December 30, 2016 | Written by: Jaime Marijuán Castro
Written by: Jaime Marijuán Castro
In the electronics industry nowadays, the average gross value of inventory (all types) as a percentage of revenue is 9,58 % (*), a pretty high ratio and proves that the cost of inventory is a matter of concern within the industry. High levels of inventory can be tackled in multiple ways, and it is widely agreed within the industry that a strong demand forecasting capability is cornerstone to healthy inventory levels.
On this post I want to share some ideas around demand forecasting framed in the context of the challenges the electronics industry is facing and some technologies that could have a positive impact in the demand forecasting.
Two Industry Shifts
- The shift in the buyer’s behavior. The users and consumers of electronics are everyday more informed, and they are more empowered too. With the spread use of social tools and increasing connectivity, they can now take better purchasing decisions, they can influence product development and they can radically affect product demand and brand image based on their perception posted on social networks.
- The shift in the industry marketplace. Convergence is blurring boundaries between industries, new disruptive competition and technology advancement are re-shaping the industry and profit margins are shrinking. It is therefore important that companies look after cost and prevent inventory imbalances.
A closer look to demand forecasting
Electronic companies have been constantly looking for ways to improve demand forecasting, better manage supply in order to trim inventory levels. As opposed to what many people think, technology should have helped with better predictions, however many organizations are hitting the lowest demand forecast accuracy in years, and they are taking higher inventory carrying costs.
Throughout this year I’ve been thinking on how cognitive analytics and blockchain could help demand forecasting and the entire supply & demand loop. I certainly see a lot of potential on these 2 technologies as game-changers within the discipline and I would like to share some thoughts and ideas:
Cognitive Demand Forecasting
I have no doubt that a good cognitive analytic engine can help improve demand forecasting, by finding out new patterns and data insights in almost any time-series or causal-series methodology applied. But the real value of cognitive analytics will come from its application into any of the demand forecasting qualitative methods.
Cognitive analytics has the ability to distill the social sentiment from social & collaborative networks. It also provides relevant insights from news feeds or analyst reports and get key information from unstructured data sources, like for instance, previous planning cycles files, sales ops. inputs or any other corpus of data coming from mails, text messages, chat messages, audio, video and other. Any source of data is valid input for cognitive demand forecasting and the best out of it is that it can be used repeatedly, in near real time and with unprecedented speed.
Some interesting use cases could be:
- Generation of demand signals. Cognitive analytics could enrich the quality of near real-time inputs to demand sensing techniques, thus improving demand forecasting accuracy.
- Enhanced Delphi Method, Why not helping experts be more experts? Cognitive analytics could enhance this method providing a well structured, near real-time insights on what the buyer is commenting about the product and brand.
- Cognitive monitoring of macro-economics. The ability to analyze chunks of text, unstructured data of any sorts (image, video, audio, etc.) delivers a rounded flavor to macro economic analysis.
- Risk-weighed demand models. Adding new cognitive analytics around risk insights to perform trade-off analytics, next best action in demand models.
These are realizable ideas that will improve demand forecasting accuracy. Moreover, the cognitive analytical layer can be applied in other areas like demand planning, supply planning, demand management, S&OP’s, Sales quota planning, etc.
Blockchain in Supply & Demand
As long as the supply chain becomes more intricate with globalization, the number of input sources and players participating on demand forecasting grows. Multiple tiered suppliers, several manufacturing sites, warehouses and other distribution centers scattered across the globe and managed by 3PL’s and EMS. The marketing, sales and distribution organizations also are becoming more complex and with more business partners and distributors in the game. Last but definitely not the least, the consumers / users are empowered as I mentioned previously. This leads to multiple record keeping efforts and lots of inefficiency in reconciling files and information.
A permissioned blockchain – as one type of distributed ledger system – is the other technology I believe can bring lots of value to demand forecasting as a mean to keep one shared record of information around the supply and demand triad:
- Demand forecasting.
- Supply planning & commit.
- (actual) demand management.
Having the outputs of these three elements shared in one distributed ledger ledger would eliminate a lot of unproductive reconciliation work and it would allow organizations to focus on what matters the most: to react to unexpected demand swings, and to run forecasting cycles with higher accuracy.
Supply and demand gains accuracy with cognitive analytics and permissioned blockchain
Meet some of the most prominent experts in Blockchain and Cognitive Analytics at CES 2017 Vegas.
(1) IBM’s Institute of Business Value Benchmark on a sample size of 875 electronics companies.