From White Lab Coat to Sales Forecasting... Via Chaos and Management!
By Steve Alker, Published: August 14th 2007
From being a graduate in Chemistry and Biochemistry, I went into instrumentation sales (The lure of a company car, a handsome bonus and an expense account defeated the attractions of the lab coat and company Biro) I moved into sales and marketing management and thence to senior positions such as General Sales Manager of a large electronics firm.
Whilst in these roles, I developed an interest in CRM, through the implementation of bespoke databases. I also had responsibility for organising and managing the forecasting process becoming firstly a fan of spreadsheet models and later realising what a time consuming tyranny they imposed on the sales force and management alike.
I came to the conclusion that as all sales people had to keep records of their calls, visits and demonstrations; we could simplify matters and improve reporting, analysis and forecasting. I therefore looked to implement the reporting and forecasting process via the customer record in a CRM system and to use the same system to input the leads generated by my marketing department, so they could be tracked to a sale or a failure to sell.
That was the birth of my involvement with Prospect Relationship Management (PRM) and Sales Performance Management (SPM for forecasting) It is only recently that I have come to discover that the majority of CRM systems are used for anything other than CRM and that has led me to a more analytical approach to clients in the expectation of being able to offer them an exceedingly swift return on their investment.
I was always bewildered by the maths used by my sales managers to analyse my performance. Apart from the fact that they contained a lot of statistical nonsense (A very useful tool when it was my turn to present figures to back my own initiatives!) they also utilised totally invalid mathematics to ascertain both the success and failure of sales and marketing activities. My bewilderment turned to amusement when later in my career I was asked to comment on forecasting systems used by larger companies on their "Big” ERP and CRM systems as well as in bloated spreadsheets containing macros, which in turn contained formulae, the purpose of which had often been forgotten.
The symptoms of these very sophisticated (looking) forecasting tools were that when they were tested with differing inputs and assessments from sales and marketing teams, they started to produce inexplicable results which were so far from any possible reality that, had they been followed, they would have resulted in the company instantly doubling its capacity only to discover that 4 months later, they would need to close down for a year. Having used simple linear regressions in the past to get an idea of future performance, the management naturally wanted to build a model which would take account of the weekly forecasts, the forecast marketing activity and forecast changes to the sales teams. So they built complex algorithms to model the actual forecast sales from real activity inputs and even used the model to determine where they would need to deploy further resources such as PR campaigns, additional sales people to meet the apparent forecast demand and production capacity to manufacture goods.
Then some bright spark pointed out that sales, profit and turnover were not without limits and that high levels of PR and Sales success one month could impact on the ability of the company to achieve a similar figure the next month or quarter or whatever. Well done. They had recognised that you can get saturation in the market, in the marketing coverage and in the field. So they built in negative feedback into the algorithms to stop the sales model from expanding like Fibonacci's rabbits. (Without any constraints to their breeding or any predation, a pair of rabbits would after a remarkably short period of time, cover the surface of the earth in a blanket of rabbits, several kilometres thick, expanding outwards at greater than the speed of light!) It was these models which threw up disturbing nonsense whenever the input variables or imposed constraints were varied, as they are meant to be varied.
They had just discovered the beauty of the strange attractor! When I pointed out that their programme would never work due to chaos dynamics, they wouldn't believe me. The analysts and IT guys all ganged up to say, "How can a deterministic model based on a computer produce random results over time, it's controlled by relatively simple equations and the computer by logic?”
So I showed them the equation for the population of a fruit fly, which was as near as damn it, in functional terms, the same as one of their iterative equations. They couldn't believe their eyes when we demonstrated, on a small spreadsheet and on an excel graph, that for a range of figures, tiny changes to the input data would produce huge difference to a forecast outcome which were to all intents and purposes, random.
That was one of the defining moments when I decided to form my own CRM and forecasting company. I still work with Maximizer; I'm still very keen on activity reports and analytics and passionate about the benefits of forecasting. Hence my decision to team up with SymVolli
I'm not pretending that the algorithms in SymVolli are totally accurate portrayals of the exact maths which will produce a true forecast, but they are continuously changeable by the management so that they will fit the reality you find on the ground and they are designed to mirror the sales process of the client company and, if known the buying processes of their customer.
What they are is consistent and capable of providing rigorous analysis. Even my first sales manager's report-analysis system, filled with mathematical and statistical rubbish as it was, was consistent, and monitoring how well a sales team was performing was at least predictable. OK a few rogues got away with murder and some good guys got the sack based on flaky analysis, but it did work as a management tool and made the company one of the most feared sales machines in the UK amongst its competitors. I know what they would have made of SymVolli because it would have saved them about 100 hours a week and revolutionised the accuracy of their forecast for production purposes.
When I automated the process for their successor company, some 10 years after leaving them, I kept all the best bits of his reporting system for sales management and saved them 150 man-hours a week in reporting time and data collation.
I believe that if we can crack simple reporting, activity analysis and forecasting on a single system that we can make savings and improvements in both efficiency and sales effectiveness that the PC based accountancy packages brought to corporate culture by moving the finance functions from a mainframe and endless spreadsheets onto a network of personal computers.
I remember with pride a quotation from the idiot who was my Finance Director when I developed our first sales based customer database (My first PRM and SPM and CRM project, all done in RBase!) "That's all very well for your salesmen types, but you will never find me entrusting my accounts to a network of personal computers”
15 years later, that company is one of my biggest clients (Has been for 7 years) the accounts department runs on 30 PC's and the entire accounts system inputs sales figures daily into a 35 user Maximizer CRM system. Their forecasting is still shaky, but I've not been allowed to look at that yet. Oh and the aforementioned idiot got the sack.