TOPIC: Agile: Team experiment not sustainable in production, What PM do

In an agile environment what should the PM do when the team wants to experiment with a new technology that is not sustainable in production ? It is not sustainable because the developer is a contractor who would be the only one who knows how to use the new software (single point of failure), who has not produced a developer manual for this new technology, and the software may not be compatible with our baseline software that will host this new software (baseline software is periodically upgraded) and our IT department runs patches weekly that may or will affect the new software or related systems. The new software is all custom code. Should the PM allow the experimentation? Company policy is to use OOTB technology with existing software or use the approved developer software where we have all developers trained so that there would be no single point of failure. What should the PM do? (BTW our company is not really in a agile environment. We use predictive and are trying to be agile).

Hi Sharon,
Agile is all about delivering value (business, technical etc.) so the question that needs to be asked is: What is the business value that new technology would deliver? Will it reduce technical debt which will result in improved application performance? In Agile environments these types of experiments are generally conducted via Spikes to see the results. We shouldn't discourage the team from exploring different ideas as long as they are aligned with the strategic goals of the organization. In your example if a developer is a bottleneck since only he/she knows the new tool/technology, its better to have cross training efforts going on first to eliminate this dependency before actually starting a full fledged adoption/implementation. Moreover you have to bring your Business stakeholders on board to get the required funding before undertaking this effort so a proof of concept (generated in a Spike) would help selling the idea if they see the value of it.

Thank you for the detailed response. I requested that they conduct a pilot separate from the production environment. Would that be considered a spike? They are meeting with the stakeholder to "sell" their idea; not sure if the stakeholder realizes that will require funding. I requested that our reliable developer be involved since she is the only developer who has stayed longest and has inherited all past developer projects. Per your advice we will need more than one other person trained. Thank you and I will follow through with your advice.

Since I don't know your current environment in which projects are being carried out (Agile/Predictive/Hybrid), I would call it PoC that your team is planning to do to get the buy-in of the business.
In a typical Agile environment, if the team wants to do an experiment then instead of regular (2/3/4 weeks of iteration) they do 1 week of iteration which is called 'Spike'.
As the PM, I would suggest that you should also be part of this initiative to facilitate the conversation between the team and the business stakeholders. Again, before taking on the initiative its better to cross train other team members as well to eliminate the dependency on one single person.

I agree we should get more people trained at a minimum so prevent a bottleneck. I recommended that but it doesn't seem like they will do that. Since I have no authority (weak matrix) over him, I will sit back and pick up the pieces when the train crashes. Its very frustrating.