Project Management

Currently, the majority of our business comes from referrals or repeat business. Although all Gradient team members are empowered and encouraged to seek out new business, the Head of Research and Research Managers have a primary role in securing opportunities.

All opportunities should be written using the SOW Template. To ensure that the proposal accurately reflects the client’s needs and is properly scoped to reflect said needs, we take significant care to write detailed (but not unnecessarily) proposals that enumerate every step of the project, especially the specific analyses (e.g. type of model and the requisite data) and outputs that will be produced.

Once a proposal has been approved and signed by a prospective client, the project will be assigned to a Project Manager to initiate the kickoff process. At this point, the Project Startup Workflow commences.

Gradient takes a different approach to sequencing research projects: we start with the model specification. Then everything else flows downstream from there. This doesn’t have to be the exact specification with the exact variables. It can be a rough sketch, literally, but it must convey how data will be used and what type of data are needed in the model. We do this because we want to be intentional with the data we collect and how we analyze it. This is more efficient, as we do not waste time analyzing data that never even had the potential to add any value, and also avoids a situation where we lack analytical specificity at any point in a project.

Once we’ve determined the model and/or analytical approach, then we create a “ghost deliverable”, which simply outlines what the final product will look like sans actual data. This can be done with simulated data or mock-ups. The intent of this phase is to specify a priori how an analysis will be interpreted and summarized. The client also receives the ghost deliverable for review and approval—this gives them an opportunity to see how the final deliverable will be composed and voice any additional or different needs before data collection has taken place.

At this point, we proceed with designing a survey and/or creating a dataset. And since we’ve already specified the core of the analysis, most of the hard thinking has already been done and we know exactly how the results will be summarized.

This may feel unnatural at first (and still does at times) but has proven to reduce the “drowning in data” feeling that can afflict even the most talented of researchers.

Given Gradient’s remote workforce and project-based contracts, we’ve developed a project management system that is 1) transparent to the entire company, and 2) reduces the risk of simple mistakes (a.k.a. human error).

We plan all of our work in 1 week sprints that start on Monday and end on Sunday. Every sprint is planned in a firm-wide sprint planning session every Monday. The purpose of the planning session is to ensure that every single task to be completed in the sprint is known, recorded, tracked, and linked to any dependencies required to complete the task.

Prior to the start of Sprint Planning, everyone is responsible for updating the issues in their sprint, coming to the meeting ready to share with the group.

We produce a wide variety of deliverable types. The type is entirely dependent upon what the client desires. Some want a scored .csv. Others want a lengthy slide deck. Some want a memo. A few really like dashboards and simulators. It’s up the Project Manager to determine the appropriate deliverable during the upfront synthesis phase.

Feel free to scroll through some of our past deliverables here.

One of the most unique aspects of Gradient is our commitment to Principle #5: an obsession with constant improvement. We bring this to life in the form of kaizen, a Japanese principle that means “continuous improvement.” Kaizen is a sacred act at Gradient and requires a great deal of self-maturity and respect for your colleagues.

Kaizen is how we get better at what we do. Operationally, we meet monthly to discuss where something could have been better—it can be a small annoyance like forgetting to reply to an important email or a major oopsie like inverting the sign on a coefficient when preparing a synthesis. We also review novel successes that may be worthy of incorporating into our standard process. When anyone notices a kaizen, they complete this survey (obviously it has to be a survey) to provide additional context around the issue, which we then discuss as a firm at the monthly kaizen sessions.

This is when we deeply diagnose the problem and develop methods to prevent it from happening again. In many cases, the solution involves more than just a “let’s do better next time” resolution and produces some type of concrete process change to avoid a repeat occurrence.