How to Evaluate an AI Demand Forecasting & Planning Solution

    

Demand Forecasting and Planning ClipArt

AI Demand Forecasting & Planning solutions are redefining traditional practices. They're improving inventory placement and reducing overall inventory costs, returns, and markdowns. Already prominent retailers such as Walgreens and PACSUN are reaping the benefits of these solutions. But not all solutions are created equally, and people struggle to evaluate them. Many advocate pilots, and while that is a viable option, even free pilots cost your company time and resources. Pilot or not, you need methods to down-select solution providers.

Approach it like other complex software buys
AI Demand Forecasting software is complex, and you must evaluate it like other software purchases. Develop a business case, get references, and find reviews. For forecasting, determine what improving forecast accuracy would mean to your bottom line. The Institute of Business Forecasting & Planning provides a calculator you could leverage. In the end, the more you dedicate to the software & vendor evaluation, the better the outcome.

Does the vendor understand your business?
Success comes down to the people and the software you choose. Does the vendor understand your industry, strategies, and situation? Do they bring in expert insights, opinions, and points of view? Can they clearly explain how the solution will help your decisions, processes, and goals? Overall, step back from the feature and function checklist and evaluate the vendor's ability to drive beneficial change within your company.

Can the vendor simplify the complexity?
Well-architected AI and machine learning products are complicated. The vendors behind them are knowledgeable about the pieces that make up the software, but that matters little if they can't explain the solution in understandable terms. How do you know if they did this successfully? After all their discussions, see if you can do it without them. Can you explain what the solution will do, why, and how? Can you explain how the solution will directly improve your operations? Can you explain how your team will be successful in the solution adoption? Many vendors are eloquent and persuasive in their speech, making you feel great in the moment. But you need to understand it, in your words, for your people.

Does the vendor sell or coach?
Sellers sell. Coaches serve. Vendors must be part of your journey and embedded in your success. Ensuring this outcome isn't found in contract incentives; it's in the attitude. Vendors should educate and assist with the solution and buying process. Great vendors help you make the best decision for your team, person, and company while taking away as much friction from the process as they can. One test, after the evaluation, "Would you be comfortable calling the vendor to ask them questions or to provide advice, even if you didn't select them?"

Is the solution flexible for non-traditional data points?
Master data is essential for forecasting, but solutions need other non-traditional data points like weather or events. The last two years highlighted that there could be many unimaginable demand signals. Others, such as social, viral videos, or fictional television episodes, affect demand in ways people never envisioned even a few years ago. So ask what demand signals are used today and how new ones could be included in the future. The last thing you want to do is entirely replace a forecasting solution because of its inability to adapt.

How is poor data managed?
Data is never perfect. Yes, the better the data, the better the results. No one argues that, but dirty data and forecast anomalies will occur. Knowing how it is handled builds trust in the overall solution. When you understand how a solution operates and how issues are highlighted, you become more comfortable with the AI and machine learning algorithms behind them.

How does the solution fit into your Organizational Framework & Enterprise Architecture?
Every business is unique, and the nuances matter. No matter how similar the company, each has a unique culture, organizational structure, and enterprise architecture. A foundational demand forecast solution must be adaptable to these conditions. Some retailers have a data science team and may want to infuse their expertise into a demand forecast. Other retailers don't have that scale and require a complete forecast to drive their inventory decisions.

Can the forecast create a unified, omnichannel plan?
Too often, data is siloed and distrusted across departments leading to misalignment between finance, merchandising, supply chain, and marketing. It's the age-old challenge of syncing all parts of the business. If promotion planners expect an 80% unit lift from a promotion, but the supply chain group only expects a 60% lift, there's bound to be mistakes and finger-pointing. When evaluating forecasting, see if it can serve as that unified view of demand to drive coordinated, omnichannel operations.

And go
Evaluating AI Forecasting & Planning solutions can be a daunting experience, but these tips will help focus that process. It's valuable to break free from the feature checklist and know the people and company that will work alongside you. The core of every inventory decision is based on forecasting the demand; you want to get it right.