badges look cheap in a way that’s hard to unsee.
Anything that does not exist yet
Two things people would happily pay for are deliberately not mentioned anywhere they could be mistaken for shipping features. Not on the paywall, not in the store listing.
Partly because selling something you haven’t built is a review rejection waiting to happen. Mostly because it isn’t true yet, and the day you start describing plans as features is the day your own documentation stops being reliable.
A second tier
The category leader runs two paid tiers, with its AI features reserved for the expensive one. We had the same structure drawn up and dropped it.
Splitting a small feature set across two tiers means every one of them has to be explained and defended, and the honest version of that explanation is usually "because we wanted more money from the people who would pay it". One tier means one decision for you and one thing for us to get right.
Countdowns and scarcity
No "only 100 left", no timers, no urgency copy. Aside from the review risk in promising a number you then have to honor, it’s the tone of a business that doesn’t expect you to come back.
The pattern
Read the list again and it’s the same call made six times. The market we’re in treats a long feature list as evidence of value. Most of the time it’s evidence that nobody was willing to say no.