Daily Archives: November 18, 2010

What’s the Right Number of Approvals?

In a recent piece by ChainLink Research on how a legal department can add value, the author noted how gaining efficiencies is not only about technology, but about process. Referencing Cisco’s big push to get to “one-approver per function”, the article noted that it’s important to ask what is the real ROI of having additional approvers and what is the related impact on revenue and customer satisfaction. It’s important to ask how much time the extra approvals take and what the time-value of money is for holding up orders for that many extra days. And what is the cost to the organization if approver number 17, who is the least affected by the purchase, decides to reject the order 7 days into the process when the product is needed on day 10?

While it’s probably impossible to build some hard and fast rules that will always apply, it is important to set some ground rules as to when another approval is needed, and when an approval can be skipped or automated. For example, does every order over $10,000 need to be signed by three approvers? What if the order is for four new servers at a cost of $20,000 and the purchase has already been approved in principle in the budget (for an amount up to $25,000)? Should not the CTO’s approval alone be sufficient once the product has been selected (provided proper procurement policies have been followed)?

At most there should be one approver per function, and the approval of functions that are minimally impacted should probably not be required at all if at least one of the approvers is a senior manager or the purchase is not high dollar and at least one of the approvers has deep product and/or service knowledge. And any approvals that can be automated should be. For example, a $500 spend on office supplies for approved products from an approved supplier should probably not require three manual approvals.

Any thoughts as to what the right number of approvers is?

Share This on Linked In

An Integrated View Is Needed, But Integrated Dashboards Are Deadly

A recent piece from ChainLink Research on going from complexity to clarity suggests a “management dashboard” that allows a manager to see the status of the end-to-end supply chain and the potential implication of a decision with respect to its impact on key metrics is the key to getting a grip on your complex supply chain.

It sounds great in theory, but it’s very dangerous in practice. Why? In addition to all the reasons I’ve already given you on why dashboards are dangerous and dysfunctional (in this post and this post), when you start chaining dashboards from different systems, you introduce the following additional risks:

  1. inconsistent views
    Different systems may calculate metrics in different ways. For example, the WMS (Warehouse Management System) may present an on-time delivery rate of 90% while the SIM (Supplier Information Management) System has an on time delivery rate of 85%. Which is right? What if they’re both right? For example, the WMS may calculate on-time as percentage of shipments that arrive on the designated day using arrival time while the SIM calculates the on-time as the percentage of shipments that arrive complete on the designated day.
  2. propagated errors
    What if the dashboards propagate erroneous metrics that are used in calculations to produce even more erroneous metrics? For example, what if the WMS incorrectly calculates on-time using date and not delivery time, and doesn’t capture the reality that everything after 11:00 am is late (as the truck can’t be unloaded during the normal shift if it doesn’t arrive by 11:00 am)? An inflated metric is then passed to the IMS (Inventory Management System) which uses this metric in its perfect on-time metric, which calculates this metric using parts that pass visual inspection but not quality testing. An inflated metric is then passed to the SIM system which might calculate perfect orders using orders that pass initial component testing, but ignore failures or returns within the full integrated QC (Quality Control) testing process.
  3. overconfidence
    The more information you have, the less likely you are to notice missing information. For example, if you have a dashboard that tells you your highest spend categories, current sourcing projects, upcoming payables, on-time orders, missing orders, expiring contracts, current and past-due project tasks, etc. you might not notice that your logistics costs are going through the roof.

In other words, integrated dashboards don’t necessarily improve visibility, but they do increase risk!

Share This on Linked In