Producing accurate sales forecasts

Forecasting sales for the next financial year can be a bit of an art but it doesn’t need to be overcomplicated. The simple principles of quantifying uncertain market forces are related to the management of risk and being sensible about the treatment of that risk in your sales forecast by applying prudence to future forecasts of revenue. The Ansoff matrix below shows how risk changes with different types of business:

The following paragraphs suggest how different types of sales prospects can be handled in the sales forecast:

1. Ongoing contracts, long term projects and support and service contracts are likely to continue as long as you continue to perform and deliver to the customer’s needs. However, the business should not become complacent and continue to offer existing customers new improved services and products. If you don’t your competitors will. These are at the top of the list when preparing your sales forecast with confidence levels between 90 and 99%.
2. Repeat business has a high probability of coming in at the quarters you predict assuming you have a good, close relationship with your customer and are aligning your delivery schedules with their business models, including their business uncertainties. Offering new services and products with compelling propositions that your customer is interested in will consolidate the customer relationship and increase the probability that they will not look elsewhere for a better deal particularly in times of austerity. These should be second on the list when preparing the sales forecast with a confidence level of 80-90% of retaining the follow on business.
3. New business with an existing customer/ market comes next. The risk of not achieving in-year sales with this type of business is high. Even if you do everything right there is a risk of the business going to a competitor (see Porters competitive forces), customer deferring the purchase, customer CAPEX or OPEX budgets being cut or even customers strategy, plans or business lines collapse. The market may also be subject to change as it is dynamic and open to the pressures of external environments (PESTC). Any of these can impact on the probability of a new business sale and until you have a signed contract the confidence level should be between 50 and 80%. The confidence level can be adjusted as you get closer to contract closure and as your customer intelligence predicts. This applies to new business where the contract value is near to average value for the market you are in.
4. High Value new business with an existing customer / market. In this case the high value of the contract can be distorting to your business forecasting figures. The forecast should therefore be treated as a digital (win / lose) opportunity and taken out of the forecast completely. It can be treated as a windfall should you be lucky enough to land it. Luck in these types of contracts is a big factor even if you make your own luck through hard work and diligence. When the value increases customers get nervous about placing the contract and making a mistake, there is an increased level of scrutiny in your customer’s organisation and if it is a government contract there is likely to be political involvement in the vendor selection decision. This all leads to this type of contract being a lottery in both time and content. A success in this type of contract can be transformational to your business and it is necessary to have a detailed plan of how to cope with this outside of the traditional sales forecast. Needless to say the confidence level should initially be treated as low (30%) and regularly reviewed and adjusted as market intelligence is refined and you approach the contract award date.
5. Existing business products with a new market or new customer. When forecasting this type of business you will have high confidence of pricing and commercial terms for your current market but may be unfamiliar with the procurement rules and commercial norms in the new market. As a new entrant to the market there will be a lower probability of winning the contract than a current incumbent supplier so your product or service will need to be well differentiated and something that the customer has expressed a real interest in. The confidence level in the forecast will be low initially (50%) but can be improved as your presence and relationships in the market/ customer develop.
6. New business with a new market or new customer. This is often referred to as diversification and is the hardest type of business to forecast accurately. That is because you are simultaneously introducing a new product that by its nature, you have sparse data for and are entering a new market where you do not understand the procurement, pricing and commercial rules.

The confidence level can be treated as a weighting in the forecasting spreadsheet with values ranging between 0 and 1. Once you have a qualified opportunity to put into the sales forecast the decision about the likely profit, resource demands and cash-flow can be forecast. This requires information about the ability of the organisation to deliver the business to time and cost in addition to the sales predictions of revenue and time (contract commencement dates). Alternatively, the confidence level can be applied through a probability distribution with higher confidence levels having a narrower spread of outcomes. This method is often simplified to three point estimates that result in a most likely sales forecast and two extreme forecasts representing favourable and detrimental outcomes.

For items 3-6 above there may be more benefit in using rolling forecasts rather that the traditional annual forecast. The traditional annual budget and the ensuing updates, estimates or forecasts to the budget, are the primary tools in use today to manage the financial and operational performance of an organisation. However, due to a low level of detail and extensive politicising of the process, the traditional annual budget often takes too long to create and is out-of-date by the time it is completed, leaving an organisation without a functioning plan for much of the fiscal year. In an attempt to address this problem, the concept of a rolling forecast is gaining traction and is being adopted successfully by various organisations spanning a range of industries around the world. 

Traditional forecasts combine recent actual data and updates to the budget to recompile full year projections. As another month of actuals are available, one less month of forecasts are updated. These x/y forecasts often are referred to as 1/11, 3/9, 4/8 or 8/4 forecasts indicating that there are x months of actual and y months of forecasts, but x + y always equals 12, meaning that the focus is always on one full year. In other words, the May forecast will be four months of actuals and eight months of forecast, while the September forecast will have eight months of actuals, but only four months of forecast.

In contrast, a rolling forecast keeps y constant so that it is always looking forward by the same number of periods. As rolling forecasts become adopted and ingrained, this is adapted slightly, but the core concept stays the same. The benefit of this approach is that the number of forecast periods is dictated by real business drivers such as business cycle, competitive forces, price sensitivity, vendor reliance, business cycle and technology adaptation. Many of these real issues cannot be treated as constant over a 12-month period as they very often are in the traditional budgeting process but they are critical for managing future operations. The benefits of rolling forecasts are shown in the table below and can be summarised as:

• Enables a consistent forward looking business perspective that aligns to the cadence of the business rather than the financial calendar
• Enables management to regularly monitor and course correct
• Presents a stronger basis in reality rather than the aspirational nature of annual budgets

Historical data can aid with the forecast in reasonably stable markets. This can set the base line for forecasting next year’s annual sales. If there is a regular pattern to annual sales and the market dynamics are not changing, then there is low risk in forecasting based on the performance of previous years so long as you can accurately predict the year on year or season on season trends.

Using big data analytics in sales forecasting 

It often seems to the rest of the organisation, outside of sales, that the sales team is involved in an elaborate process of guesswork. This suspicion by the board and the CEO turns to paranoia when markets become more dynamic due to austerity measures and customers delay, defer or cancel proposed contracts midway through the financial year without warning. 

From their perspective, one day the pipeline is strong, the next day panic has set in as it looks like sales will miss its targets, followed the next day by claims “it’s OK, we’re good.”   It is difficult to argue with their criticisms of unscientific behaviour and what appears to be complete guesswork.   Despite automation like CRM, SFA, and ERP systems, analytics, and Sales Operations teams engaged in sophisticated analysis, it still remains guesswork.

This concern for the reality of the numbers going into the sales forecast is further compounded by inconsistent sales team behaviour when managing their pipeline. People outside of the sales team ask perfectly sensible questions about why Sales don’t report on some potential sales opportunities, underreport others, and on occasion are overly optimistic as they manage how their leaders perceive their pipelines.  This makes the head of sales’ job hard as they try to optimize their team’s productivity, and it has a ripple effect from finance to marketing.  The way most sales leaders manage this lack of visibility is to add layers of managerial oversight, daily reporting, frequent detailed account and pipeline reviews, and to generally micromanage the whole process which also compounds the uncertainty as confidence builds up over periods of months and is often subject to undulations within that period. This is analogous to the daily oscillation of share prices compared to their relatively predictable trend performance over longer time periods.   It has been said that CRM systems do next to nothing to help the manager manage the process and that spreadsheets are the state of the art tool to help sales managers support the forecasting process.

A large part of this inconsistency in the sales pipeline results from a lack of accurate data from market intelligence. However, in a digitally connected world there is now a wealth of real data available that can be collected in ERP and CRM systems. A number of big data analytics tools are able to draw information and intelligence out of the raw data. Analysis of big data may uncover patterns that are not immediately obvious when looking at the raw data. Companies that manage by focusing on changes in business patterns have a valuable early warning system; a core success factor for businesses operating in dynamic markets. These companies are able to respond faster and smarter as sales predictions unfold.

Big data tools and applications enable Marketing and Sales to spot changes in buyer behavior and sales cycle patterns at macro and micro levels.  Knowing that a buyer is about to disengage from a sales cycle, based on past patterns of behavior, enables a sales manager to either quickly disqualify the opportunity or get it back on track to win.  In addition, the ability to spot a change in a sales rep’s performance enables contextual coaching to be delivered when it can have the most impact, in near real-time.  That’s a more effective strategy than ‘one size fits all’ sales training or the quarterly termination of the bottom ten percent of the sales force.

Applying big data analytics techniques and getting closer to your customers is the most effective way of ensuring that data going into the sales forecast is accurate and useable for managing the business. This can also empower the sales team by:

• Matching your selling processes to how your buyers want to buy.
• Cleaning up your data to prevent insights being wrong, and costlier.
• Optimising your campaign-to-cash processes.
• Automating everything to free up Sales time to spend with buyers.
• Building Sales operations to act on big data insights versus building complex spreadsheets
• Training sales on how to make that first call more successful by leveraging social media.
• Inviting Marketing to use the big data application to improve the quality of your leads.

Graham Jones