Sales Forecasting; keep all stakeholders on the same page by rewarding accuracy May 3, 2014Posted by Bernard Lunn in Enterprise Sales.
Tags: enterprise sales, sales forecasting, software business
This is # 10 in a serialized book called Enterprise Sales for the Digital Age, delivered here as 12 blog posts. You can get value from each in isolation, but if you really need to understand enterprise sales, reading the whole series is worthwhile. You can buy an improved version, neatly printed and bound, for $6 from Amazon.
Forecasting new business sales revenue is hard. As any sales manager will tell you, that is the ultimate “no, duh” statement. Yes forecasting is very hard.
The reason is obvious – the future is uncertain.
Sales revenue forecasting is also enormously important. Ask any CEO who got hammered by their Board for missing their numbers. Forecasting drives so many critical decisions. Without good forecasts you cannot have a good relationship with investors and you cannot plan your business.
If the company is big and old, you have lots of data to guide your forecasts and errors become rounding errors. However if you run a company that gets revenue from say 5 sales executives, you cannot rely on the usual statistical models. In startups the forecasting is also a lot tougher because there is a step ladder of forecasting difficulty:
– Very Easy: add-on sales to existing accounts. As a start-up you don’t have much of this.
– Fairly Easy: new accounts within a geography and a niche where you have been selling for years. It is unlikely you will have many of these.
– Hard: sales of a well established product into a new geography or a new horizontal or vertical market.
– Really Hard: sales of a new product into a market that is not even well-defined yet. These are the blue ocean markets that allow startups to get traction and scale, but this is a very tough forecasting challenge.
Forecasting recurring revenue contracts such as maintenance can be automated quite easily. You can apply standard assumptions about decay (how many will cancel) and the growth will be based on new contracts.
The problems all come from forecasting new contracts. These are outside your direct control. You are extremely dependent on the judgment of your sales team. SaaS subscription models make new contracts less critical, but investors are still mostly looking for the new contracts (and churn) as the signals of success or failure. Whichever way you cut it, your VP Sales (Sales Director, Chief Revenue Officer, Chief Hustling Officer, whatever you want to call her) has a tough job where everything is on the line every day.
You obviously want more sales. Perhaps even more, you want to know what is likely to happen. You want accuracy.
Attempts to automate new contract revenue forecasting usually do more harm than good. The standard approach is to apply closure rates to the sales funnel. The idea is to make assumptions about how many calls it takes to get meetings and how many meetings it takes to prepare a proposal and how many proposals it takes to get a contract. Then you can say we have 10 deals at 40% probability, 5 deals at 60%, 3 deals at 80% and one deal at 90%, based on where your deals are in the funnel. Put all that in a spreadsheet and hey presto you have a revenue forecast.
This approach appeals to engineers and accountants. It appears to be scientific. The problem is that it generates a false sense of confidence and is very susceptible to gaming as in “lets bump up the number of meetings until we get the desired result”. It is a classic “garbage in, garbage out” problem.
It is better to build a system around what good sales managers do in the real world. What they want to know from a sales guy is “will this deal close this quarter?” In the real world it is always binary – it either closes or does not close. 90% closure does not hit the revenue numbers and 2x 90% is still not worth any money.
Of course this leads to “sandbagging”. The sales guy may have 2 deals that can close in the quarter. He will tell his manager that one will definitely close and keep the other one in reserve. If his “committed close” blows out he hustles to close his back-up deal. If his main deal closes, he can either get his back-up deal in this quarter and be the star of the quarter and pick up some nice accelerator commissions, or push it into the next quarter and get ahead of the game.
Everybody sandbags right up the CEO providing “earnings guidance” to public market investors. Is this a problem? As one Board Director put it, “I love getting sandbagged, it means surprises are much more likely to be positive rather than negative”.
Whatever system you put in place, it will be gamed. The trick is not to try and avoid gaming as that runs against human nature. The trick is to get game theory working on your side by explicitly focussing on accuracy in two ways:
1. Measure input accuracy. The old saw, you cannot manage what you don’t measure, applies here. How accurate was salesman x in the past? Note that this is not the same as “did salesman X make target? The question is “at end Q2, salesman X forecast $1m for Q3. Now at end Q3 what was the actual result?”
2. Reward accuracy. Revenue is always rewarded, but with accuracy being so critical to the company why don’t we explicitly reward accuracy? This can be in “attaboy” gifts; rewarding accuracy with cash when a sales guy is way below target could be counterproductive. Yet they must be good gifts – such as the holiday in the sun all expenses paid for top accuracy.
One reason that we do not measure and reward accuracy is that we are too focused on budgets and targets. These are only plans. What we really want to know is what will happen this quarter? Accountants and spreadsheets can measure the difference between actual, forecast, budget and target and the gaps can be used to kick ass. But don’t confuse that with the main objective of getting accuracy.
Many stakeholders are involved in the sales process and can add value in the forecasting process. During the regular sales review meetings all stakeholders should have a say. For example, the head of Customer Support may chime in with data about a nasty problem that Customer X is reporting that will not be easily fixed. That is likely to delay closing. It may also elevate that problem in the fix priority. Or, a salesman may say “POC for Prospect X starts next week”, but the Head Of Professional Services who provides resources for POCs may so “no, we cannot start next week”. The key output from these meetings is a company view on where each prospect is in the sales Funnel (eg in POC, in contract negotiation, Proposal presented, first meeting).
However that must not replace a simple financial forecast from each sales executive by month. You record accuracy over time. Then you can apply simple metrics. For example:
Sales Exec # 1: 90% accuracy, forecasts $1m in February, you record $900,000
Sales Exec # 2: 50% accuracy, forecasts $1m in February, you record $500,000