Spaced repetition software — Anki, Duolingo's review system, Memrise — is built on genuinely solid science. Hermann Ebbinghaus mapped the forgetting curve in the 1880s, and every SRS system is essentially a digital implementation of his insight: review things just before you forget them, and you'll forget them far more slowly.
That science is real. The problem isn't with spaced repetition. The problem is with what most learners put into their spaced repetition systems — and what they leave out.
Problem 1: decontextualised cards
The standard Anki card looks like this: "perro" on one side. "Dog" on the other. You see "perro" and say "dog." Correct. Card scheduled for review in 4 days.
What you've actually learned: that the string of letters p-e-r-r-o translates to d-o-g. What you haven't learned: how "perro" is actually used. You don't know its common collocations (pasear al perro, el perro caliente, vida de perros). You don't know whether it changes register. You don't have a memorable sentence anchoring it to anything meaningful.
Research consistently shows that words learned in rich context — with example sentences, collocations, and situational usage — are retained significantly better and are more accessible for production than words learned as isolated translation pairs.
Problem 2: no thematic structure
Many learners — especially Anki users — build decks by adding words randomly as they encounter them. One session adds words from a cooking show. The next session adds words from a grammar book. The session after adds vocabulary from a news article about economics.
Your reviews then look like a random walk through the dictionary: frigorífico → sin embargo → biodiversidad → el codo → gestionar. There's no network being built. Words accumulate without structure, and unstructured vocabulary is significantly harder to retrieve under the time pressure of real communication.
Thematic organisation — all your cooking words together, all your environment words together — mirrors how your brain actually stores semantic knowledge. It creates networks, not lists.
"SRS is an optimiser. It will make you very good at reviewing the cards you put in. It will not fix bad cards. Garbage in, garbage out — efficiently spaced."
Problem 3: reviews become a chore
Anyone who has used Anki for more than a few months knows the feeling: the review pile grows. You miss a day. Now you have 200 cards due. You miss a week. Now you have 600. The SRS system — which was designed to reduce cognitive load — becomes a source of anxiety that actively discourages study.
This isn't a user failure. It's a design problem. Systems built around daily streaks and cascading review piles optimise for engagement with the system — not for the actual learning outcome. The learner becomes a servant to the algorithm rather than the other way around.
Problem 4: passive recognition masquerading as learning
Most SRS flashcard formats test recognition: you see the Spanish word and produce the English meaning, or vice versa. But exams — and real conversations — test production: you need to access the Spanish word from meaning, under time pressure, in a grammatically correct form, in appropriate register.
A learner who has reviewed "gestionar" 50 times in a flashcard may still pause for three seconds when trying to use it in a sentence. The card was always showing them the destination — never asking them to navigate to it.
What to do instead
Spaced repetition is worth using — but it works best as one component of a richer system, not as a standalone method. Here's what actually works alongside it:
The bottom line
Spaced repetition is a good tool. The problem is how most people use it — isolated translations, random decks, passive recognition only, and a review system that eventually becomes more stressful than useful.
Fix the input, and SRS becomes powerful. Keep the flawed input, and SRS just makes you very efficient at the wrong kind of learning.
The best vocabulary learners we've seen use SRS as a maintenance system — not a learning system. They encounter words in context, they write sentences, they practice production, and then they hand the long-term retention work to SRS. In that configuration, it genuinely works.
For a complementary perspective on why context-based vocabulary is so critical, read: Why vocabulary is the foundation of every language skill → And for the science behind how input and output fit into your routine, see Input vs Output →