With sales forecasting, traditional CRM solutions have not been able to address the industry’s needs, according to Geoff Birnes is the SVP of Customer Engagement and co-founder of Atrium. Birnes explains how many companies struggle with how time-consuming and inaccurate forecasting can be.
The consequence is that many organizations are beginning to combine the impact of data, analytics and more specifically, AI to make decisions faster, make employees more productive and ultimately make customers happier.
However, AI remains a new and uncharted territory. Birnes provides advice on to figure out how to leverage AI in forecasting.
Digital Journal: What are the complexities around sales forecasting?
Geoff Birnes: Forecasting is complex and many companies struggle to maximize the value of this process. In large companies producing for mass markets, sales forecasting is viewed as a formal and highly structured activity involving manual calculations in spreadsheets that aggregate historical direct and channel sales results, with cyclical growth assumptions applied. Unfortunately, these approaches tend to be inaccurate, manually intensive and not actionable in real-time. A blend of traditional and machine learning (ML) approaches are necessary to develop a trusted, actionable sales forecast. These include propensity-based predictions, based on current pipeline data, as well as overall sales forecast predictions using linear regression methods.
DJ: What are the weaknesses with traditional CRM solutions?
Birnes: When it comes to sales forecasting, traditional CRM implementations have not been able to address the industry’s needs. For years, companies have cursed inaccuracies and blamed bad data while identifying the forecast as an all-important output. In these cases, management objectives do not connect to the needs of the field. For a traditional CRM system to work, there needs to be buy-in across the organization and the processes in place to support it. The CRM must provide value-add at the rep level. We can enable this with a data-driven approach to pipeline management and forecasting, where insights and recommendations are provided to sales reps in the context of their workflow. Without this approach, the CRM will become an expensive garbage dump for data
DJ: How can AI make predictions more accurate?
Birnes:In order to make predictions more accurate, a blend of traditional and ML approaches are necessary to develop a trusted, actionable sales forecast. These include propensity-based predictions based on current pipeline data, as well as overall sales forecast predictions using linear regression methods. With the continuous cycle of improvement in AI methods, companies can drill down from their macro-level forecast into opportunity-level data to better understand demand signals without bias. With this level of insight and understanding, companies can change the conversations they are having internally. They can use their forecast as a way to go beyond accuracy and actually improve win rates.
DJ: Which types of companies are best using AI?
Birnes:I would say that companies who know what they are trying to achieve with AI are the ones reaping the most benefits. We are seeing high interest and adoption in banking, insurance, life sciences, and manufacturing. However, one cannot invest in AI just for the sake of it. Many companies ask us how to get started with AI, which of course is the wrong question. You must know what business outcomes you are trying to achieve, then we can move the conversation to the how.
DJ: What are the challenges with using AI?
Birnes:While artificial intelligence has a long way to go, organizations are already solving every day challenges with impressive applications of machine learning. According to Gartner’s 2019 CIO Agenda Survey, the number of organizations that have deployed AI has grown from 4 percent to 14 percent between 2018 and 2019. The major challenge most companies face is building a data pipeline such that AI can be deployed incrementally and managed efficiently. Often, companies try to boil an ocean of data due to a misconception that they need more data than is necessary, and at a higher level of hygiene than is really needed. The trick is to start small, continuously improve, and focus on simplicity of deployment. Cloud technologies have come a long way to simplifying the barriers to success.
DJ: Where can AI develop next?
Birnes:There’s virtually no major industry that AI hasn’t already affected. In most cases, industries such as transportation, manufacturing, healthcare and education are at the start of their AI journey using machine learning. Regardless, the impact AI is having on our lives is hard to ignore. In the foreseeable future, cloud computing and AI will move from a fragmented technology space to a more complete ecosystem, with the major cloud platform providers leading the way. This will open the door to AI capabilities that are both more contextually aware and actionable by blending structured and unstructured data. This, in turn, will lead to more accurate predictions and prescriptive workflow across companies’ business processes. The technologies are already working together to reach that goal. AI will never be as easy as turning on a light switch, but when done right, it will replace the light switch altogether.
