There is a famous thread on chess.com: “I've been doing 50 puzzles a day for 58 days and I'm not improving. What am I doing wrong?” Fifty puzzles. Nearly three thousand in two months. Zero rating points.
That player wasn't lazy — he worked harder than almost any club player. His mistake wasn't effort, it was arithmetic: puzzle gains don't add up like a runner's mileage. The formula is different, and it has three limiters.
Limiter one: concentration runs out before the puzzles do
A puzzle trains you exactly when you calculate honestly: candidates, opponent's replies, verification. That is expensive attention work — and attention degrades fast. You calculate the first puzzles, “estimate” the tenth, and by the twentieth you're guessing: click the most natural move, got it, next.
Guessing isn't just useless — it's harmful: you are training the habit of moving without calculating. Fifty puzzles a day all but guarantees most of them were solved exactly that way. We've written about how fading attention eats your games — with puzzles it's the same, only faster: ten minutes of honest calculation are worth an hour of clicking.
Limiter two: without review, everything you solve evaporates
The second mistake of the volume approach is assuming a solved puzzle stays with you. It doesn't: the forgetting curve wipes a fresh pattern within days unless a review disturbs it. Fifty new puzzles a day is a conveyor belt filling a leaky bucket.
So the right daily portion is not “as many new ones as I can take” but scheduled reviews first, new puzzles second. A review at the right moment — a day before the pattern would fade — locks it in far more cheaply than ten new puzzles. How that schedule works is covered in our piece on spaced repetition in chess; the consequence that matters here: most of a day's value sits in the first few reviews, not in the thirtieth new puzzle.
Limiter three: the goal must survive your bad days
The third reason not to inflate the number has nothing to do with the brain — it's about the calendar. Growth comes from continuity, not heroic Saturdays: a pattern you return to daily never has time to fade. And continuity rests on a goal you can hit even on your worst day — tired, traveling, in ten minutes.
A 50-a-day plan lives until the first crunch at work. One miss breaks the streak, a broken streak becomes “I'll restart on Monday,” and a month later you're at zero. A goal you close in 10–15 minutes is psychologically fireproof: the “no time” excuse has nowhere to stand.
Where 12 comes from
Now add it up: 10–15 minutes of honest calculation before attention degrades; scheduled reviews first, then new material; a number that survives a bad day. The intersection lands at about a dozen — our daily goal is exactly 12, assembled from the reviews due today plus new puzzles targeting your weakness.
That is not “too few.” Twelve honestly calculated puzzles a day is 4,380 a year, each one anchored by review instead of dissolving by Thursday. The forum player's two months of grinding bought him less than one month of such dozens.
Two footnotes to the number. Want more? Play a game: applying a pattern in battle trains harder than ten extra puzzles (which is why our day only counts as complete with a game played). And replaying familiar puzzles for fun is welcome — beyond the goal and off the books: in 99chess, practice mode costs no attempts and no credits, while the daily plan in your dashboard — scheduled reviews plus puzzles built from your own mistakes — is what your streak actually stands on.
Rating grows not with the number of puzzles solved but with the number of positions calculated. Twelve is how many you can truly calculate.