Disciplined Agile

Improving SAFe’s WSJF

Weight Shortest Job First (WSJF) was defined by Don Reinertsen as a way of ordering work to be done to lower overall cost of delay. SAFe modified Mr Reinertsen’s definition in WSJF 3 ways:

  1. business value is defined as business value,  time criticality, risk reduction and/or opportunity enablement,
  2. using relative numbers instead of value (and normalizing and adding these together)
  3. using effort instead of time

There are significant challenges to these changes as well as a way DA FLEX’s Minimum Business Increments (MBIs) can improve the business value calculation.

Defining Business Value.  Adding time criticality, risk reduction and opportunity enablement is a positive step. However, business value as a single number does not help go beyond the current challenge that different business stakeholders have different ideas of what business value is. DA FLEX calculates business value by taking the objectives in the Lean Portfolio section and using them as what to measure business value on. This provides a consistent measure of dimension across the stakeholders.

Using relative numbers.  By bringing the measures down to one very odd things can happen. Unitless measures should not be combined in this way.  It’s possible that all of the business values are higher than any of the time criticality, but if the time criticality has very high variance, the time item with the highest time criticality could come out on top. For example, consider I have time criticality varying from 1 cent to 10 cents while business value varies from 1 to 4 dollars. Some of the high criticality items will come out on top regardless of their business value.

Using Effort Instead of Time.  There are two challenges with this. First, considering effort takes us out of the paradigm of lean – which is focusing on time not effort. More to the point, Cost of Delay is about delay, not effort and the two are not always correlated (team size can vary slowing down or speeding up delivery).

WSJF on epics and features doesn’t make sense.  One of Don Reinertsen’s most valuable pieces of advice is “If you only quantify one thing, quantify the Cost of Delay.” This means to look at the loss of value realized incurred by a delay.  That is, if a release is delayed by two months, and the value we would achieve is $100,000 a month, then our cost of delay is $200,000.

WSJF is a great method of sequencing the work to be done based on cost of delay. However, it requires that the items being sequenced have realizable value. A feature that does not provide customer (internal or external) value by itself  will, by itself, not affect the cost of delay. While it does have value for the purposes of feedback, this is not cost of delay, nor what Reinertsen means. WSJF should only be applied to items that will realize value when released. Doing WSJF on epics ignores the fact that we’re not releasing the entire epic.

Since the features that are not MBIs have no realizable value for the customer, doing WSJF on features is a non-sequitur. This focus on features also has an unintentional side effect in that this has people focus on finishing features and not finishing value that is deliverable and realizable. One can counteract this with an MBI Mindset. It is important that sequencing of work be based on value realized to customers or internal personnel. This is another reason for using MBIs, they provide us with a focus on value realized which improves alignment.