Skip to main content

Marketing AI: Q&A with Salesforce Einstein’s Jim Sinai

In the age of AI, it’s important to get back to basics.


It feels like Artificial Intelligence is everywhere, integrated and sold whenever possible, often without clear or explained benefits.

Just adding “AI” to your landing page isn’t enough anymore. Every company needs a defined strategy for demonstrating AI value in their product.

Last month we sat down with Jim Sinai, former VP of Marketing of Salesforce Einstein, their integrated AI product suite currently making 2B+ predictions daily for customers.

Let's start with the basics: why do SaaS companies even lead with AI?

Companies are beginning to understand how valuable their data can be. Google and Facebook are each worth billions of dollars because of the reams of data they’ve collected, and enterprise companies are looking to follow in their footsteps.

There are two main advantages to leading with AI in your marketing materials: first, companies will latch onto your AI features as a competitive differentiator against their peers. Enterprise software is sold more than it is bought, which means you need to craft a narrative about why your product is different.

Second, it sets clear expectations that your data will be leveraged and used in some productive way. While consumer apps can afford to hide or mask how they use personal data, buyers of enterprise products are far more sensitive to the risks associated with leaks of their own data to the public or competitors.

Why is marketing AI products more difficult than normal SaaS products?

Marketing AI is difficult because an AI or machine learning (ML) algorithm’s value is unclear on Day 1. A lot of the benefit of great AI features comes through small UX improvements that are hard to measure.

And more robust features need customer ROI metrics. But the dirty secret behind most ROI calculations: they’re wholly dependent on how you currently run your business. If you’ve built great business processes and enable them with software, your ROI will be massive. If you’ve allowed terrible business processes and attempt to solve them with software, everything will fall apart. This problem is even more difficult with AI. You have no proof that your predictions are going to solve each customer’s specific problem until they implement it.

What advice would you give to AI marketers?

Let the product team lead the story. Don’t just do AI for AI’s sake. The teams and marketers that are succeeding are talking about AI-enabled features, not just AI.

We'll be at a world where marketers don't talk about AI in three years, just like we all stopped talking about “mobile” years ago. AI will be commonplace and standard in software.

We'll be at a world where marketers don't talk about AI in three years, just like we all stopped talking about “mobile” years ago.

Jim Sinai

This inevitable decline means one thing: it’s more important than ever to focus on the fundamentals of product marketing. At the end of the day, all of your customers just want one question answered: “Am I going to be successful with this product?”

What’s been most surprising about marketing AI products?

First customers need to see it to believe it. But it’s far more difficult to build AI demos for potential customers. For most form-based software, you could quickly generate or find fake data. With AI tools, there are far more roadblocks building fake datasets that can create meaningful outputs. You’ll often need to partner with a data scientist or engineer’s help.

Another surprising learning: companies need ways to remove unintended biases from their AI after the model has been created.

AI bias is real – if the underlying dataset is biased based on conscious or unconscious prejudices, the AI itself will become prejudiced. One resume screening tool built on years of hiring data found that the “two factors to be most indicative of job performance: their name was Jared, and whether they played high school lacrosse.” Not good.

This is a massive conversation that AI SaaS companies need to involve themselves in. Empower your customers to remove biases – with features that exclude gender or ethnicity from datasets when applicable, for example – instead of hiding your head in the dirt.