Four Kind of Team Structure for Analytics and pros & cons associated

I am Gopal, a keen learner, trainer and observer of analytics practices. My youtube channel (https://www.youtube.com/user/gopalprasadmalakar12/videos) and udemy publications (https://www.udemy.com/user/gopalprasadmalakar/)  will give you more detail about my interest in analytics.

I have seen four different team structure prevailing in organizations for analytics. Let me describe you each one along with it’s pros & cons.

Structure 1 : Process oriented team –

Here you will find modelling team, policy team, MIS team and so on. Within each vertical you will have people aligned with different business verticals (like credit card, debit card, home loan so on). Usually each process sits together (like one area for modelling, one area for policy and so on)

Pros

Such team usually have strong process capabilities. Like if you are talking of modelling, they will have

?      Strong variable selection skills

?      Great Model validation process

?      Availability of highly automated procedures / macros / tools for modelling

?      Availability of subject matter experts on statistics, econometrics ..

Similarly if you are talking of reporting, they will have

?      Strong automation skills

?      Great efficiency in report deployment

?      Availability of subject matter experts on automation

Cons

• Most of the time, such team lack business understanding, due to which they end of working on some obvious stuff for very long
• For business, at times this kind of team generates statistically great stuff, which are of little business use
• Lack of employee motivation. A number of them feel that they are more of machine during a repetitive job.
• High employee attrition – due to above

Way out

• Such team needs to have fortnightly presentation on business details
• If this is a CoE kind of environment, they should do cross functional meeting on project before starting to work on them. They need to know, why they are being asked to work on? What is the business need?

Structure 2 : Function oriented team –

Here you will find Card analytics team, Home loan analytics team and so on. Within each vertical, you will have people doing acquisition, portfolio management, collections analytics etc.

Pros

Such team usually have strong business understanding. Like if you are talking of card team, they will know

?      Which metrics are bothering business

?      What is the need of model here

?      How the model should be used

?      Availability of resources – who are good at business understanding

Cons

• Most of the time, such team lacks strong process skills, due to which they apply crude way of doing things. Like they

1. Won’t know advance variable selection process.
2. Model validation process
3. How to generate model statistics quickly

• Their models can be suboptimal
• They lack subject matter experts on process
• Great amount of disparity of skills in various team due to which the variance in time taken by teams to do similar jobs can be huge. One such examples can be that one team can take two days to develop classification tree along with volume and bad rate validation, whereas other teams can take 2 months.
• Lack on process innovation – they keep following outdated techniques and procedure
• Difficulty in skill development – a vicious circle, where similar folks are hired

Way out

• This kind of business should get few strong process oriented folks, whose jobs should be to train various teams
• These folks should provide procedure, tools etc
• They should be encouraged to understand best practice from one process and bring it to entire organization level
• A repository of tools / process / training material / video presentations on these tools and techniques for greater knowledge sharing
• A strong process to recognize innovation + sharing attitude

Structure 3 : Process and Function oriented team –

Here you will find Card analytics team, Home loan analytics team and so on. Within each vertical, you will have people doing acquisition, portfolio management, collections analytics etc. Also you will find verticals for modelling team, MIS team and so on

I call such structure chaotic, where

• Everyone is trying to follow up with others
• Everyone is thinking that this is someone else jobs
• Too many people doing very less actual work
• High friction – due to which they end up having highly polluted political environment. The top leader gets involved in too much of friction related problem solving
• Attrition of good guys
• Reward to bad folks (specially political ones)

Structure 4 : Nuclear team –

This is an extension of functional team. Here say with home loan – acquisition team, you will find same person developing reports, models, policies etc.

Pros

• Exceptional business orientation
• Very high speed of execution – as the person know that data, which was used for metrics that has bothered business. Person will be able to develop models etc quickly
• High employee motivation – as they do meaningful job
• High retention – they love to come to office as they know what they are solving

Challenges

• It is difficult to create such teams as you require all analytical superman capable of doing many things
• A good business environment, where such analytical folks are considered horse power of business innovation and business loves to involve them in their discussion
• A local business, where business and analytics team are sitting side by side
• A very clear understanding of who will do what