Opinion
Imagine if you had to go to IT to get prebuilt spreadsheets and those were the only spreadsheets that you could have. Sounds pretty useless? Having been through numerous enterprise system implementations for more than ten years, I am of the opinion that most IT systems that we have today are exactly that. This is particularly the case in emerging areas such as analytics, where there is a lot of uncertainty, confusion, and hype; and yet, organizations need to take definite steps and make implement new solutions.
Anatomy of a Typical Analytics Implementation
Say that your company or business team needs analytical forecasting software. The typical process to get such software implemented is as follows:
- Survey software in the market
- Create a proposal, along with the benefits case, and a budget
- Obtain capital approval for the project
- Obtain IT resources for the project — both technologies and people
- Create and evaluate proposals from multiple software vendors, and negotiate price
- Select implementation partners to study your demand planning process and configure the software accordingly
- Implement and test the software
- Maintain the software along with all the system integration links, post implementation
- Call implementation partner and software vendor to make changes to the system
While this may sound like the process only large billion dollar companies would through, the truth is that even the smallest of companies has to do this; it is also the case for many enterprise analytics and planning solutions.
So What Is Wrong With The Above Process?
Slow
Over the last 15+ years, we have repeated seen long capital approval cycles with excessive justification requirements. Software and implementer selection processes are also unnecessarily long. This leaves business teams frustrated and averse to undertaking projects
Rigid, Linear Process
The way applications are implemented today, is rigid and linear, and banks heavily on getting it right the very first time. The software, hardware, implementer teams are brought together for the duration of the project, and you have one chance to get it right. It’s no surprise then that less than 20 % of IT projects deliver the value they originally promised to deliver.
Learning and Buy-In
The entire implementation process is hurried, especially because implementers are charging by the hour. This does not leave enough time for organizational learning. It also fosters a mentality of implement-everything-now, even though the business team has not had enough time to process the full impact of the features being implemented. Many times, these later turn into process and maintenance nightmares.
Too Late To Change
Many companies find themselves in situations where they need to modify or enhance applications due to change in business conditions. The traditional model of application deployment makes this really difficult and expensive. The problem is worse in the case of mid-tier and start-up companies that do not have dedicated resources to maintain such applications, and may want numerous changes to the applications as business and growth rates change.
Is There Another Way?
What if business teams could instead start by simply creating a working prototype in a matter of a few hours? Business teams that are users of these applications can decide how to prioritize which features they wanted rolled out and set their own criteria for how long it would take their team to absorb the new capability before adding more features. If such teams needed changes to the application, they would be in a position to themselves make and deploy the changes to their team in a short amount of time. Here are the key features such solutions would have
Compose-able Applications
What if the business team could create a forecasting application, by simply pulling in the products to be forecasted into a collaborative workspace? Even if the only things they initially deploy are shared views of the same data, visibility to each other’s inputs, and chart views of history and forecast, it would be a great step forward for the team, especially if it could be done in a few hours of time.
Analytics
Instead of following the traditional model of letting the consultant configure advanced algorithms, companies can now start with simple analytical models, with standard, out-of-box forecasting models used as good starting points. Business teams can also roll out the implementations by product groups or by geographies depending on whether data is available, how ready and sophisticated business users are etc.
On-Demand Infrastructure
Unlike traditional solution deployments, many such rapid deployments may be experimental and modified and even torn down by internal teams without any detrimental effects to the business. It is important that business teams be able to create and share solutions without having to worry about technology issues such as provisioning and scalability.
Can Self-Composed Applications Compare With Commercial Software?
Our experience has been that in enterprise settings, collaborative self-service applications are as effective as complex analytical applications deployed by external implementers. The very fact that these solution were developed by the business for the business, gives them legitimacy and a high degree of user buy-in. Further, if the solutions are modular, internal teams can bring in external experts at the right time to enhance the core algorithms powering the solution.
Our 2 Cents
We are in the midst of multiple waves of revolutionary changes in technology, including cloud services and machine learning. The traditional approaches of configuring and deploying applications will soon become a thing of the past. Only when internal teams are able to create and use analytical solutions, i.e. only when they become not only consumers but also producers of solutions, will analytics truly become pervasive throughout the organization and deliver on its promise.
A lot of these solutions will be simple solutions, but will yield tremendous value to the organization. And that’s the whole point — your internal teams know about these problems and are best positioned to solve them!
