In 1971, Richard Nixon was president, All in the Family debuted on CBS, and the first Starbucks opened in Seattle. It’s also the year the Harvard Business Review featured a story on how to pick the right forecasting model for your business.
Much has changed in 49 years, but the relevance of forecasting models remains strong — particularly at a moment overcome with uncertainty. Though revenue forecasts, which predict business income over a fixed period, aren’t always fool-proof, they can help businesses anticipate challenges before it’s too late.
In fact, one of the most important aspects to remember about forecasts, as marketing software provider HubSpot reminds us, is perfection is not the goal. Instead, use forecasts to plan for your business and adjust as needed. Before you can tap into the power of forecasting models, you have to choose one. And with so many options, we’re diving into everything you should know.
What Forecasting Models Look Like
When it comes to forecasting models, they boil down to two different methods: traditional statistical methods, such as regression analysis, and modern machine learning methods, like neural networks. The one you choose depends on who is responsible for forecasting within your organization and what that person’s background is. Someone with experience in data science may opt for machine learning, while business professionals are more likely to rely on statistical methods.
To identify the best fit for your company, engage employees in a conversation before you make your final decision, according to the global consulting firm Bain.
Here’s an inside look at the most common models:
- Statistical models: These projections show the impact of a given variable in the future — and how certain elements impact that variable. They require a modest data set and are particularly suited for forecasting retail sales, according to Bain.
- Machine learning models: These show nuanced relationships between variables and work best with large data sets, so you’ll need help from a CRM vendor. Retailers and consumer packaged goods (CPG) companies use machine learning and predictive analytics to help create more accurate demand forecasts as they expand into new sales channels, according to SAS, a statistical software suite.
- Expert forecasts: As the name implies, these predictions rely on the knowledge and experience of industry experts to make sense of a business within the context of the market. They require minimal data, as the projections hail from people. Remember: this model is subjective and prone to bias.
- Combination models: Combine different models for enhanced accuracy. However, this will require working with a knowledgeable partner who can guide you through the process, especially if you don’t have data scientists and AI capabilities.
These basic forecasting models are the tip of the iceberg. In terms of more sales-specific methods, you can also consider:
- Opportunity stage forecasting: This method looks at all potential deals and the likelihood they may close to determine an overall projection. However, HubSpot notes the results are often inaccurate as the model fails to consider the shelf life of a given opportunity.
- Length of sales cycle forecasting: This model uses the relative age of an opportunity to determine when it is most likely to close. You will have to carefully track data, which likely means working with a CRM vendor.
- Historical forecasting: If data tracking or CRM isn’t your style, you can use historical data to figure out what may happen. Once again, HubSpot warns this is not a perfect method as it doesn’t consider seasonality and assumes buyer demand is constant.
- Multivariable analysis forecasting: Perhaps the most accurate model, this one combines predictive analytics with business data to make its forecasts. However, this requires substantial data and advanced analytics, so it can be cost-prohibitive for some users.
The one aspect these models have in common is data. No matter which method you choose, remember it’s only as good as your data. Therefore, it’s essential to get buy-in from your sales team, be in regular communication, and hold staffers accountable, so the data they input into your CRM platform is accurate and actionable.
For more on why data-driven marketing is crucial:
Forecasting Models in a Pandemic
This is once again where 2020 is an exception. Because we’re in the midst of an incredibly challenging time to plan for anything, it may be tempting to think forecasting models are useless. However, the pandemic has made forecasting more important than ever as it gives business leaders a roadmap to chart the course during this unprecedented time.
If you’ve looked around and realized your existing forecasting tools are inadequate, this may be a smart time to reevaluate what you’re using to find the best fit for your brand. Though no one likely accounted for the COVID-19 pandemic in their 2020 forecasts, it’s crucial businesses re-assess their projections for the year as soon as possible. This will give key stakeholders the best insights to figure out the way forward.
It’s best to use machine learning models in this endeavor, according to KPMG, a professional services network. It also suggests ultimately automating the forecasting process, so you’re ready with the most advanced tools available before the next crisis. Also, remember that to update your 2020 forecasts, you’ll need to pull external data related to the pandemic, such as how governments respond and the potential impact on consumer demand.
The Closest We Have to a Crystal Ball
Unfortunately, forecasting models are not crystal balls. They don’t show us the future as it will be — and they definitely didn’t foresee a global pandemic. But they’re the closest tool we have to a clearer picture of what the future may hold.
While the options may seem endless — and they are to an extent — this only underscores the importance of searching for the right partners, which will help you find the right forecasting model, tap into its power, and drive your business forward.
For predictions on post-pandemic trends to watch out for:
Angie is the Content Marketing Manager at AdRoll. Prior to AdRoll, she was a Content Writer at various digital marketing agencies. A writer by day and a reader by night, Angie’s other hobbies include cooking and learning useless movie trivia.