The number of startups focusing on machine learning and artificial intelligence is exploding, both in Israel and worldwide. Plenty of them come through the Battery offices with promises to change the world – so it can be difficult to figure out which companies are actually onto disruptive innovations, and which are trying to build excitement with the latest buzzwords.
What separates the heavyweights from the lightweights? We’ve identified three key considerations when evaluating an AI company. Let’s move through them, from macro to micro.
There’s a large addressable market.
Seventy-one percent of companies that exceed their revenue goals have documented customer personas, and that’s no coincidence. Knowing more about your audience will give you insights into their pain points and why they would buy your product over the competition’s. Ask yourself: Who is your target customer, and how likely are they to buy?
Estimating the addressable market means digging into questions like: how much are my customers willing to spend on my solution? How easy is it for a company to switch providers? How many competitors are there? Are there “look-alike” customers who might be just as motivated to buy this product and expand the overall market?
A promising-sounding idea might not make for a great business if the addressable market is small. Personalized shopping algorithms offer a case-in-point. Amazon has proven the value of personalized shopping, from recommending “buy it again” purchases of consumable items as well as items of interest based on your user profile. A number of startups are creating a machine learning engine that could fit generically into e-commerce sites to personalize shopping. But given that retailers are focused on top-line growth and feeling cost-cutting pressure, the addressable market seems neither large nor inclined to spend. With a great idea but an addressable market of questionable size, spending time with decision-makers within your potential customer base to understand their approval process and available budget could bolster your case for inferring a large market.
They offer a realistic solution to a pain point.
It’s surprising how often pain points are overlooked or underdeveloped in AI startup pitches. Often AI founders emphasize that the pain point they’re addressing is being solved today via a fully manual process. They (the company) plan to build an engine that will automate the entire thing. They’ll emphasize their proprietary dataset they’ve annotated in such a detailed way as to train their AI model efficiently, as well as the early feedback from potential customers willing to try them out in parallel to their manual processes.
Many companies say they’ll use the current financing round to fund R&D and hiring, as well as start building a sales team. If we ask about business model, the answers generally revert to an approximate current cost of the manual process they want to replace, with some discount to make the company’s solution attractive vis-à-vis the competition. But this is putting the cart before the horse. Who cares if you can automate a manual process if it’s not causing serious pain?
All great innovations start with identifying a need—a gap in the marketplace—that will spark demand once the product is released. When done right, this unique angle becomes a competitive advantage and blossoms into a start-up’s value proposition. It answers the question: why are businesses likely to buy your product over someone else’s? The success of a product is often proportional to how well it addresses that pain point, how unique it is in the marketplace, or a combination of both.
Take sales-call automation, an emerging space in which AI helps salespeople optimize what they say to buyers to close sales faster. The key players in that space are addressing a true pain point and laser-focused on solving that problem. Contrast this with the ELK stack hosting space. Many players are enriching their core hosting with machine learning / AI frills that customers just aren’t interested in. Startups whose innovation isn’t truly novel, or whose pain point is obscured by extras, will end up competing for a slice of the market and being commoditized.
They’ve developed a solid business model.
Even the most exciting innovation, when it finally bursts into the market, is subject to the laws of business. It stands to reason, then, that startups that have given enough thought to developing a sound business model early in their process—in addition to developing great new technology–are more likely to succeed. This process starts with knowing and capitalizing on a real value proposition. A startup’s pitch should also include basic financial information, such as how much a customer is likely to pay for the product. Also: What will it cost to develop it? How will you bring the product to market so that you can make a significant profit margin in a reasonable amount of time?
It’s understandable that early-stage founders often don’t think about unit economics when they’re seeking early funding. But they should have a good sense of how they are going to make their business model work financially. Think about how you will develop your business to fulfill customer needs at a competitive price and a sustainable cost.
Consider the many AI companies trying to improve radiology. With MRI technology, radiology imagery is only growing in importance – and radiologists’ workload is huge and growing, too. Doctors really need workflow-automation help. The addressable market could be sizable, and the pain-point is real. But how will you make money today?
You’re selling this product to hospitals or radiology centers who are already paying doctors. How will you convince them to pay for software, too? Will you charge on a per-scan basis? That can get expensive. Should they stop hiring radiologists or cut salaries to pay for your software? Fine, but then the remaining radiologists will go on strike. You have to go to market with the radiologists, not against them. Your business model must address these issues of practical transition and how to make money in the market as it exists now.
While these three points may seem fundamental, startup founders trying to capitalize on AI/machine learning technology don’t always address at least two out of the three. Startups that spend too much time on technical explanations are likely not spending enough time talking about why their product represents a breakout opportunity in the marketplace.
The bottom line is that you don’t want to spend time with investors getting feedback on your technology. You want to spend that time helping the investor to grasp how your product will improve the customer’s cost structure, either by increasing revenue or automating to reduce costs. You want them to feel confident that you have deep insight into the customer’s pain points.
For AI startups, there’s always a transitionary period between inception and maturity. During this period the algorithm will need “babysitting” as it learns from additional edge cases. As a final point, I’d recommend spending a few minutes explaining how the company will operate from a business model perspective during this period. If you believe you will skip this period, spend some time explaining why.
Battery Ventures provides investment advisory services solely to privately offered funds. Battery Ventures neither solicits nor makes its services available to the public or other advisory clients. For more information about Battery Ventures potential financing capabilities for prospective portfolio companies, please refer to our website.