The goal of a startup is to find a sustainable, repeatable and scalable business model. And so much of a startup’s success is dependent on the early “beta period” where you provide access (to your product) to a limited group of prospects. If the beta period is a complete flop (no one uses the product, feedback is poor, etc.) it doesn’t mean the startup is a failure, but the company has to be in a position to learn from those setbacks, adjust and try again. If the beta period is a roaring success, then fantastic, you’re most likely ready for the next step (which typically involves opening up access, starting to charge money, etc.)

I’d bet that most beta periods end up somewhere in the middle. And many go off the rails through the process because of a startup’s eagerness to scale (before proving sustainability and repeatability) and a lack of focus.

Getting people to sign up for your beta is getting easier and easier. GoInstant has ~2,000 people on its waiting list. I’ve seen startups with 100x that number. But finding the perfect people or customers in that list is another story. This is where it’s so important to have a hypothesis and assumptions around the ideal customer. This is true whether you’re B2B or B2C (but more relevant for B2B companies.) Without a strong definition of your supposed ideal customer, it becomes too tempting and too easy to hand out beta accounts like candy. Being able to say you have thousands of “customers” at the beginning sounds great and may feel like momentum, but it’s the worst vanity metrics possible. For starters, they’re not customers (unless they’re paying), and secondly, it’s so early in the process you really have no clue if any of them will use your product successfully.

For each beta account that you hand out you want to be actively soliciting feedback and working with them. Adding too many people into an early beta increases your workload, while decreasing your focus. And if the customers range quite extensively, their use cases are different, and their requirements, product demands, etc. are all over the map. Now you’re left confused and uncertain as to what you should be doing, and you’re suddenly building a product with too many masters.

With a strong hypothesis around your ideal customer and ideal use cases, you can be extremely selective around who comes into the beta. Screen your beta customers. Make it clear that its a selective process, and your goal isn’t big numbers of mediocre users, but a small number of insanely passionate and successful users.

Think of it as an initial cohort that you’re going to actively work with. If the results are less than stellar, you can evaluate the reason(s) for that. Maybe the product really isn’t ready, but hopefully the feedback from this small and similar group of beta customers is consistent and can drive the appropriate learning and change. Or maybe they’re the wrong type of customer. You can then find another group/category of beta customers and try again.

You can also let in a few outliers into your beta program — chase a few of the (potentially) interesting but diverse leads — and see what happens. One of these may turn out to be your ideal customer and lead to a lot of significant learning and change for your startup. But only do this in an extremely controlled way. And make it clear internally, that you’re taking on added risk by doing so.

Your initial beta customers have an incredible amount of influence over the direction of your startup, whether intentionally or not. If the group is too big and de-focused you run the risk of losing yourself in the noise. The experiment (which is exactly what a beta program should be) will have too many variables to tease out the most important lessons, and leave you with some great big vanity metrics (like # of users) but incredibly poor actionable metrics that really matter.