Reading Time: 10 minutes

by: Zank, CEO of Bennett Data Science

The purpose of this document is to evaluate the benefits of artificial intelligence (AI) to early stage startups that are not AI-specific. It will show how AI can provide the competitive advantage young companies need to help them optimize operations, serve their customers better, and secure important funding. Given the hype surrounding AI, it’s essential to understand what its power really means for early stage startups and how to leverage it effectively, if at all. Here are a few common questions executives may ask:

  1. What does the power of AI really mean for an early stage startup?
  2. Is AI just another buzzword?
  3. How can a young company unlock the transformative potential of AI?

In this post, I’ll answer each of these questions, starting with how I’ve seen the differentiating power of AI help launch Cordial, a very successful messaging company, right here in San Diego.

I’ll also look at ways I’ve seen it go wrong. Really wrong. Next, I’ll address the current buzz around AI. Finally, I’ll talk a bit about the potential of AI and the approaches successful companies take towards optimizing their offerings and changing an industry.

Artificial intelligence will fundamentally transform every aspect of business in the coming decades. Are you ready?

- Fortune Magazine

1. What does the power of AI really mean for an early stage startup?

Here’s a bit of a reality check. Let’s pause to level-set and look at what AI really delivers to companies: When applied properly to your products or offerings, AI leverages your data to give you a competitive advantage. It helps you stand head and shoulders above your competition. AI usually isn’t the product itself, but an enhancement to the product(s). It’s best to think of AI as a tool that’s capable of driving a 30% boost to revenue, not a 3x boost.

The rest of this document will focus solely on startups where AI is not the product or the core offering. Examples are marketing, e-commerce, services, wellness apps, or any company that makes and/or sells products or services.

For companies like these, AI can provide two main paths towards increased revenue:

  1. Process optimization – a roadmap to the most profitable ways to sell products or services
  2. Subject matter expertise – a secondary effect that provides data-driven strategic guidance (most often used by messaging and marketing teams)

Here are some examples where AI provides the lift, not the product.

Segment
Product Offering without AI
Boost Provided by AI
Marketing
Weekly emails
Increased revenue: personalized emails increase engagement
E-commerce
Recommend popular or trending products
Increased revenue: 1:1 personalized recommendations
Services
Online Mortgage Application
Quick/instant qualification notice
Wellness Apps
Report steps or calories burned, etc.
Provide weight loss plan based on similar users
Ride-sharing
Send closest available ride to rider. No concern for supply/demand
Increased rider satisfaction: intelligent supply/demand optimization
Finance
Fraud alerts based on a few metrics and a heuristic
Automated, real time fraud alerts, based on predictions using thousands of signals
Services
One size fits all messaging and sales pipeline
Increase conversion & retention: onboarding data to customize messaging, customer journey and up-sells

Clearly, there’s a lot of opportunity to utilize AI for growth and differentiation. For startups, there’s a question buried in the table above:

With cash in short supply, when’s the right time to add AI?
Here are some example questions to ask before jumping into AI:

  • For a ride sharing app, is it essential to balance supply and demand right out of the gates?
  • For a marketing company is it imperative to send one-to-one personalized emails on day one?
  • If my e-comm company has 1,200 customers and I’m selling, 750 products, what kind of lift would I see by using an AI driven recommender system?

Those are questions best answered by each individual company. But I’ll offer this small piece of advice: It’s a heck of a lot better to start with a simple (non-AI) model based on really simple assumptions than to jump into a complicated black-box solution. We sometimes call simple models heuristics. And heuristics don’t require a data scientist, yet they can get teams focused on using data to drive decisions. This is a huge win, no matter how simple the heuristic may seem.

As a marketing example, sending the same collection of popular products in weekly emails is hardly personalized, but could go a long way towards boosting sales. That’s a lot better than doing nothing, and is a great first step until smarter solutions are required. You’d probably be surprised how many companies rely on popularity as a proxy for personalization! And why? Because it works!


When is a startup ready to use AI?
A good indicator that a startup is ready to unleash the power of AI is when a company starts generating substantial volumes of data that, unfortunately, nobody ever looks at. Each time a user interacts with your product, subscribes to a service, visits your website or opens an email, you are presented with an opportunity to learn something potentially useful and important.

Looking at the trends and patterns in this data can help you to efficiently utilize your scarce resources and allocate them to the initiatives that can deliver the highest value. This desire to working the next most important thing is something we see over and over again in startups.

To answer the “when” question, it really comes down to prioritization. How important is it to increase recommender accuracy by 50% or increase open rates by 35%, compared to building out that new product feature that everyone’s been asking for? That’s a decision for product owners.

So, what does the power of AI really mean for an early stage startup? It means that once the start up has their product on a solid footing, AI can provide the differentiator that increases user engagement, reduces churn and drives incrementally more revenue. But it may not be necessary right out of the gates. Later in this document I’ll talk about how you know when the time is right.

2. Is AI just another buzzword?

I’m biased. This is all I do. All day long. Every day, for 20 years. So, um, NO, of course it’s not a buzzword, doomed to expire and be replaced by the next big thing! AI is getting a lot of notoriety, and frankly, after spending most of my life in this field, some of it is nice to see. But be sure, AI is no panacea. Simply, when done right, AI provides an incredible boost to a company. When done wrong, it’s wasteful at best. At worst, it’s a needless diversion that can drain a start up of its valuable capital.

Let’s look at what some outlets say. The following bullets come from a Forbes article, 3 Reasons AI Is Way Overhyped.

  1. Many CEOs are being scared into caring too much about AI
  2. There are very few examples of high-payoff AI applications
  3. Very few companies can afford or find good uses for AI

I take all these with a grain of salt; next to this article on Forbes sits: 10 Successful Applications Of AI In Business. 😛

Let’s look at some of the reasons AI gets a bad rap:

  1. Inexperienced practitioners – this is a new field. I’ve seen data scientists with two years of experience interviewing for senior data scientist jobs! Wow. That’s hardly ‘senior’! This will settle more as young analytics professionals gain experience. With the emphasis that universities around the world are putting on data science programs, data science certainly isn’t going anywhere, and we’ll have a bigger, more experienced pool to choose from every year.
  2. The hype – wow, this is a big part of it! And when combined with the previous reason, we see a lot of companies hiring inexperienced data scientists and “hoping for the magic”, which invariably leads nowhere good. AI is exotic. It’s new and exciting. And it comes with fantastic visualizations and irresistible infographics. It must be good, right?!
  3. The magic bullet – our belief in a cure-all is as old as humanity itself and our penchants for fads is not going to disappear any time soon. Frequently, CEOs believe the hype and declare that they have found a tool (AI) that will help them to find order in chaos with minimal effort. This couldn’t be further away from the truth. AI will not suddenly turn mercury into gold. However, your data scientists can become your trusted advisors who can tell you where to start digging if you want to find some true gems.
  4. It’s incredibly complex – There’s no doubt that AI requires a complex and deep set of skills to execute well. This complexity can result in strained communication between management and technical teams. And when no one knows or understands what data scientists are working on…that’s definitely bad (but happens all the time!)

It’s expensive– For the aforementioned reasons, it can be expensive. But there’s a positive, when done right, good AI should provide more lasting value than it costs to create.

3. How can a young company unlock the transformative potential of AI?

So where does AI come into the picture? And what value can early stage companies, that aren’t directly developing an AI product, expect to see from AI?

How should young companies proceed? In two words…very carefully!

Is your company ready for AI? Here’s a short checklist to help guide the decision making process.

Do you:
❠ have a relatively stable/mature product or service?
❠ know exactly how to measure the value of your products or services?

❠ have a good understanding of what a boost to these metrics would mean for your business?

If you can answer yes to all these, you might be ready to start using AI to grow your company. Let’s dive into each:

Make sure you have a relatively stable/mature product or service
If your product or service is subject to fundamental change, then there’s probably not much optimizing to be done. In other words, product changes should be done before optimization to your product will provide any value. Young companies pivot a lot. The Lean Startup method of pivoting and failing fast is very popular. The point I want to make here, is to avoid getting caught in the cycle of trying to apply AI optimization to a process that will change fundamentally in two weeks. It’s just better to wait.

Know exactly how to measure the value of your products or services
What are you main KPIs? The ones you report to your board; the ones that mean you’re winning. Data scientists need to know the answer to that question before they can help you. They need something to “optimize”.

This might be:

  • More revenue?
  • Higher engagement (a proxy for revenue)?
  • More weekly active users?
  • Higher user lifetime value (LTV)?
  • More time spent on site?

The answer to this will determine the data and approach a data scientist will use to help you get more of your most important KPIs. I’m always surprised how few companies have a good grasp of the metrics that drive their products or services.

Understand what a boost to these metrics would mean for your business
Ask yourself how much you would pay for a boost in your main KPIs. Keep in mind that double (or half) a small number is still a small number!

If you boosted customer LTV by 100% from $3 to $6, and you only have 10k users, would the additional revenue be worth six weeks of expensive AI work? Probably not. But with an LTV of $400, those numbers become a lot more fun to think about.

However, there are also more subtle ways that AI can help you stand out from the crowd of competitors. Usually, the data you already have at hand can unlock answers to a lot of important questions and help you to put the strategic direction you are following to the test.

Do your current users represent type of users you intended to attract? Why are your customers leaving? What’s the best way to convince your customers to stay? You cannot always put an exact price on this type of knowledge, but you can find remedies to or avoid completely faulty strategic decisions.

AI won’t (usually) invent a new product for you that magically “harnesses the value of big data” to rocket your company revenue to the moon. (Yes, I’ve received that request. More than just a few times!) That’s where the “hype” becomes a pain. It’s just not what we do.

Conclusion

We see young companies succeed when they focus on the fundamentals, while paying close attention to the longer game, making sure to collect and properly store the data for use later.

We’ve had a lot of success bringing AI to early stage startups, but only when:

  • products are well defined
  • there are establish KPIs in place
  • there are enough users to merit statistical approaches

If you think AI is a good fit for your company, please contact us. We’d love to talk with you. Our first consultation is always on us.

Hugs and thank yous:
Big thanks to Cody Barbo of Trust & Will for the inspiration for this article. Cody is a kind person and one of the fantastic leaders that makes San Diego such a wonderful place to live and work.

References:

Zank Bennett is CEO of Bennett Data Science, a group that works with companies from early-stage startups to the Fortune 500. BDS specializes in working with large volumes of data to solve complex business problems, finding novel ways for companies to grow their products and revenue using data, and maximizing the effectiveness of existing data science personnel. https://bennettdatascience.com

 

Signup for our Newsletter