You are working on a project where, starting from an extraction of a client’s hi
ID: 390152 • Letter: Y
Question
You are working on a project where, starting from an extraction of a client’s historical data, your team needs to produce a model to be deployed to production. From the SOW, you had 8 weeks for DS work, starting on March 5th, and ending on April 27th. The client sent the data in 5 separate tables on March 13th, and upon validation from your team, a new extraction was requested, which was sent by the client on March 22nd. This new extraction still has some issues, but you believe it’s solvable on your side. Besides data issues, your team also realizes that there is no perfect joining condition across the 5 tables, and in particular with the table containing the labels.
On a tight deadline, with your small team already stretched thin, and struggling to deliver one model, you get word that the client is now requesting for a second model (one for card-present and another for card-not-present transactions) to go live on the same deadline.
How do you handle this situation:
How would you communicate and say to the project sponsor? Try to anticipate their rebuttal and contain it.
How would you communicate and say to the head of your company's DS team who isn't involved day-to-day with the project.
How would you communicate and say to your project team?
Explanation / Answer
As mentioned in the case, a model needs to be created which extracts historical data and prepares solution which is ready to deploy. The timelines given are already a stretched and aggressive plan and this has been made further complicated by the delay in handling over of data. For building the Design Specification (DS) a period of 8 weeks starting from March 5th and ending on April 27th. The final version of data was shared by client on March 22nd. The data shared by client is not proper and still has many issues with the proper condition. And in such a scenario, the client has asked to prepare another model.
In such a situation, the delays and the dependencies have to be properly communicated with the project sponsor. There should be fortnightly based governance meeting with the client and sponsor along the business and development leads to have a transparency of the project with all stakeholders. During such a meeting I would highlight the delay and issues faced by the team to the sponsor. In addition to this, I would speak to him along with the development leads to ensure the un-factored delays can be considered to re-baseline the project timelines. The project sponsor would ask me and development to work overtime and complete the project but in such a case, the delays and data validation issues should be properly escalated and CR should be raised to capture the new timelines due to delay from client`s side. Also the additional model which was asked by the client would also have to go through a CR process to ensure the effort which goes into development are captured and rewarded.
I would have frequent discussion with the head of DS`s team to ensure he is the loop while discussing the DS areas of the project. I would ask him for his support to try to complete the project and building the model within the given time frame and additional models. I would also request him to join the discussion with sponsor to ensure the effort which goes into DS are properly highlighted and the delays have been due to the error from client`s side.
With the development team, the discussion will be a tricky one. The timelines are already stretched and the work pending is huge. In addition to this, a new model needs to be created. In such a case I would ask the team to follow and agile way of development and ask them to release the features in sprints as soon as they are ready to test and deploy to ensure the changes are done faster. This would also mean that the team can work on creating the second model parallel while the first model is being tested.
Related Questions
drjack9650@gmail.com
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.