If your retail organization is looking to implement or replace your planning or optimization solutions, you have three options.

The point of this article is to help ensure you don’t overlook three important considerations in choosing the best option for your company.

Here are your three options:

• Develop a solution in house
• License a third-party best-of-breed solution
• Rely on an analytical platform that supports your business processes.

Each option has advantages and disadvantages.

Before we get into the pros and cons, let’s be sure we agree on the meaning of an “analytical platform.”

In our context, an analytical platform is software on top of which companies can create their own planning and optimization solutions.

Such platforms typically offer these capabilities:

• Drag-and-drop or low-code capabilities to enable specific business processes.
• Artificial Intelligence (Machine Learning, Data Mining, and more).
• Optimization of merchandise assortments, stock-levels, and the like.

You can build the application through a variety of resources, including these:

• The retailer’s internal team
• A services team that works for the platform provider
• A third-party business partner

Here are the three key considerations you don’t want to overlook in your selection process:

1. Which skills do you need in house?

There’s a big difference between running a successful Proof of Concept (POC) and operating in production mode 24/7.

Maybe the best engineers and data scientists can get a production system running and keep it going. But to do so will incur added cost.

So in building your investment model, be sure to consider these costs:

• The cost of a separate DevOps team to deploy and maintain the production environments.
• The cost and risk of depending on people who are in high demand. Because such people may be hard to replace, you’ll be at high risk if you don’t have backup resources.

Compare these costs to the annual premium you must pay to a platform provider or best-of-breed software provider to supply these services for you.

You will need an internal data science team to analyze the growing volume of data you collect.

Consider a hybrid approach to providing the right skills

Rather than relying on your data science team to do all the work – from designing the system to implementing and running it – a hybrid option may offer an attractive alternative.

Here’s how a hybrid approach might work:

• Your internal data science team creates analytical models for your system pilot. This team also works to keep improving your planning and optimization solution after it’s live.
• Your software provider deploys and maintains the solution.
• Experts on your internal team collaborate with the software provider (or partner) to evaluate best practices and to coach on recent techniques.

2. How much operational risk can you accept?

Ensure compatibility across cloud service providers

Suppose you choose a cloud service provider, then you want to change providers.

If your solution provider doesn’t support the cloud service you prefer, your options will be limited.

Why might you want to change your cloud service provider?

Maybe the cost is higher than you expected. Or you’re not getting the level of support you need.

It’s possible you don’t want to do business with Amazon Web Services (AWS) because you consider them a competitor.

You may have similar concerns about doing business with Google Cloud, because Google influences shopping behavior.

What about Microsoft Azure or another cloud service provider?

To mitigate the risks of switching cloud service providers, choose a solution provider that will work with any cloud service.

Be sure your provider can address critical issues fast

What if you’re unhappy with the support you get from your software provider?

A major bug could affect your stock positions and cause lost sales or overstock.

Will your software provider train your team to take over or help with support?

Have a backup plan to change your solution provider if necessary.

You may decide at some point you want to change to a different solution provider.

The reasons could vary widely, including these:

• The provider doesn’t meet its service-level agreements.
• They don’t provide adequate support.
• They change their product strategy.
• They increase their pricing.
• You have concerns about reliability or data security.

Here some ways to ensure you can switch vendors with minimum cost and pain:

• Be certain that you can easily extract your data.
• Maintain full control over your processes and design documentation.

3. Do you have reliable access to expertise in AI and Optimization?

You want the latest technologies embedded in your Retail Planning and Optimization solutions:

• Machine Learning for forecasting
• Deep Learning for automatic replenishment adjustments

Your organization may not be ready to embrace such technologies right away.

To deploy them successfully, you must have access to experts who can advise you. So be sure the software vendor you choose can provide the counsel you need. Or identify other experienced consultants who can.

The recommendations of your experts will vary with these factors:

• How much and what kinds of data you have
• The quality of your data
• Your team’s ability to understanding and adopting new models and techniques

Note that some advanced techniques such as Machine Learning and Deep Learning will work better with some product categories than with others.

Ideally, the solution you choose should enable you to embed AI techniques as you develop your data, processes, and people.

It may serve you best to start with a simple approach at first, then add more sophisticated techniques in future phases.

You want a solution that allows your internal teams – data scientists, data engineers, and IT – to participate in its continuous development.

Anticipate risks and plan to mitigate them

When you undertake any project to transform or innovate, you inevitably incur risks.

Some risks are easier to mitigate than others.

To recap, you can reduce the risk of choosing the wrong retail planning system by answering these sometimes-overlooked questions:

• Which skills will you need in house?
• How much operational risk can you accept?
• Do you have reliable access to expertise in AI and Optimization?

Implementing a successful Retail Planning and Optimization system requires a true partnership with potential providers.

So think hard about what kind of partner each provider will be over the long term.

Other than product features and functions, what would you add to the list of important considerations in choosing a partner?

About Slim Kallel

Over the last 15 years, Slim Kallel partnered with retailers in the United States and Europe to choose, design, and implement Retail Planning and Optimization Solutions.

Slim is now a senior business solutions manager for SAS Institute’s Paris office, where he focuses on the Retail & CPG vertical.

Before joining SAS Institute, Slim worked for Predictix LCC. He then worked for Infor Corp. after the company acquired Predictix in 2016.

Slim started his career with Bechtel Corporation. Slim holds a Master of Industrial Engineering degree from Georgia Institute of Technology and a Bachelor of Science degree from The National Engineering School of Tunis.

Slim supports Tunisian Start-ups using Artificial Intelligence and is part of the Tunisian American Young Professionals Association.