Outsourcing Analytics? How does it work? Illustrated by sales forecasts

Anna Plemyashova, the Head of AI and analytics department at Beltel Datanomics.

Modern approaches to building software architecture allow you to independently deploy different analytical services, to integrate finished products into existing accounting or platform systems. At the same time, either your own team or a third party can take on the role of the developer of these solutions, and the choice in favor of one or the other party can be made according to the principle “invested efforts – obtained result”, and do not forget to assess the risks in both cases.

Also find out: How the microservice approach to analytical decisions implementation impacts business

Budgets are estimated based on sales forecast. Having elaborated it, you can access any level of budgeting: production with a forecast of the workload of work shifts, consumption of raw materials and purchases, as well as marketing plans and other derivatives. The more operations, factors influencing demand we have, the more difficult it is to elaborate a sales forecast.

As a rule, forecasting is carried out:

• by the capabilities of the ERP system, but the limitations of algorithms and data may affect the quality of the forecast;
• in specific software for forecasting purposes. In this case, it is required to purchase licenses, possibly to increase server capacity and to have trained employees who will work using this software. The results need to be integrated into accounting systems;
• in Excel tables. The quality of forecasting and the labor consuming nature of the process will sooner or later force you to look for new solutions;
• through integration of cloud forecasting services into accounting systems or platforms.

Development of cloud computing enables implementing of low-cost, fast-scalable, high-performance solutions that can be easily integrated into existing systems. IT experts define predictive services designed or customized for each specific business as a global trend for building the planning and budgeting process.

Do I develop my own product, or do I look for an existing solution?

First, it is necessary to estimate the cost of solution deployment for each of the cases.

An existing product has a number of advantages:

• Reduced time for development
  It’s faster and easier to « adjust » an existing product of a third-party designer for your own business specifics rather than to develop your own.
• Access to highly qualified specialists
  No need to spend time and money on training of your own employees, on hiring new highly-paid specialists. SLA could provide you with qualified assistance upon agreed terms up to 7/24.
• Industrial expertise
  Solution developers could enrich current business processes with new ideas thanks to their “trained eye” and experience with other customers.
• Transparent solution budget
  As early as at the stage of quotation you can get the idea of the lead times, specifications, solution and support costs, which is often important for defending of analytical projects in front of the company’s management.

It is necessary to estimate the lead times and costs having given indicators of the forecast quality metrics and a technical task that describes the functional requirements.

Having sorted out quantifiable values (lead times and costs), we can move to the qualitative evaluation of the contractor.

Whether you choose a corporate partner or look for a Head of predictive service development department among your staff, you need to ask approximately the same questions.

Remember to find out:

• What mathematics methods and algorithms are used by the implementer?
• What stack of technologies do they operate with?

The answers to these questions will make it clear whether the potential developer is advanced in terms of analysis and is technologically flexible. At the same time, these answers will help you to understand if choosing one candidate or another will require purchasing additional software or training the team.

• Portfolio of cases over the last 2-3 years

Onrush of IT technologies and algorithms over the past 5 years makes forecasting experience of 10 years ago irrelevant. Certainly, having long records in forecasting is an advantage, however, if this experience is not reinforced with knowledge of creating, let’s say, current cloud architecture, than by choosing such a contractor, the customer may end up with a retro-system that would have problems with scaling and fault tolerance.

• How will security and data protection be organized?

This question is especially relevant when choosing a third-party developer. If this is a cloud solution, which cloud provider is used, should the customer get an individual subscription with the provider, how and where the data is stored. A non-disclosure agreement (NDA) must be necessarily signed.

• How reliable is the developer and what are his plans for further development?

Whether you are choosing a legal entity for outsourcing or looking for a team to join your company, you need to focus on long-term cooperation (at least 2 years), therefore, you should make inquiries about the company through open sources, receive recommendations, understand the plans and priorities of developers for the future to make your choice. Losing a key specialist, as well as receiving a notification from a company that it is closing its business, is equally unpleasant at the height of the project.

And one more important factor. The success and quality of the solution largely depends on efficient interaction both within the development team and cross-functional integration and communication between the contractors and users of the service, so do not forget to evaluate professionals’ ability to articulate their actions and common friendly manners during a conversation.

More

You will find more articles about digital transformation in our blog:

• The ways to reduce risks of digital transformation projects can be found in this article;