📖 Part 8: FSRS & Deck Options — Tuning Retention & Workload

Part 7 covered what the answer buttons do. This part is about the brain behind them. Modern Anki ships with FSRS, a machine-learned scheduler that models your actual memory — and a screen of Deck Options that let you decide how much you want to remember and how much work you're willing to do each day. Get these two things right and Anki quietly runs itself.

📚 What You'll Learn

By the end of this part, you'll understand:

What FSRS is and why it beats the old ease-factor scheduler
How to enable FSRS and set your desired retention — and the workload tradeoff behind that number
How to optimize FSRS to your own review history
The key Deck Options and presets, and which ones actually move your workload
How to read the Stats screen to sanity-check your settings

In This Part

What FSRS Is (and Why It's Better)

FSRS — the Free Spaced Repetition Scheduler — is a modern scheduling algorithm built into Anki since version 23.10, and it's now the recommended scheduler. Where the old system multiplied a hand-tuned "ease factor" against intervals, FSRS is a genuine memory model. For every card it tracks three quantities:

📖 Definition

Stability: how long the memory currently lasts — roughly, how many days until your chance of recall falls to a set threshold.
Difficulty: how inherently hard this particular card is for you, which affects how fast its stability grows.
Retrievability: your estimated probability of recalling the card right now.

Because FSRS predicts your probability of recall over time, it can schedule each card to reappear exactly when your recall chance drops to a target you choose. That's a fundamentally smarter approach than a fixed multiplier.

Why it's better than SM-2

It personalizes to you. FSRS learns from your review history, so its predictions reflect how your memory actually behaves — not a one-size-fits-all default.
It's more efficient. For the same level of retention, FSRS typically schedules fewer reviews than SM-2, because its intervals are better-placed rather than conservatively short.
It largely eliminates "ease hell." There's no ease factor to spiral downward, so a few Hard or Again presses can't permanently sabotage a card's future intervals the way they could under SM-2 (see Part 7).

🧠 Mindset

You don't need to understand the math to benefit — FSRS is designed so that the defaults are already good. Turning it on and leaving retention near 0.90 puts you ahead of most users. Everything below is about fine-tuning, not prerequisites.

Enabling FSRS & Desired Retention

You turn FSRS on from Deck Options (click the gear icon next to a deck, or open a deck and choose Options). Scroll to the FSRS section and switch it on. It applies collection-wide — once enabled, FSRS schedules every deck, not just the one whose options you opened.

Desired retention

The most important knob FSRS gives you is desired retention. The default is 0.90, which means: at the moment a card comes due, you want about a 90% chance of still recalling it. FSRS then spaces every card to hit that target. The number is a direct dial on the tradeoff between remembering more and working less:

Desired retention	What it means	Effect on workload
0.95 (high)	You'll remember ~95% of cards when due — you forget less.	Much more work: shorter intervals mean noticeably more reviews per day.
0.90 (default)	Remember ~90% when due — a well-balanced sweet spot.	Recommended for almost everyone. Leave it here unless you have a reason.
0.85 (low)	Remember ~85% when due — you forget more often.	Less work: longer intervals, fewer daily reviews, more "Again" presses.

⚠️ Important: Pushing retention from 0.90 to 0.95 does not cost 5% more work — it can cost dramatically more, because the last few percent of retention require much shorter intervals. Chasing 0.97+ is a classic way to drown yourself in reviews for a tiny memory gain. Unless you're prepping for a high-stakes exam, leave retention around 0.90.

Optimizing FSRS parameters

FSRS ships with sensible default parameters (the weights of its memory model), but it works best when those parameters are fitted to your review history. In Deck Options there's an Optimize button (often shown as Optimize All Presets); clicking it re-fits the parameters to the reviews you've logged. Do this periodically — the more review history you have, the better the fit, so it's worth re-running every month or few thousand reviews. If you're brand new and have little history, that's fine: start on the defaults and optimize later once you've built up data.

✅ Tip

Optimizing changes future scheduling, not your cards or your history — it's completely safe to click. There's nothing to lose by running it, and Anki will simply tell you if you don't yet have enough reviews for a meaningful re-fit.

Deck Options & Presets

Deck Options aren't stored per deck directly — they live in presets. A preset is a named bundle of settings that can be shared across many decks. This is powerful and occasionally surprising: if two decks use the same preset and you edit it, both decks change. When you want one deck to behave differently, clone the preset (the dropdown at the top of Deck Options) and give it its own copy.

⚠️ Watch Out

Before you change a number, check which preset the deck is using and whether other decks share it. A lot of "why did all my decks suddenly change?" confusion comes from editing a shared preset without realizing it. Save with the Save button (top-right) — changes aren't applied until you do.

Here are the settings you'll actually reach for, and what each one does:

Setting	Default	What it does
New cards/day	20	How many brand-new cards Anki introduces per day. This is the biggest lever on your long-term workload — lower it if you're overwhelmed.
Maximum reviews/day	200	Caps how many due review cards you're shown per day. Usually leave it high enough to clear everything; a low cap just hides reviews and builds a hidden backlog.
Learning steps	1m 10m	The short intra-day steps a new card passes through before graduating to Review. "1m 10m" means one minute, then ten minutes.
Graduating / lapse behavior	varies	Controls what happens when a card graduates from Learning, and what happens to a Review card when you lapse it (relearning steps, minimum interval).
New card insertion order	Sequential	Whether new cards are introduced in the order added or shuffled.
Display order	varies	The order cards are shown within a session (e.g. new cards before/after/mixed with reviews).
Desired retention (FSRS)	0.90	Your target recall probability at due time — the main FSRS dial (see above).
Maximum interval (FSRS, optional)	~100 yrs	A ceiling on how far out any interval can be pushed. Rarely needs changing; lower it only if you want to be forced to see very mature cards more often.

Of all of these, only two really change your life: new cards/day and (under FSRS) desired retention. The rest are refinements you can leave at their defaults for a long time.

The Workload Truth & Stats

Here's the single most useful thing to internalize about Anki: your steady-state daily review count is driven mostly by two things — how many new cards you add per day, and your retention target. Every new card you introduce today becomes a stream of future reviews. Add 20 a day and, after a few months, you settle into a manageable daily load. Add 200 a day because you're excited, and you are writing a check your future self has to cash — the reviews compound and eventually bury you.

⚠️ Watch Out: The Enthusiasm Trap

New users almost always crank new cards/day too high in the first week, feel fine (new cards are easy), and then get crushed three weeks later when those cards mature into a wall of daily reviews. The workload you feel today is not the workload you're signing up for. Add sustainably.

Reading the Stats screen

The Stats screen (click Stats from the main screen, or press S while viewing a deck) is your dashboard for sanity-checking all of this. A few charts matter most:

Future Due — a forecast of how many reviews are coming in the days and weeks ahead. If this is climbing relentlessly, you're adding new cards faster than you can sustain. Lower new cards/day.
Retention / True Retention — how often you're actually recalling cards correctly at due time. If it's well below your desired retention, your cards may be too hard or you may need to optimize; if it's far above, you could afford a lower retention target and less work.
Reviews (count & time) — how many cards you've done and how long it took, day by day. Great for spotting whether your daily load is trending up or holding steady.

⚠️ Important: Check the Future Due chart when deciding whether to raise your new cards/day. If the forecast already shows a heavy load ahead, adding more new cards now will only make it worse. Let the forecast — not your enthusiasm — set the pace.

🛠️ How To: Turn On FSRS & Set a Sustainable Pace

What you'll do: enable FSRS, set a sensible retention target, choose a new-card pace you can actually keep, optimize the parameters, and save.

Step by step

On the main screen, click the gear icon next to any deck and choose Options.
Scroll to the FSRS section and turn FSRS on. (It applies to your whole collection, so you only do this once.)
Set desired retention to 0.90 — leave it here unless you have a specific reason to change it.
Set New cards/day to a number you can sustain including the reviews they'll create later — for most people starting out, 10–20 is plenty.
Click Optimize (or Optimize All Presets) to fit the parameters to your history. If you're brand new, skip this for now and run it in a few weeks.
Click Save in the top-right corner to apply everything.

💡 Tip

Start conservative and increase, never the other way around. It's painless to bump new cards/day up when you have spare time, but cutting it back after you've buried yourself means living with a backlog for weeks. Set it low, watch the Future Due chart for a couple of weeks, and raise it only if the forecast stays comfortable.

Best Practices

✅ Do's

Turn on FSRS — it's the recommended scheduler and it's smarter than the old system for almost everyone.
Leave desired retention near 0.90 unless a specific goal justifies changing it.
Set new cards/day to a pace you can sustain with future reviews included — start low (10–20).
Re-run Optimize periodically as your review history grows.
Check the Future Due chart before raising your new-card limit.

❌ Don'ts

Don't crank retention to 0.95+ "to be safe" — the extra work is huge and the memory gain is small.
Don't set new cards/day to 100–200 in a burst of enthusiasm — you're scheduling your own burnout.
Don't edit a shared preset without checking which decks use it — you may change all of them at once.
Don't cap Maximum reviews/day too low to "feel done" — it just hides a growing backlog.
Don't forget to click Save — unsaved changes don't take effect.

💡 Pro Tips

Think of new cards/day as a "reviews faucet" — every new card you add today drips into your daily load for months.
If your True Retention sits well above your target, you can safely lower desired retention and reclaim time.

📓 Learning Journal

Keep a journal as you work through this guide — digital or paper. After each part, jot down:

Key ideas you learned
Things that clicked for you
Questions or confusion points to revisit
Ideas you want to try
Your progress and how you feel about it

✍️ This part's prompt: Be honest with yourself about time. How many minutes per day can you realistically give Anki — not on your best day, but on a busy, tired, ordinary day? Work backward from that to a new-cards-per-day number you could keep for months, and write it down. What would it feel like to add fewer cards than you're tempted to?

📝 Part Summary

🎓 Key Takeaways

FSRS (built in since Anki 23.10) models each card's stability, difficulty, and retrievability and is the recommended scheduler.
It personalizes to your history, schedules more efficiently than SM-2, and largely eliminates "ease hell."
Desired retention (default 0.90) is the main dial: higher = more work, lower = more forgetting.
Optimize fits FSRS to your reviews — run it periodically as your history grows.
Your steady-state workload is driven mostly by new cards/day and retention — add sustainably, and watch the Future Due chart.

🎉 What You Can Now Do

You can open Deck Options without flinching: enable FSRS, choose a retention target that fits your goals, set a new-card pace your future self will thank you for, optimize the model to your own memory, and read the Stats screen to check your work. You now control the two dials that decide how much you remember and how much you work — everything else is polish.

❓ Common Questions

If I switch to FSRS, will my existing cards get messed up?

No. Enabling FSRS reschedules future reviews based on your history; it doesn't delete cards or lose your progress. Anki even uses your past reviews to seed FSRS's estimates, so the transition is smooth.

How often should I click Optimize?

There's no harm in running it whenever you like, but roughly monthly — or every couple thousand reviews — is plenty. More history makes the fit better, so early on it matters less than later.

What new-cards-per-day number should I actually use?

For most people, 10–20 is a sustainable start. If you have a hard deadline (an exam) you might go higher temporarily, but remember the reviews compound. When in doubt, start lower and raise it once the Future Due forecast stays comfortable.

Is a higher retention always better?

No. Higher retention means fewer forgotten cards but disproportionately more reviews. Around 0.90 balances memory and effort well for almost everyone; chasing 0.95+ usually costs far more time than it's worth.

🔭 Up Next

In the next part, we'll zoom out from a single deck to organizing at scale — decks and subdecks, tags, and filtered decks — so a growing collection stays tidy and easy to study.

📚 Additional Resources

🌟 Keep Going

You've just learned the setting that separates people who quit Anki from people who stick with it for years: a sustainable pace. Turn on FSRS, keep your new-card faucet modest, and let the scheduler carry the load. Come back to this page whenever it's time to re-tune.