CiteLadder asked Claude Sonnet 5.5 20 buyer questions about language learning apps. Each question ran three times in three conditions: from memory, with web search available, and told to search first. Duolingo was the brand Claude named most often in every condition. When Claude searched, its shortlists got shorter and some names dropped away. This page reports what Claude said through the Anthropic API. It is not a ranking of which app is best, and the Claude apps can answer differently.
What we asked Claude
The 20 prompts read like real buyer questions. They covered five kinds of intent: best overall, a specific buyer or use case, budget or free, feature-led, and comparison shortlists.
Some examples, quoted exactly:
- "What's the best language learning app right now?"
- "Best app to learn Japanese including kanji for a total beginner?"
- "Language learning platform a company can offer employees for training?"
Other prompts asked about apps for kids, seniors, business French, Korean, offline use, AI conversation practice and pronunciation feedback. A separate model listed the brands in each answer. Code did the counting. The requests gave no current date.
Brands Claude named from memory
With no tools, Claude gave 60 answers. It named 91 distinct brands, with 7.9 brands per answer on average.
Duolingo appeared in 68% of answers. It was named first 27 times, with an average position of 2.3. Pimsleur followed at 57%, Babbel at 55% and italki at 53%. Anki was fifth at 48%, though it was never named first. Preply came close behind at 47%.
Consistency across the three runs was moderate. Run agreement was 54.9%. Pimsleur, Babbel and italki were each named in every run for 9 prompts. Duolingo was named in every run for 8 prompts. Memrise appeared in 32% of answers but was named in every run for only 1 prompt. Lower on the list, the names shifted more from run to run.
When search was only available
When web search was available but not required, Claude did not search at all. It ran 0 searches across 60 answers. Every answer in this condition came from memory.
The brand order stayed close to the no-tools results. Duolingo appeared in 75% of answers, Pimsleur in 70% and Babbel in 67%. These answers named 129 distinct brands, with 9.8 per answer on average. Since no searches ran, these differences reflect normal variation between answers, not new information from the web.
For this category, a brand's position in Claude's answers depends on what the model already knows, unless the user or system asks it to look things up.
What changed when Claude was told to search
When the system prompt told Claude to search first, it searched in all 60 answers. It ran 266 searches in total.
The answers got shorter. Claude named 5.6 brands per answer on average and 83 distinct brands overall. The top three brands took 29.8% of all mentions, up from 22.7% from memory.
Duolingo stayed the leader at 60% of answers, but it was named first only 11 times. Babbel moved to second at 55% and was named first 8 times. Pimsleur appeared in 52%.
Four of the five brands at the top from memory stayed there: Duolingo, Pimsleur, Babbel and italki. italki fell to 25% of answers. Anki left the top five, dropping from 48% from memory to 15%. Preply entered the top five at 35%, down from 47% from memory. It moved up because others fell further.
Some names appeared that were not in the top ten from memory. Speak appeared in 17% of answers, Talkpal in 13% and Rosetta Stone in 13%. Run agreement was 51.4%.
What Claude searched for
Claude's queries were mostly short and generic. 59.9% of queries contained the word "best". Examples, quoted exactly:
- "best language learning apps 2025"
- "Duolingo vs Babbel vs Pimsleur which is best"
- "best completely free apps to learn German 2025"
For narrower prompts, Claude often named specific brands inside its queries, such as "best Japanese app kanji Renshuu WaniKani Anki comparison" and "Deutsche Welle Learn German app free Nicos Weg". This means brands it already knew shaped what it looked up.
37.8% of queries included a year. 2025 appeared in 100 queries and 2026 in 1. The requests gave no current date, so Claude chose the year on its own. Pages that rank for "2025" queries were more likely to reach it.
Sites Claude's searches returned
The sites that came back most often were review blogs, app makers' own content and reference pages. polychatapp.com appeared in the results for 35 answers, or 58% of searched answers. en.wikipedia.org followed at 53%, then taalhammer.com at 50% and ilampadmanabhan.medium.com at 48%. lingq.com and testprepinsight.com each appeared for 47% of searched answers. apps.apple.com appeared for 40%.
Several of these sites belong to language learning companies that publish comparison articles. Claude did not add inline citations to any of its answers. The sources shaped the answers, but readers would not see them linked.
What a language learning brand can take from this
- Memory matters most by default. Claude never searched when search was optional. Your presence in its training data decides most answers.
- Search narrows the list. When Claude searches, it names fewer brands. Brands with a steady place in comparison articles held up. Anki fell sharply.
- Comparison pages carry weight. The sites returned most often were "best apps" round-ups and competitor blogs. Being covered on pages like these gives Claude something to find.
- Dated content gets found. Many queries carried 2025. Keeping round-ups and comparison pages current with the year helps them match.
- Use-case prompts open doors. Prompts about kanji, free German or AI conversation brought in niche names. A clear use-case story gives a smaller brand a way in.
CiteLadder tracks these patterns over time, so you can see where your brand appears and which sources feed the answers.
Brand shares in each condition
Share of the 60 answers (20 prompts, three runs) that named each brand. "Prompts named in every run" counts the prompts where the brand appeared in all three runs from memory.
| Brand | From memory | Prompts named in every run | Told to search |
|---|---|---|---|
| Duolingo | 68.3% | 8 | 60% |
| Pimsleur | 56.7% | 9 | 51.7% |
| Babbel | 55% | 9 | 55% |
| italki | 53.3% | 9 | 25% |
| Anki | 48.3% | 5 | 15% |
| Preply | 46.7% | 8 | 35% |
| Memrise | 31.7% | 1 | not in top 15 |
| Busuu | 28.3% | 2 | 21.7% |
| Tandem | 25% | 3 | not in top 15 |
| HelloTalk | 25% | 3 | not in top 15 |
| Speak | not in top 15 | 16.7% | |
| Talkpal | not in top 15 | 13.3% | |
| Rosetta Stone | not in top 15 | 13.3% |
Sites Claude's searches returned
When told to search. Share of answers whose searches returned at least one page from the site.
| Site | Answers | Share of searched answers |
|---|---|---|
| polychatapp.com | 35 | 58.3% |
| en.wikipedia.org | 32 | 53.3% |
| taalhammer.com | 30 | 50% |
| ilampadmanabhan.medium.com | 29 | 48.3% |
| lingq.com | 28 | 46.7% |
| testprepinsight.com | 28 | 46.7% |
| apps.apple.com | 24 | 40% |
| lingopie.com | 22 | 36.7% |
| copycatcafe.com | 22 | 36.7% |
| guide2fluency.com | 20 | 33.3% |
The 20 prompts
- What's the best language learning app right now? (best overall)
- Which app actually gets you speaking a new language fastest? (best overall)
- Most effective app for learning a foreign language from scratch? (best overall)
- Top rated language apps that people actually stick with long term? (best overall)
- Best app to learn Japanese including kanji for a total beginner? (for a specific buyer or use case)
- What language app works well for kids around 6 to 10? (for a specific buyer or use case)
- Best app for brushing up business French before a work trip? (for a specific buyer or use case)
- Which app helps intermediate learners finally get conversational? (for a specific buyer or use case)
- Good language app for seniors who aren't very tech savvy? (for a specific buyer or use case)
- Best app to learn my partner's native language, Korean? (for a specific buyer or use case)
- Language learning platform a company can offer employees for training? (for a specific buyer or use case)
- Which app is best for learning a language in 10 minutes a day on my commute? (for a specific buyer or use case)
- Best free language learning app without constant paywalls? (budget or free)
- Cheapest way to get one-on-one lessons with a native tutor online? (budget or free)
- Are there any completely free apps for learning German that are actually good? (budget or free)
- Which language apps use AI to have real conversations with you? (feature-led)
- App that focuses on pronunciation with speech recognition feedback? (feature-led)
- Language apps that work offline for travel without wifi? (feature-led)
- Compare the most popular language apps — which is worth paying for? (comparison shortlist)
- Gamified language apps vs audio-based courses, which ones should I shortlist? (comparison shortlist)
Frequently asked questions
Which language learning app did Claude name most often
Duolingo. Claude named it in 68% of answers from memory and 60% when told to search. This shows what Claude said, not which app is best.
Did Claude search the web when it was allowed to
No. With search available but optional, Claude ran 0 searches across 60 answers. It only searched when the system prompt told it to, and then it searched in all 60 answers.
Which brands lost ground when Claude searched
Anki dropped from 48% of answers from memory to 15% and left the top five. italki stayed in the top five but fell to 25%. Preply entered the top five at 35%.
Do these results match the Claude app
Not necessarily. CiteLadder ran this study through the Anthropic API with low effort and no current date. The Claude apps can use different settings and may answer differently.
Track how often AI names your app
CiteLadder tracks your brand's mentions, position and competitors across Claude, ChatGPT, Gemini and Google AI Overviews for the buyer questions you choose. Every result is tied to its recorded answer, so you can see how your app's shortlist position shifts over time.
Visibility over time
| Series | 8 Sep | 15 Sep | 22 Sep | 29 Sep | 6 Oct | 13 Oct | 20 Oct | 27 Oct |
|---|---|---|---|---|---|---|---|---|
| Acme Analytics | 27.9% | 29.6% | 30.2% | 31.8% | 32.5% | 33.1% | 34.2% | 38.4% |
| Northwind | 46.0% | 45.2% | 45.8% | 44.6% | 44.1% | 43.5% | 42.9% | 41.7% |
| Lumen | 30.4% | 31.0% | 29.8% | 30.6% | 29.9% | 29.2% | 28.8% | 27.5% |
412 answers to 120 prompts across 4 engines, week of 27 Oct – 2 Nov 2026
Brands
| Brand | Visibility | Change |
|---|---|---|
| Northwind | 41.7% | down 1.2 pts |
| Acme Analytics | 38.4% | up 4.2 pts |
| Lumen | 27.5% | down 1.3 pts |
| Fathom Labs | 16.9% | down 0.2 pts |
Acme Analytics by engine