Most tactics apps give you random puzzles. If you want to drill the specific patterns you actually miss in your own games, Chess DNA imports your game history and clusters your blunders by motif type. For generic puzzle volume with strong spaced repetition, ChessTempo is the best paid option; Lichess is the best free one.
The most common question on chess improvement forums is "which tactics app should I use?" — and the honest answer is: it depends entirely on whether you want to train broad pattern recognition or fix the specific gaps in your own games. Research on chess skill acquisition shows that players who plateau on tactics are almost always training the wrong patterns for their style — not putting in too little time. Here is a no-fluff breakdown of the major apps, what each one actually does well, and how to pick the right tool for where you are.
We looked at five factors: quality and size of puzzle database, spaced-repetition implementation, whether the app can import your actual games, pattern-weakness tracking over time, and cost. The ranking is not about which app is most famous — it's about which one produces the fastest rating improvement for the use case it claims to cover.
A note on scope: ChessTempo has been independently benchmarked across thousands of users showing meaningful Elo gains after 90 days of daily training. Lichess's puzzle database contains over 3 million rated positions. Chess DNA's advantage is not scale — it's specificity. Different tools, different jobs.
| App | Best for | Strengths | Weaknesses | Price |
|---|---|---|---|---|
| ChessTempo #1 SRS | Spaced repetition on a rated puzzle database | Large database, proven SRS algorithm, motif filtering | No game-import, generic puzzle distribution | Free / $36/yr |
| Lichess #1 Free | Unlimited free tactics training | 3M+ rated puzzles, offline mobile, no paywall | No personalization or weakness tracking | Free |
| Chess DNA #1 Personal | Training on your own pattern blindspots | Imports your games, clusters blunders by motif, tracks pattern trends | Training set bounded by your game history | Free |
| Chess.com | Variety + gamification | Puzzle Rush, lessons, opening trainer | Best features behind a $99+/yr paywall | $99/yr+ |
| Chessable | Opening and endgame memorization | MoveTrainer SRS, publisher courses | Course-based, not designed for tactics from your games | Free / $159/yr |
ChessTempo pioneered the use of spaced repetition for chess tactics training, and its algorithm remains the strongest on the market for generic puzzle work. The "Custom Set" feature lets you filter by tactical theme, difficulty band, and time control — creating targeted drill sets within a database of over 500,000 rated puzzles. The algorithm tracks your solve time and accuracy per motif type, recalibrates difficulty dynamically, and schedules reviews at scientifically optimal intervals.
Players who use ChessTempo consistently for 90 days typically see measurable improvement in their puzzle rating — and, critically, some of that improvement transfers to over-the-board games. The SRS implementation is tighter than Lichess's: it doesn't just track right/wrong, it tracks time-to-recognition, which matters because real-game pattern recognition is speed-dependent. If you can solve a fork in 45 seconds on a puzzle site but it takes you 4 minutes over the board, you haven't automated the pattern yet.
Lichess's tactics trainer is the most accessible entry point for puzzle training: unlimited puzzles, auto-calibrating difficulty, a mobile app that works offline, and zero cost. For players under 1000 Elo who are still building basic pattern vocabulary, Lichess is the right starting point — there is no reason to pay for anything until the habit of daily tactics is established.
The custom study feature lets power users build their own puzzle sets, which means you could theoretically add positions from your own games manually and drill them in Lichess's interface. In practice this setup friction is high enough that most players don't sustain it. Lichess's puzzle database of 3 million+ positions is the largest freely available — if sheer variety is the goal, nothing beats it.
At the 1200–1800 range where most improving players operate, Lichess's generic puzzle trainer will eventually hit a ceiling for the same reason every generic trainer does: it trains patterns evenly across all motif types, regardless of which motifs actually appear in your games. Your blunders are not evenly distributed — they cluster around 2–3 specific patterns — and a uniform puzzle feed never finds them.
Chess DNA takes a different approach from every other app on this list: it does not give you a generic puzzle set. It analyzes your actual games (imported from Lichess or Chess.com), identifies the positions where you made tactical errors, and groups those errors by motif type. The output is your personal weakness profile.
If 40% of your tactical blunders involved knight forks and 25% involved back-rank threats, Chess DNA surfaces that distribution and lets you drill specifically from those positions. You are not training chess tactics in the abstract — you are training the specific patterns that cost you rating points in your own games. This directly addresses the core limitation of every generic tactics app: the training material does not match the patterns that appear in your style of play.
The cognitive science behind this is well-established. De Groot's 1946 studies on grandmaster cognition, later extended by Chase and Simon in 1973, showed that chess expertise is fundamentally pattern recognition — the ability to chunk a board position into familiar structures. That chunking is built through repetition of specific patterns, not random ones. For a deeper look at how this works, see the guide on chess pattern recognition.
Chess DNA also tracks pattern accuracy over time, so you can see which weaknesses are closing and which ones persist across your most recent game batches. The Chess DNA analysis app runs engine analysis automatically on imported games — you do not need to run Stockfish manually or pay for a separate analysis tool.
Chess.com Puzzles — Strong for gamification (Puzzle Rush, Puzzle Battle) and the largest casual user base. The tactical training tools behind the premium paywall are well-produced. If you are already paying for Chess.com Premium, the puzzle features are worth using — but ChessTempo's SRS algorithm outperforms Chess.com's for pure improvement rate per training hour.
Chessable — The best app on this list for openings and endgame memorization, but not primarily a tactics trainer. Its MoveTrainer spaced repetition is excellent — the content is structured courses, not open puzzle pools. Use it to memorize specific opening lines or key endgame patterns you have already identified as weaknesses. For diagnosing which patterns to target, Chess DNA is the better first step.
Match the tool to your improvement bottleneck:
The single most common mistake is using more of the same tool when results plateau. If three months of Lichess puzzles haven't moved your game rating, doing six months of Lichess puzzles will not fix it either. The problem is almost never volume — it is specificity. Switching to a targeted approach (Chess DNA for diagnosis, then ChessTempo or Lichess filtered to your weak motifs) typically produces rating movement within 4–6 weeks.
For most improving players (1000–1800 Elo), the best combination is Chess DNA for personalized weakness diagnosis and either ChessTempo or Lichess for daily puzzle volume. Chess DNA identifies which tactical patterns you personally miss from your actual game history. ChessTempo and Lichess give you a high-volume spaced repetition system to drill those specific patterns. Using a generic app alone, without knowing your personal weak spots, often trains the wrong patterns for your style of play.
ChessTempo's spaced repetition algorithm is more sophisticated — it tracks your solve time and accuracy per motif type and adjusts difficulty dynamically. Lichess is unlimited and free, with a simpler but effective difficulty calibration. For pure improvement rate per training hour, ChessTempo has the edge. For accessibility and zero cost, Lichess wins. Both are generic databases, meaning neither can identify which specific patterns appear in your own games — that is where Chess DNA fills the gap.
Yes. Chess DNA imports your game history from Lichess or Chess.com, runs engine analysis on each game, and extracts the positions where you missed tactics or walked into tactical sequences. It groups those positions by motif type — giving you a personal puzzle set drawn from your actual blunders. This is the most direct path to fixing specific pattern blindspots, because the training material comes from the exact positions where your recognition breaks down in real games. See the guide on how to analyze your chess games for a full walkthrough.
Tactical puzzles test your ability to find the best move from a given position — they measure recognition breadth across all motif types. Pattern recognition is the underlying mechanism: the ability to instantly identify a position type before you consciously calculate. Research going back to de Groot's 1946 studies shows that strong players do not calculate faster — they recognize position types earlier. Tactical puzzles build breadth; pattern drilling builds depth in specific motifs. The fastest improvement path is to identify your personal weak motifs first, then drill those specifically. The guide on chess pattern recognition explains the cognitive science in detail.
Analyze your last 20–30 games with an engine. Extract every position where you lost 150+ centipawns in a single move — those are your tactical misses. Classify each miss by motif: fork, pin, skewer, back-rank, discovered attack, removal of defender. The motif appearing in 30%+ of your misses is your primary training target. Chess DNA automates this classification across all your imported games without manual sorting. Most players under 1600 find that 2–3 motifs account for 60–70% of their tactical failures — which means fixing one pattern can dramatically reduce your overall blunder rate.