Machine Intelligence · SaaS sales

How does a customer gain confidence in a SaaS machine learning product?

Vineet Sahni Senior Consultant - Advisory Services at EY

May 1st, 2016

Machine learning is finding its way in many core business principles but one challenge is the end buyer’s / customer's confidence in validation, accuracy and fit of the model to predict and optimize key business functions.

I have worked on tech M&A transactions to perform due diligence for machine learning products and understand these will bring a facelift to traditional enterprise software.

There are workarounds to assess the fit of a traditional model but how do we deliver value and confidence on a real time SaaS machine learning product for accuracy?

We understand customer testimony & past work do make a difference but we are looking for inputs on a sales pitch for prediction & accuracy of the ML model.

Any suggestions, comments, criticism are welcome.

Tom Cunniff Founder at Cunniff Consulting, B2B Brand Consultancy

May 1st, 2016

My experience from marketing companies with machine learning at their core is that confidence can only really be built through customer success stories, third-party validation (Gartner, Forrester,et al), and adoption by companies who have a reputation for being smart and ahead of the curve.

There are three core problems with attempting to sell prediction and accuracy.

First, very few buyers and end-users will have sufficient expertise and patience to evaluate your claims -- even if you were to reveal 100% of the source algorithms and data science practices that go into training your specific AI.

Second, you cannot reveal all of that in any case. Many aspects of what you do will have to remain trade secrets. If you have access to more and better or more unique training data, that's probably a trade secret. If you have proprietary ways of using that data to learn and improve, that's probably a trade secret too.

Third, the real value of any machine learning is that the AI gains intelligence over time. In theory, if the system is good all of your numerical proofs should quickly become outdated.

In the end, buyers -- even those who insist they will be unmotivated by anything but hard proofs -- will find they are forced to use heuristics to make a decision. Does it do things I can't easily do using other tools? Are customers in similar businesses to mine finding success with it? Will they enthusiastically talk about that success? Do companies I admire or respect use this SaaS service? Does the company look and sound like it is smart? Do they provide serious thought leadership, yet speak in language I and the people in my C-suite can understand? Do they look, sound, and act like a sophisticated leader?

Hope this is useful. Happy to jump on a call or exchange emails if that would be useful. Cheers, Tom

David Ronald Vice President of Marketing at Foxit Software, Inc

May 1st, 2016

It's an interesting question. What type of persona you targeting? Have they had any prior experience with sophisticated software products? Will "machine learning" be a completely new concept or one they've had some exposure to?

Massimiliano Versace CEO & President, Neurala Inc.; Director, Boston University Neuromorphics Lab

May 2nd, 2016

Hello Vineet, I have worked on Machine Learning/Neural Networks/Deep Learning for about 20 years, between academic and non academic (startup) environments, in both commercial & DoD. This is the main issue of them all: validation. In general, people, at all levels (from a consumer drone pilot, to the designer of a Deep Learning-enriched SaaS piece of software to crunch medical data, to a platoon commander controlling 20 ground robots, to the engineer working on self-driving cars) want human-like performance from their AI algorithm, either implicit or explicitly. Ultimately, performing at 'acceptable' level against human operator(s) on a specific task is the ultimate proof. Happy to chat more (versace@neurala.com) Max -- *************************************** Massimiliano Versace, PhD CEO, Neurala Inc. 8 St. Mary’s Street, Suite 613 Boston, MA 02215tel. 617.418.6161 x701 fax. +1.617.418.6161 Neurala | TEDx talk | Youtube ************************************** CONFIDENTIALITY NOTICE: This e-mail may contain information that is confidential and proprietary to Neurala, and Neurala hereby designates the information in this e-mail as confidential. The information is intended only for the use of the individual or entity named above. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution or use of any of the information contained in this transmission is strictly prohibited and that you should immediately destroy this e-mail and its contents and notify Neurala. ---- On Sun, 01 May 2016 16:35:03 -0400 Vineet Sahni <reply+dsc+5161@founderdating.com> wrote ---- FD:Discuss New Discussion on How does a customer gain confidence in a SaaS machine learning product? Started by Vineet Sahni Senior Advisory Consultant, Ernst & Young LLP. Entrepreneur, Purdue grad. Machine learning is finding its way in many core business principles but one challenge is the end buyer’s / customer's confidence in validation, accuracy and fit of the model to predict and optimize key business functions. I have worked on tech M&A transactions to perform due diligence for machine learning products and understand these will bring a facelift to traditional enterprise software. There are workarounds to assess the fit of a traditional model but how do we deliver value and confidence on a real time SaaS machine learning product for accuracy? We understand customer testimony & past work do make a difference but we are looking for inputs on a sales pitch for prediction & accuracy of the ML model. Any suggestions, comments, criticism are welcome. FOLLOW DISCUSSION or Reply Directly to this email to participate in the discussion Manage your email notifications

Vineet Sahni Senior Consultant - Advisory Services at EY

May 3rd, 2016

Thanks for the response. 

David - the audience would be sales and marketing folks, IT folks / CTO. Few might have strong experience in this area and the others might be looking into this as a area of exploration.