How To Pivot An AI Company

The questions to ask yourself when it's not working

Happy Sunday and welcome to Investing in AI! I’m Rob May, CEO at BrandGuard, and host of the AI Innovator’s Podcast. I’m also a very active angel investor in the AI space, and today I want to discuss some of the things I’ve seen work in the past for pivoting AI companies.

I made my first AI startup investment in June of 2015. It’s been over 100 investments later and I’ve been quite lucky that when those early AI companies failed, they teams and/or tech usually got picked up by someone who needed AI talent. A few have turned into nice outcomes. But now that AI is more pervasive, it won’t be so easy to land a failing company somewhere. You may have to pivot. In my experience, this is something many entrepreneurs do poorly, so I want to provide a list of questions to ask yourself if you are going through it that may help you find the right new direction.

1. If your direct model isn’t working, is there a channel model that works? It’s very common for AI companies to struggle to sell direct, but it’s the way most SaaS companies were built so most of the sales and marketing folks in tech these days are most familiar with a direct model. But I’ve invested in a few AI companies that struggled to sell direct but partnered with various types of solution providers as a channel partners and had good success.

2. What value does your data have? The most practical way to pivot is to find a good use case for the existing data. Consider whether you can combine it with other forms of data for some uncommon or non-obvious predictive value.

3. How valuable is your customer as a data generator? Some data sets related to certain types of people or professions are difficult to collect. If you built a good user base or deep customer relationships in a group like that, and they have some trust, consider what data they may have, or be able to generate, that you could get that others couldn’t

4. Is there a data equivalent of lead gen for your business? In the early days of the web, people made a lot of money building simple lead generation businesses. For example, build a mortgage calculator, and sell the user leads to a mortgage company. Building apps or games to collect data sets is a similar opportunity.

5. Can you be a plugin to some other app? VCs don’t generally back lightweight apps that could be features of other larger platforms, so you may have been initially discouraged from this. But that can actually be a great place to start if you can build off that in new directions some day.

In general, my advice is to pivot into your market learnings, not into what you’ve built. There is a temptation to always play into the sunk cost fallacy and say “we have this tool, who can we sell it to?” That works sometimes, but startups are really learning machines and there is a chance you’ve learned a lot. Do the harder pivot into what you’ve learned.

In fact, “pivotability” is something I consider in my angel investments. I go into it thinking the company is probably wrong about many things, and I try to make a guess for how easy it will be to pivot into other areas. Is the tech unique? Is the customer profile valuable and underserved? Is the team fast and flexible? If so, there is a good chance I’ll do the deal even if I’m not crazy about the initial idea.

Investing around pivotability has been wildly successful, and lead to some of my best investments. So don’t be afraid of the pivot. Many of the best companies have one in their past.

Thanks for reading.