There has been a lot of market activity around vendors offering sales-forecasting products (or functionality to address sales forecasting) as part of a wider technology offering for sales and revenue management. As I have discussed in my Analyst Perspective: The Art and Science of Sales from the Inside Out, the pandemic accelerated the prior trends that are now forcing sales leaders and sales teams to reexamine traditional notions of how B2B sales are conducted. In addition, with the rise of the subscription business model and digital e-commerce, a more holistic approach to identify where revenue is coming from and how to manage and optimize a predictable revenue stream is becoming a pressing need. I cover the basic premise of this management of revenue streams in my Analyst Perspective: Revenue Management: The Opportunity for Innovation and Optimization.
These market trends will have a significant impact on what organizations will need to think about when it comes to sales forecasting and should persuade many organizations to reevaluate their current processes and software to ensure they are more scientific than not.
The traditional process by which organizations understand where they may land in terms of expected revenue is the sales forecast. Based on judgment and experience, the sales forecast process focuses on salespersons and management’s view of the likelihood of a deal being closed “won” within a defined period, typically the current quarter. Management takes a salesperson’s committed deal and applies their own judgment not just in whether it will be closed won but also the size of the deal. Whereas there is much commentary on how to conduct a sales forecast and, from vendors, why their functionality is best to get to a more accurate number, very few address whether this approach is fundamentally the right approach. Apart from the need to know the value and content of an individual deal, even the best efforts in using a top-down approach based on the monthly historical performance of open pipeline to closed deals is insufficient. Top-down forecasting cannot capture the short-term variability in deal flow but can use the historical conversion rate against a bottom-up individual sales professional historical pipeline conversion to average between the two to get a projection. But, even then, it is a crude approximation at best.
A sales forecast is often not, in fact, a forecast at all. It is not an attempt at an accurate projection of revenue at the end of the forecast period. In fact, when I see accuracy statistics quoted, I wonder exactly how they are being measured. Given that most deals are not committed until they are well advanced, it leaves out many opportunities that could close win from being counted in the forecast. In addition, what is being forecasted? Final product mix, volume, price or revenue? Often an early-stage deal either does not contain product level info or, if it does, the final outcome is very different from the initial view. This difference was best summed up by a sales operations leader who told me (and I am paraphrasing), “Yes, I know you can probably automate the Sales Forecast process, but if you do that, and the sales manager misses, they will turn around and say that it was the machine’s number, not mine. I want that contract with the sales organization, so we can hold their feet to the fire.”
When it comes to a revenue forecast (new sales, sales to existing customers, renewals), the typical sales forecast excludes as much as it includes. If only a small percentage of deals are committed at the beginning of a forecast period, what about the others? What about deals for which there are not current opportunities, bluebirds or smaller deals? Are upsell and cross deals to existing customers included? What about renewals? What about partner sales? Self-service digital sales?
In addition, there is potentially useful data that is not typically sourced from a SFA system. Such data could be firmographic data or regional economic data. When it comes to including subscription-based business models, the relevant information is more than likely to be in the subscription-management and billing system. Likewise, digital commerce information and partner information is likely to be found elsewhere. All this data is needed to derive the necessary revenue forecast as needed by the revenue operations team and the office of the CFO.
In conversations with finance or supply chain leaders, the general consensus is that sales forecasts are unreliable. Finance and supply chains are looking for a forecast that delivers an expected outcome so they can plan ahead to assess the impact to cash positions and income statement projection. What goods and services need to be positioned when and where, in the case of supply chain? What resources are required for delivery or implementation? Where and what sales engineering or technical resources are needed, and when and where?
So, the difficulty here is that both groups are motivated by differing uses of the outcomes of this sales-forecasting process.
While there is certainly overlap between a sales and a revenue forecast, because of the different motivations behind the use and purpose of these forecasts, companies should employ both approaches – one for the sales organization and one for the rest of the company. Make sure that when you are thinking of reevaluating your sales and revenue forecasting approach, you include all aspects of revenue. Question claims of accuracy: what is being measured to derive accuracy? How important are product and service mixes to your organization? Think about separating sales forecasting for the direct sales team’s purposes from the company’s need for revenue forecasts. Any new approach and tools should support all your revenue-forecasting needs, not just the sales team. Look to vendors who recognize this and do not focus solely or predominately on a traditional bottom’s up sales forecast. Leading forecasting applications will help you blend techniques: bottom up, top down and AI assisted. Better applications will enable revenue sources not directly linked to individual salesperson activity to be used in AI-assisted projections. A good AI-assisted forecast should not be a black box but should instead explain why the results are projected to be as they are with the ability to allow overlay judgment, but one supported with evidence and not just as “plug”. Combining attributes of the seller with opportunities, external data and machine learning (ML) can begin to guide the likelihood of all types of deals closing and thus can provide a more evidence-based means to a unified forecast across all types and channels of revenue.
Regards,
Stephen Hurrell