Misconceptions about Machine Learning consultants affects your business. Staying informed allows you to move forward with confidence and intentionality.
When we hear about the growth in the gig economy, it’s often in the context of individual consumers, but the gig economy is growing for companies too. Companies are feeling more pressure than ever to reduce full-time employee commitments for work that’s not in the core mission in favor of contractors, gig workers, and consultants.
The pressure is even stronger when the work has an irregular demand or requires in-depth expertise.
Machine Learning and Data Science (MLDS) are firmly in that zone for most companies. While the insights, predictions, and automation they bring can be powerful differentiators, they are still complex networks of specialized tasks done by experts who need to update their skills and tools regularly. With that said, business leaders who aren’t quite sure how to work with MLDS consultants can still be reluctant to partner with them.
If this describes you, read on. This article will address a few common concerns that might be keeping you from using a valuable resource.
How is Machine Learning Different?
Consultants aren’t typically hired for MLDS projects because the assumption is machine learning is more intimate than the typical software and data project commonly delegated to consultants.
Most software and data projects use a discrete business process composed of a series of mechanical steps. In short: data in, data processed, data out.
In comparison, Machine Learning interprets data to a degree and at a speed that the human mind cannot fathom. Machine learning is doing just that—learning.
It continuously evaluates ever-changing data and the many, many variables that can be taken into account by the process. It studies the interplay between variables, the previous results, and how those affect predictions. The same machine learning problem can produce markedly different results after each iteration.
Common Concerns on Partnering with Machine Learning and Data Science Consultants
Hiring a consultant makes logical sense to many companies, but their leaders don’t pursue consulting partnerships due to a few important concerns.
We address a few of those concerns here in hopes of making partnering easier.
Your expenses only increase if you also consider re-training, specialized software, and the hardware needed to keep MLDS competitive. Ultimately you will need to make a decision, but this partnership is no different than any other. Let someone else manage the complexity while you pay for what you use in smaller increments, often with greater efficiency than an employee and without the long-term costs or commitment.
- Machine learning models are trade secrets that differentiate a company in the market
- Machine learning consultants may learn too much about a business
Intellectual property rights are a reasonable concern any time you enter into a partnership with another company. What makes things unsettling around MLDS is the unfamiliarity which leads to a knee-jerk reaction to keep everything, everywhere 100 percent locked down.
However, whenever you enter into a partnership, you are likely to have a non-disclosure agreement (NDA). The typical NDA protects your rights to customer and operational business information shared with the partner. Alter the NDA, as necessary, to also preserve precise citations of your business information in case you list any business rules inside of a script. Protect the outputs (fitted models, scoring logic, customer clusters, and more) of embed information about your customers or the way you operate.
Everything else is likely in the shared license zone. The vast majority of what’s left is tools, techniques, and expertise for MLDS. You picked a partner in large part because they had a good set of tools, techniques, and knowledge from their prior work.
I Don’t Want Them to Sell My Groundbreaking Innovation to Someone Else
Technically, the umbrella of intellectual property covers any innovative discovery, but there are a few differences to mention.
The rule of thumb is business practices change every time an employee quits. Very few things are important enough to incur the additional rigor and expense of trade secret protection or patent protection. If you have a patent, it’s already protected. If you have a trade secret, you might consider adding rules for identifying and protecting trade secrets to your NDA.
On the flip side, there might be a hidden opportunity here. If it’s groundbreaking, you might also consider having your partner help you turn it into a product! Your partner might even help share the cost of development for the opportunity.
The fact is, from a partnering perspective, Machine Learning and Data Science engagements are not much different from any other consulting engagement. Just a few tweaks and you won’t have to miss out on the value and be left behind!
About the Authors
Jim Schaller has been in the IT & Management Consulting field for more than 20 years and, before his current role, specialized in clients in the utility industry and their billing systems.
Jim is responsible for all the risk management items in the company. This mostly includes his involvement in all contracts with our clients and partners as well as tax and other compliance issues. He also administers all of the companies benefit plans, including medical, dental, vision, 401(k), etc.
Data & Analytics Practice Lead, Jeff Kanel is a business-oriented leader who brings more than 17 years of industry experience in IT project and team management. His background in statistical analysis, application development, and the full data warehouse lifecycle gives him a distinct edge in managing a technical team.
He has performed in a variety of implementation roles ranging from hands-on BI implementation activities to strategic BI advisement. Jeff has a proven track record of partnering with business stakeholders to identify high-value opportunities and solutions to deliver on senior management priorities.