By the end of 3rd grade, the standard target is automatic recall on both addition and subtraction within 20, plus multiplication and division within 100. Practical benchmarks:
40 mixed facts in 90 seconds with 95+ percent accuracy.
The inverse of multiplication. Once times tables are fluent, division comes within 4 to 6 weeks.
Should already be in long-term memory from 2nd grade. Add quick weekly check-ins.
One session, every day. Same time each day, ideally before something the kid likes.
Warm-up. 15 to 20 facts the child already owns. Build momentum.
Focus set. The 5 to 8 slow facts from yesterday. Repeated retrieval until each one drops under 3 seconds.
Mixed sprint. 30 mixed facts including one or two from last week to test the spacing held.
2s, 5s, 10s. Easy wins. Most kids reach fluency on these in two weeks.
0s, 1s, 11s. Pattern facts. These should be quick adds.
9s and squares. The 9s have the finger trick to start; squares anchor the diagonal.
3s, 4s, 6s, 7s, 8s. The hard middle. This is where spaced repetition does most of the work.
Mixed practice and division. Division comes much faster once multiplication is automatic.
No clean trick or pattern. They need the most spaced repetition. Expect 3 to 4 weeks of focused work.
6×7 = 42, 7×6 = 42, 42÷6 = 7, 42÷7 = 6. Teach all four together so the child sees the structure.
Reframe: the timer is for the facts, not the kid. Track personal bests, not class rank.
For the underlying method, see multiplication fact fluency. For the parent strategy guide, see how to memorize multiplication tables. When you are ready to put this into action, start a 5-minute session in MathBuilders.

By the end of 3rd grade, students should be fluent in addition and subtraction within 20 and have multiplication and division facts within 100 (typically 0×0 through 10×10). Common Core 3.OA.7 calls this out explicitly.
5 to 10 minutes a day is the sweet spot. Daily consistency at this length outperforms 30+ minute sessions twice a week, because retention follows frequency, not volume.
Drop the test framing. Use a timer to measure individual facts (under 3 seconds = known), not as a pass/fail event. The number on the screen is a diagnostic, not a grade.
Yes, when they are adaptive. The right app spends most of the practice time on the few facts the child is slow on, instead of cycling all 100 facts every session.