Anthropic's Marketing Plan

Anthropic has the best coding model in the world ¹, and nobody knows it. That’s not a technical problem, it’s a marketing problem.

Overall I’ll attempt to do my best Ben Thompson impression - they’re right in the middle of trying to be a platform, yet also an aggregator. Below I’ll go over some areas of success and opportunities that they have.

The Aggregator Approach -> Becoming the Go to coding/writing LLM tool

Anthropic runs Claude and Claude Code. It depends on who you ask, but Claude has one of, if not the top coding model in Opus 4 and Sonnet 4. Claude Code is their command line tool, which uses the Claude models to work through your directory and write code programmatically. Again, it’s tough to really know for sure, but Claude Code is definitely a top power tool, with the integration between model and tooling becoming more and more powerful.

If the goal is to become the go to, top coding tool, there’s a few changes I would implement immediately.

1. Implement student pricing and better target universities
2. Start a marketing campaign with two thrusts: privacy and personal

In order:

This coding workflow is competing against tools such as Cursor and Windsurf. Cursor recently began offering student pricing, while Windsurf has experimented with it previously. While Anthropic has made some general moves in this area, there is room to be the leading tool for computer science programs, and the inspiration here can be found with law school.

In law school, there are two main tools for looking up law review, Westlaw and Lexus Nexus. These two tools are so deeply integrated into the whole process, with classes being designed around how to use these tools and deeply discounted pricing for students.

Anthropic could come up with a similar strategy. While they already have some university partnerships, an education-specific product, and a student builder/campus ambassador program, there’s more they could do here with offering a student discount and more deeply integrating with specific coursework.

Marketing Approach

While working on coursework and getting student adoption will help, another initiative that would help is a more traditional marketing effort. I would want to focus on two efforts - LinkedIn for influencer marketing for corporate VPs / SVPs / Directors for coding in the B2B world.

Secondly, I think some brand marketing could actually come into play. Anthropic has some good advantages going for it when compared to OpenAI / Google. My two thrusts would be around Privacy and Personal. With OpenAI training on the ChatGPT data and Anthropic not training on the Claude data, that is a real point in their favor.

I also think there could be a campaign around being your personal AI. This is again based mostly on vibes from smart people, but the Claude models have a certain smell. A bit more personal, a bit more of a better writer, etc. You can imagine that would be a great TV ad, something very top of funnel.

The fact that Anthropic is already hiring for some paid media roles shows that they understand there’s a need here. But paid media itself won’t solve this awareness problem - they need a cohesive brand strategy to better steer those ad dollars.

The AGI path of the platform

Option 2 is for Anthropic to be the path of the platform. So this is the full steam ahead path to keep competing with everyone and push the frontier of AGI capabilities.

The first thing I would do is to go broad. Rather than not giving access to Windsurf, it would be providing a layer of intelligence across all tools. Work with tools like Windsurf and Cursor, but also tools like Aider, Roo, other command line tools that compete with Claude Code. Why keep your insights for yourself, go ahead and share best practices to further solidify your place as a layer across all sorts of intelligence.

I think there’s also a really interesting developer first strategy here. You could lean even further into the API business, reduce pricing to match Google’s strong value, have better pricing for startups, sponsor hackathons as well as student-led initiatives. The goal should be to have Claude be the first tool people think of when trying to build something.

I would also try to compete with OpenAI and Google in terms of having more models available. For instance, Anthropic recently put out a v2 of their voice model using ElevenLabs. The old version was not great, and whisper has been open source and basically solved it forever. The new one is definitely better! Imagine being able to move this all in house - it would be a terrific value add.

I also think an image model could be powerful for helping with multimodal workflows - anything to start getting more data and figure out what people enjoy.

The Strategic Confusion in Action

The current growth tactics reveal the core tension. Anthropic is running consumer referral programs - ‘Win 4 months of Claude 4’ ($400 value!) - while simultaneously trying to be infrastructure for other tools. This is growth hacking 101 for a B2C app, not how you build the AWS of AI. You can’t referral-program your way to being a platform."

There’s also a point to be made on pricing. I can try OpenAI and Google’s Deep Research tools for free, but Anthropic only recently allowed Claude Code and Deep Research to come to the $20 subscription. I’m sure it’s worth it, but if it’s behind a paywall for a majority of people that means it doesn’t exist. Remember, the attempt is to go wide - I’d consider reducing pricing of the API to at least be competitive with Google and OpenAI’s recent O3 price cuts, and maybe reduce the pricing of Claude as well.

The Windsurf Dilemma

The recent Windsurf situation perfectly illustrates this confusion. When reports emerged that OpenAI was acquiring the AI coding tool for $3 billion, Anthropic abruptly cut off Windsurf’s direct access to Claude models. This forced Windsurf users into expensive “bring your own key” solutions while competitors like Cursor maintained direct access. The move makes sense if you’re protecting a competing product (Claude Code), but contradicts the platform strategy of being infrastructure for all.

Wrapping up

Overall, there’s this dichotomy that I find super fascinating. It’s a tough line to walk to have both the consumer relationship as well as the broad platform relationship. While some fixes are apparent for both strategies, OpenAI is such a monster in the category that having a distinct strategy to differentiate is critical.

However, this strength in coding does appear to be one main point of differentiation. By really leaning into that, Anthropic can carve out a niche and gain further share with performance. And the performance seems to agree

Anthropic would be better served by picking a lane. Either go all in as the developer’s AI company (axe the referral program, come up with a solid student plan, enhance the APIs even further with expanded SDK support across more languages) or push to be a broad based intelligence layer (cut the price on the API, partner with every tool, even ones potentially getting bought by OpenAI, and don’t overindex on Claude Code’s success).