Autonomous Commercial Planning Competes in Supply Chain Logistics: Solvoyo Leads with Major Clients Onboard
# Automated Business Planning Is Not Only Possible, It’s Here
Steve Banker
Contributor
The views expressed in this article are those of the author.
I focus on logistics and supply chain management.
September 25, 2024, 14:40 EDT
Updated September 25, 2024, 14:48 EDT
Omer Bakkalbasi, Chief Innovation Officer at Solvoyo, stated that the company’s solutions have already provided automated business planning for companies like Unilever.
Several supply chain planning firms gave me a brief overview of their operations. All companies indicated they are investing in artificial intelligence (AI). Many mentioned using generative AI, an AI capable of creating new content and ideas, as part of achieving automated planning. Automated planning is a type of planning that excludes human intervention from the process.
However, Bakkalbasi asserts that this goal has already been achieved. “We feel we are ahead of our time,” he said. “The people who work with us truly believe in our initial belief that automating supply chains is possible. This has been our vision since 2010. Our clients have realized their dreams and benefited from it.” They have created new efficiencies for these customers, as the plans they create are largely executable.
Solvoyo has a metric called user acceptance rate. This measures the percentage of planners who accept replenishment, transportation, or inventory planning without any changes in delivery times or quantities. A 95% user acceptance rate means that 95% of the planning recommendations were implemented as is.
For one client, A101, Solvoyo reports achieving a 99.5% planner acceptance rate. I spoke with A101 in 2021; this Turkish convenience store retailer has achieved 99% automated planning for all products except perishables.
In a division of the world’s largest consumer goods company, production planning reached an 85% automation level, and proposed purchase order acceptance rates hit 95%. A Unilever department is also on Solvoyo’s journey to automated planning.
However, when presenting these results to many companies, they were skeptical. “When I spoke at Home Depot, a senior executive looked me in the eye and said it was too good to be true. It’s impossible. No one would entrust so much of their planning to an algorithm,” Bakkalbasi recounted.
“But we are not just an algorithm,” Mr. Bakkalbasi firmly stated. “We are a platform. The platform collects data and ensures master data consistency internally. If users change the plan, they record this information. Planners who do not accept plans must choose a reason code to explain why the plan is insufficient. This allows the system to learn and improve engine output quality. It’s a ‘continuous feedback loop.’ Achieving an over 95% planning acceptance rate requires concerted effort from our clients and Solvoyo’s dedicated analysts.”
The improvement process starts with reason codes. Once planners must choose a reason code, acceptance rates significantly increase.
Furthermore, achieving automated planning does not depend on highly accurate forecasts. “I’ve never cared about forecast accuracy levels for 20 years,” Bakkalbasi emphatically stated. “Forecasting is not an actionable item.” You don’t act on forecasts; you act on what you buy. “That’s an action. You produce things. That’s an action. You set target inventory levels. Accept this parameter, that’s an action. You schedule truck routes. That’s an action.”
The solution creates inventory targets by SKU channel based on historical sales and automatically updates them according to changes in weather and other external demand drivers. When forecast accuracy improves, less inventory is needed for just-in-time delivery. “That’s why I don’t care about forecast accuracy.”
The journey towards highly automated planning begins with digitization, then becomes intelligent, and finally autonomous. “And you can’t skip any stages.”
“What does ‘digitization’ mean?” Bakkalbasi asked rhetorically. “You must have a digital platform to capture all relevant data.” And this data must be “internally consistent.” Master data management is the primary requirement for automated planning. If your master data is incorrect, “you cannot succeed on this journey.”
Long-term plans are being developed, as well as sales and operations plans, but these change with circumstances. When supplier availability changes, the platform must immediately know about it. For fulfillment planning, pallet and truck dimensions, number of trucks, and warehouse capacity must be accurate and current.
Secondly, the plan must become intelligent. Algorithms used and customer priority rules must be established. You cannot constantly replace algorithms or alter customer allocation logic while aiming for automated planning.
Intelligence also relates to how solutions are utilized. The platform tracks not only plan acceptance rates but also the frequency of use on different pages. What does this mean? We have many features, many analyses, and many reports.” The system tracks who uses each page and for how long. “What’s the point of making a report that no one uses?” Bakkalbasi explained, “This type of accountability is necessary before you focus on automation.”
Bakkalbasi acknowledged that many clients view automated planning as risky. He admitted that many of their clients are just starting their journey towards automated planning. He was particularly frustrated that they have proven significant success with automated planning in one division of a global company but cannot get other divisions to pay attention.
Visit my website.
Steve Banker
—
Source: Forbes
