I removed 95% of my product’s features after talking to customers

Sometimes, removing features from your product makes the product better and helps you acquire customers.

My product was a self-hosting AI visibility tool to track a brand's citations across all popular AI models, and we had multiple models from each AI brand.

We also had many other features, such as brand website analysis, a feature to evaluate a brand's sentiment across the internet, and an action plan for brands to improve their AI citations.

We used to host the product for clients on Railway. We had a ready-made Docker file template for Railway, and customers could use it as their own product.

They owned the data and used their own AI model API keys, so they had full control over their token usage. This also gave them good control over their monthly running costs.

Even Railway account management is very simple and does not require any technical knowledge.

Most of our customers were from technical fields. As we planned, we could not crack sales with brand owners and marketing agencies.

Our initial assumption was that our product would be loved by agencies, brand owners, and solo founders because of product ownership and cost control.

But our product was overwhelming and complicated for them.

What customers actually wanted

By interacting with many potential customers, we understood that what they really wanted was to know how brands in their category get citations from AI models, especially ChatGPT.

They didn't want a fancy dashboard and many features.

They wanted a simple answer:

Where are AI models getting information about brands in my category?

That changed how we thought about the product.

The problem with "done for you"

When we thought about pivoting to a "done-for-you" model, we found another problem.

It is still not possible or effective to automate many things, like writing for industry websites, blogs, and review platforms.

There is a high chance of spamming, and it can cause a bad reputation for the brand.

From my understanding, with years of studying AI citations, I found that discovering the sources of AI citations in your category and manually placing your brand content on those sources is the most efficient and fastest way to get AI citations for your brand.

So instead of trying to automate everything, we decided to make the discovery part extremely simple.

We removed 95% of the product

We removed 95% of our product features and pivoted to a very simple and effective method of helping brands get AI citations.

Instead of a fancy dashboard and complicated features, customers get the sources of AI citations in their brand category through Telegram.

No sign-up.

No subscription.

Just send /start in the Mayin Agent Telegram channel and message your brand name and category.

The goal is simple: help brands discover where AI models are getting citations in their category.

Mayin — Find where AI talks about your brand

Sometimes, cutting features is better than adding new features.