Anthropic says experienced Claude users get better results as AI adoption broadens

New Economic Index report finds longer-term users are more successful, more collaborative, and more likely to bring higher-value work to Claude, while overall usage continues to spread into simpler and lower-wage tasks.

Anthropic’s latest Economic Index report says people who have used Claude for longer are getting better results from it, as usage broadens into simpler, lower-value tasks.

The company’s February 2026 data shows higher-tenure users are more likely to use Claude for work, more likely to bring higher-education tasks to it, and more likely to have successful conversations than newer users. Usage on Claude.ai has become more diversified and more personal, while coding activity continues to shift into the API.

The findings point to a gap between access and effective use, with experience emerging as a factor in how much value users are able to extract from AI tools.

Claude usage diversifies as coding shifts to API

Anthropic says Claude.ai usage became less concentrated between November 2025 and February 2026, with the top 10 tasks accounting for 19 percent of traffic, down from 24 percent.

Coding activity moved further into first-party API workflows, where Claude Code accounts for a growing share of usage. On Claude.ai, coursework dropped from 19 percent to 12 percent of conversations, while personal use rose from 35 percent to 42 percent.

The shift changed the overall value of tasks performed on the platform. Anthropic says the estimated hourly wage associated with those tasks fell from $49.3 to $47.9, reflecting an increase in simpler queries such as sports, weather, product comparisons, and home maintenance.

The report describes this pattern as consistent with a typical adoption curve, where early users focus on higher-value use cases before broader uptake introduces more general-purpose activity.

Experienced users show higher success rates and more work use

Anthropic says users who have been on Claude for at least six months are seven percentage points more likely to use it for work and less likely to use it for personal queries. Their prompts are associated with higher levels of education, and their usage is spread across a wider set of tasks.

The report highlights a difference in outcomes. Higher-tenure users are about five percentage points more likely to have a successful conversation in simple comparisons. After controlling for task type, request cluster, country, model, and use case, the gap remains at around three to four percentage points.

Anthropic says this pattern is consistent with learning-by-doing, where users improve through repeated interaction with the system. The company also notes that cohort effects and survivorship bias may contribute, since early adopters may be more technical and more likely to continue using the platform.

Model choice reflects task value and complexity

The report finds that users adjust model selection based on task complexity.

Anthropic says Opus, its most capable model class, is used more frequently for tasks associated with higher-paid occupations. On Claude.ai, the share of conversations using Opus rises by 1.5 percentage points for every additional $10 in hourly wage linked to a task. On the API, the increase is 2.8 percentage points.

The data indicates that users are matching model capability to task requirements, particularly in more advanced workflows.

Report points to widening gap in AI effectiveness

Anthropic’s findings suggest that differences in how people use AI may shape outcomes more than access alone.

Within the U.S., usage per capita continues to converge across states, though at a slower rate than in earlier reports. Globally, usage remains concentrated, with the top 20 countries accounting for 48 percent of per-capita use, up from 45 percent.

The report identifies experience as a factor in successful AI use, with higher-tenure users more likely to apply Claude to work tasks, collaborate with it, and achieve stronger outcomes. For EdTech and workforce learning, the implication is that AI capability may depend on how quickly users develop effective usage patterns, not just whether they adopt the tools.

ETIH Innovation Awards 2026

The ETIH Innovation Awards 2026 are now open and recognize education technology organizations delivering measurable impact across K–12, higher education, and lifelong learning. The awards are open to entries from the UK, the Americas, and internationally, with submissions assessed on evidence of outcomes and real-world application.