5 Steps to More Accurate Demand Forecasts

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.

November 6, 2019
by Jonathan Zalman
Read time: 5 minutes

Step 1: Expand Your Views on Data

Demand forecasts have traditionally been calculated on spreadsheets using historical shipment data. But historical sales data is just one piece of a much larger puzzle; it by no means represents the full scope of factors that impact sales.

The most thorough forecasting techniques includes demand-driven data. In other words, forecasts should take into account consumer purchase behavior in addition to retailer purchase patterns.

Make no mistake: Historical data is vital to calculating demand. But demand planning and forecasting for CPG requires a more robust approach to the kinds of information that’s available. Those massive data sets extend well beyond supply-driven demand signals and require tools that are much more capable than spreadsheets alone (more on this in Step 3).

Factors that impact brand consumption and forecast accuracy include:

  • market trends
  • brand (own) and competitive marketing and trade investments
  • promotions
  • historical shipments and consumption data at both chain and store levels
  • weather patterns
  • seasonality
  • ingredient and health trends
  • online search trends
  • social media mentions
  • other variables that impact brand consumption

Step 2: Know Where to Look

Even though the aforementioned data sets are readily available, most CPG brands don’t know how to correctly utilize this information and incorporate them into their business strategy and forecasting process.

The data sets listed in Step 1 must be sought out, as most of it comes from external sources. So what should CPG brands do once they’ve collected and tabulated internal* data related to historical shipments? Here are two ideas:

  • Enter into connected data partnerships with market intelligence firms that specialize in providing statistics-backed research and insights into consumer behavior.
  • Have a dedicated data science team that can conduct deep research on its own. This is a business strategy unique to each company.

Step 3: Ingest, Clean, and Organize the Data

The number of data points that can affect actual demand is both diverse and massive. Ingesting and analyzing these data pools is better suited to machine-learning models run on enterprise-grade servers that can handle the sheer enormity of data — rather than, say, an Excel model run by an analyst. Unfortunately, most CPG brands do not have the technological infrastructure necessary for this type of demand forecasting process. Rightly so, they opt to use external demand planning platforms that are intuitive and offer a high degree of automation.

(A note on internal data sets: Historical shipment data, for example, is often incomplete and imperfect due to a number of factors, including data siloing, an inconsistent SKU strategy, or the use of insufficient inventory management software. As a result, it’s critical for brands to be able to track shipments across the entire supply chain and have 360-degree visibility in real time, especially when demand planning. A data science team with experience in CPG is critical to organizing the data correctly.)

Step 4: Unleash the Data Scientists & AI

Once all data has been ingested and organized, it’s time to find dynamic relationships between various sets of data. To accomplish this, a team of data scientists will create customized mathematical models that overlay both internal and external data points in search of relationships that would help to accurately forecast sales. Then, a data science team iterates these algorithms, molding and shaping and re-jiggering them until a predictive model is created that meets and/or exceeds predetermined benchmarks (such as forecast accuracy or forecast bias). Ideally, these machine learning models become so accurate that they resemble Artificial Intelligence in their predictive capabilities.

Step 5: Make Sure the Data Is Easily Accessible Across Your Organization

Having machine learning algorithms that yield accurate demand predictions at the SKU-level is only part of the solution. The insights these models produce should be paired with intuitive, easy-to-use software that’s accessible across the entire organization.

Key Takeaways

At this point, you’ve got it all. The right data. The right technology. The right people. The right software. So how do you leverage these demand forecasts? How can your company, and the many areas of your supply chain, benefit from them? Here are few benefits:

  • Operational efficiencies across sales, marketing, finance, inventory, distribution, and manufacturing teams
  • Reduced working capital
  • Better understanding of the ROI of competitive marketing and trade investments
  • Fewer meetings; better S&OP/demand planning process
  • Improved customer service levels


Further reading: The New Rules of Demand Planning for 2020 & Beyond [Complete Guide & eBook]

Stay in the loop

Get the latest CPG insights delivered straight to your inbox.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Unioncrate respects your privacy and will not share your personal data. View our Privacy Policy.

You might be interested in this

No items found.