The Most Effective AI-Generated Forecasts Should Be Augmented by Human Insight

CPG brands need the technological dynamism and agility of AI, but also the kinds of collaborative functionality that enables ground-level human insights to be input then automatically factored into AI forecasts.

by Gaspare Bonventre | April 12, 2021

Introducing… Augmented Forecasting!

Every day, specialized algorithms, more commonly known as Artificial Intelligence (AI), are applied to more and more fields with tremendous effect, and the CPG industry is no different. Our clients witness it every day, as Unioncrate’s AI-generated forecasts provide them with the agility they need to stay ahead of demand in all situations.

Because our sales forecasts deliver impressive automation and accuracy, our clients can proactively plan to get the right amounts of inventory to the right places at the right time. These efficiencies save them incredible amounts of time and working capital costs, to say nothing of increased margins, sales, and productivity.

Powerful and precise as it may be, AI does not know everything. This is why we’ve built out a collaborative functionality that unites human intelligence with artificial intelligence for increased agility, accuracy, speed, and visibility. We call it: Augmented Forecasting.

Image from Unioncrate’s UI showing Augmented Forecasting next to AI forecasts and actuals.

The Complimentary Power of Human Intelligence

Unioncrate’s industry-leading machine learning algorithms help power the best-in-class demand forecasts, but these algorithms do not have the day-to-day interactions with clients and suppliers that humans do.

Despite advances in machine learning applications, human insights are still extremely important to the demand forecasting and supply chain planning process.

Sales and supply teams, for instance, will always possess unique insights that only they (or a select group of people) are privy to, and brands need to have access to technology that enables these inputs to be factored into forecasts.

When these human insights — new distribution gains, supply constraints, promotions, and other buyer negotiations — are able to be combined with AI, brands can potentially raise the accuracy of their forecasts, which in turn can maximize sales, increase margins, and reduce working capital costs.

On the Unioncrate platform, an Augmented Forecast is always presented side-by-side with a pure AI forecast. The reason for this is because the AI takes an objective- and pattern-based view of various datasets to generate a forecast: It always comes to the table without any of the emotion that sometimes skews human based forecasts during periods of unexpected changes. For example, AI is far less likely to overreact to an unexpected spike in demand than a person under stress.

Providing Certainty in Uncertain Times

The brands I work with have the ability to enjoy the powerfully agile combination of AI and HI. And given the uncertainty of last year — a period when historical consumption patterns were thrown for a loop and the global supply chain was regularly disrupted — I would argue that the collaborative visibility of artificial intelligence united with human intelligence is a must-have for brands moving forward.

Of course, agile planning has always been an asset to businesses across a variety of industries, and this is especially true for CPG. Consider Q2 2020, for instance. In April, after a month of unprecedented demand spikes for many goods and services, numerous high-growth companies, expecting similar demand levels going forward, overpredicted their sales numbers. For these companies, this resulted in excess inventory, which tied up working capital and decreased margins because they then had to offer deep discounts just to move goods off the shelves. Many companies also had to go back to their shareholders and revise downward some of their previous earnings and financial guidance. Internally, team members were having to deal with issues such as unanticipated higher working capital usage, decreased margins, and price cuts to move inventory faster. Very inaccurate forecasting harmed more than just their bottom line; it added stress throughout many organizations.

The faster a company could align operations with the pace of demand — and rely on sales, inventory, and distribution forecasts to get them there — the faster they could get back to normal.

Building Key Components of a Leading-Edge Technology: Augmented Forecasting

Most people recognize the benefit of both machine learning algorithms and human insight. Unioncrate set forth to combine the two forms of insight in a manner that works well within organizations of different sizes.

Throughout the R&D process, we brainstormed internally, we met with different clients to pick their brains, and we iterated continuously over many months. New technologies were developed at Unioncrate and some of our existing processes had to be changed. In the end, we arrived at a vision and implementation of a forecasting process which we term Augmented Forecasting.

From a high level, here’s how Augmented Forecasts works.

  • Start with very granular AI-powered forecasts, and enable a firm to grant access to anyone within the organization that can add valuable insight. This includes marketing, sales, executive management, logistics coordinators, warehouse managers, and internal forecasting teams. These team members can input changes that may impact future sales — including, but not limited to, distribution changes, new marketing investments, supply constraints, and other buyer communication.
  • Then add human gatekeepers to oversee and approve which human insights are added to the Augmented Forecast while providing a record of all approved forecast insights to different team members.
  • Surface those sales forecasts to key stakeholders in a clean and easy-to-use interface. Use those sales forecasts as a basis for inventory forecasts.
  • Finally, benchmark the Augmented Forecast alongside the pure AI forecast to quantify the continuous benefit of a firm’s human insights.

Once these proposed forecast modifications are approved, they will be either automatically factored into future forecasts or treated as one-time occurrences. This Augmented Forecast is visible to all allowed members of an organization, and it can be viewed with tremendous granularity down to the weekly, product, and ship-to levels.

Concluding Thoughts

Even with AI at the wheel, human insight remains an integral part of building and maintaining a holistic demand forecasting process that’s both agile and collaborative.

Augmented Forecasting plays a key role in unifying supply chain, financial, and operational data; eliminating blind spots from demand forecasts; and slashing hours from the planning process. During uncertain times, this technology can be a stabilizing — and powerful — force. When sales increase or decrease, a domino effect occurs. And a company’s ability to provide functionality that enables cross-functional teams to enter and view changes can be incredibly useful in the Integrated Business Planning (IBP) process.‍

Unioncrate is an AI-powered Integrated Business Planning (IBP) platform that delivers demand forecasts with unmatched accuracy, collaborative visibility, and actionable intelligence — enabling CPG brands to plan and execute agile supply chain strategies at the click of a button.

AI-Driven Supply Chain Planning Platform for CPG Brands