How to get the first 10 customers for your AI startup

Reaching out to potential customers. Convincing them to use your product. Signing your first recurring revenue deal.

Hey reader, Prateek here! Welcome to the 💥 free edition 💥 of my weekly newsletter. I write about building startups and careers in AI. Professionals in the early to mid stages of their career tend to find it useful. Feel free to send your questions to me via DM on Twitter. I’m happy to offer my thoughts and advice.


When it comes to building startups, AI has been a phenomenal category. Startups in this category build products that serve various needs such as a data labeling, data monitoring, platform to build models, devops for ML engineers, automating quality check, and many more.

As an enterprise SaaS founder, getting your first 10 customers is paramount. What’s a good playbook for this? I’ve outlined 8 steps here:

1. Do outbound

Outbound works best in the early days. It means you making the effort to reach out to your customers vs waiting for them to come to you by building more features.

You need to cold call people. You need to cold email them. You need to hear one-on-one what a prospective customer has to say.

2. Build qualification criteria

Before you reach out to people, you need to define what a "customer" is to you.

You need to make a list of qualities that an ideal customer would have. For example, let's say your AI product needs access to their sensor data. If they don't collect any data, then they wouldn't be able to use your product. You need to qualify them out of your funnel.

You need to score each prospective customer to evaluate if they're a good fit. Be meticulous about it. It will become a gold mine of information later on.

3. Conduct a good demo

Let's say you reached out to someone and they responded. They fit the qualification criteria. They want to know more. Now what?

You need to show them a product demo. No mockups. No wireframes. You need to have a real demo to show them.

A good demo tells them that this is real and that you're not wasting their time. It can be a simple tool hosted locally on your laptop, but it should be something.

4. Follow up, follow up, follow up

This is where a giant majority of the founders drop the ball. Let's say the prospective customer liked the demo, what do you do next?

You try to follow up a couple of times. If you don't get a response, you tend to leave it at that. Don't do this.

In enterprise SaaS, you need to have the discipline of following up multiple times. It usually takes about 8-10 touchpoints to get a deal done.

5. Analyze a small sample of their data for free

Let's say you do have the attention of the prospective customer. But they're not sure if your product is going to work for their data.

Data in the enterprise world is really messy. The world's best algorithm can't do much if the data is not accessible. Or is really bad. Or is really messy.

What do you do now? You can sign their NDA and ask them to send a small sample of their data. You can present a data readout in the form of a slide deck. This will remove any lingering doubts about whether or not your product is going to work for their data.

6. Present a clearly defined POC plan

Now that you've verified that the data is reasonable, how will you help them evaluate your AI product?

In enterprise SaaS, large companies need a way to test your product. They're not going to sign that big enterprise deal before thoroughly evaluating you. The good thing is that they will pay you to do the POC.

Present a crisp plan that will help them evaluate you. POCs are usually 90 days long.

7. Build a benefits scorecard

In enterprise SaaS, companies buy outcomes. You need to be able to measure the impact of your AI product.

Before they start using the product, you need to establish the status quo. This will help you measure the metrics that matter to them. If you don't do that, then you won't be able to measure your impact.

If there's no impact, then the customer has no reason to sign the longer term deal.

You need to track the metrics during POC. And share it with your customers regularly.

8. Convert the POC into an annual recurring deal

At the end of your 3 month POC, you need to clearly show the impact of your product.

This impact has to be financial. Either on their revenue or their margins. Help them make more money or help them cut costs. If you can do neither, then there's no real reason to continue this engagement.

If you do this successfully, you'll convert a POC customer into a production customer.