Who Maintains the Logic After Go-Live?
Suppose a supplier cuts its unit price from $10 to $9, but doubles the minimum order from 100 to 200 units. The minimum cash commitment rises from $1,000 to $1,800. The software can process the new terms perfectly while making a worse investment.
Lokad’s Supply Chain Scientists are responsible for keeping the decision logic economically sound after go-live. An initiative typically has a lead scientist and a backup who can cover absences or take over if the lead leaves. Clients using this service do not need to recruit their own engineering team.
In this example, a 10% discount raises the minimum cash commitment by 80%. The expected return must justify that commitment.
Maintaining economic integrity
The numerical recipe should favor decisions with a higher expected rate of return, accounting for uncertain demand, lead times and other risks.
For the supplier’s offer above, the scientist needs to examine how long the extra stock will tie up capital, what may remain unsold, and which other purchases the larger order would displace. The recipe must account for these consequences when deciding whether to accept the offer. Updating the price and minimum-order fields only records the new terms. The maintenance work lies in checking that the calculation still values the investment correctly.
The scientist must also follow decisions through to their outcomes. One unsold batch does not prove a purchase was wrong: uncertainty was part of the calculation. But if orders repeatedly earn less than their estimated return, the scientist must investigate. Did deliveries take longer than modeled, were unsold goods overvalued, or was a cheaper purchasing option overlooked? Each finding calls for a different correction to the recipe.
The same work arises when a competitor changes its assortment, customers become less willing to wait, or a new sales channel changes fulfillment costs. The scientist works with the client’s practitioners to understand these changes and revise the recipe. This combination of engineering and business judgment is central to the Supply Chain Scientist’s role.
IT changes continuously in a large company, even between ERP upgrades. A stock field, for example, starts including quarantined goods. The file still loads, but the recipe counts unsellable stock as available. Scientists must track such changes and adapt the way the recipe interprets the data.
Testing and documenting changes
Lokad versions both the data and the Envision scripts that process it. When a result looks wrong, a scientist can retrieve the inputs and code that produced it, reproduce the calculation, and check what changed. This is built into the platform.
A proposed correction can be tested on the same data as the old recipe. In our purchasing example, the scientist can inspect how the larger order affects the remaining budget and the purchases it displaces. Comparing both versions on identical inputs separates the effect of the code change from changes in the business.
The scientist also updates the Joint Procedure Manual, which records the assumptions and reasoning behind the recipe. The next scientist can understand why the recipe treats supplier discounts as it does, alongside the code implementing that reasoning. Lokad allocates time for the backup scientist to stay familiar with the initiative, so that a handover begins with an understanding of the current recipe.
The work after go-live
Lokad generates daily forecasts and decisions unattended. Scientists improve the recipes; they do not babysit forecasts SKU by SKU. They correct misinterpreted data, revise cost assumptions and investigate decisions that put money into the wrong stock.
The client supplies business knowledge and sets priorities. Finance and operations must tell the scientists about a new commercial strategy or service commitment; these cannot reliably be inferred from transaction records. The maintenance section of Introduction to Supply Chain explains this continuing collaboration.
All supply chain software incurs maintenance costs. With configurable products, planners spend time maintaining parameters, exceptions and spreadsheets that cover the product’s shortcomings. That labor rarely appears on the software invoice. At Lokad, maintaining the recipe is an explicit responsibility of the Supply Chain Scientists. Code, versioning and documentation reduce the time needed to trace a problem and revise the logic. The test remains economic: whether the decisions continue to justify the money committed to them.