Sales forecasting methods and tips for B2B marketing teams

There are many sales forecasting methods you can use to predict future revenue.

If your financial year runs from April to March, you’re probably focused on sales forecasts right now. Or maybe you’ve been asked to recast 2021 because the finance director doesn’t think it all adds up?
 
If you’re new to sales forecasting, it can be quite a daunting process, and one that’s not made any easier by the ongoing COVID-19 pandemic. How are you supposed to predict the future when the pandemic keeps shifting the goal posts? 

But, breathe. 

Sales forecasting – an activity that helps you predict future sales figures over a set period – 
can be quite straightforward if you approach it systematically. 

Responsibility for creating a sales forecast may often lie with the sales team, but if your company is structured in such a way that forecasting falls to the marketing team, we’ve got some tips to help you.

Why do you need sales forecasts?

When you get down to basics, a sales forecast reveals what your cash flow should be. This is what drives all businesses, regardless of size. Large corporations may use banks, bonds and shareholders to manage cashflow, but they still need an accurate projection of cash coming into the business. For small and medium-sized enterprises (SMEs), cash is the lifeblood of the business. A company can be both profitable on paper and have a negative cash flow hampering its ability to pay employees and suppliers. Eventually, without free cash flow, that company’s operations will have to stop.  

Sales projections underpin your understanding of cash flow. They tell you when you should expect cash from customers, allowing for things like payment terms, which tell you when and how payment will be made. This means if your business has seasonal fluctuations and a high fixed-cost base, you can make provision for cash when there is a period, according to your forecasts, when cash is not coming in. 
 
While cash flow is crucial, sales forecasts also drive the rest of your business. A forecast tells operations when they need to provide the products, services and solutions that drive your business, procurement professionals when to order raw materials and stocks, HR when to hire new people, and logistics when to have trucks ready for deliveries and distributors. All these activities are ultimately driven by sales forecasts.
 
Forecast metrics are also a measure of your company’s performance, driving bonuses and telling shareholders what their dividend will be. So the accuracy of your sales forecast figures is important.  

Sales forecasting methods: what’s the best approach to take?

When it comes to what sales forecasting method – or approach – to choose, the past tends to be important in estimating the future. If your business has been trading and selling for a while, you will have a sales history and therefore access to some form of business data. However, if you have a newly-launched enterprise with no trading history then you won’t have that sales history and data. 

Broadly speaking, there are two approaches you can take:

1.    Look at what has gone before and use this as the basis for your forecasts. 
2.    Model the sales empirically and look at the future period you are forecasting for as greenfield, whether you have a sales history or not.
 
The best option is to combine the two approaches, but if your business is new, or your marketing team is developing a marketing plan for a new product or solution, this will not be possible. 

Sales forecasting using existing data 

If you have access to time series data, you can use this as a base for future sales. Though not an entirely scientific method, a popular approach is to examine time series data from across the previous 12 months. 
For example, if you sold 1,000 units a month at an average of £1million a month, that amounts to 12,000 a year and £12million in revenue. 

During periods of relative economic stability, business leaders may request sales to increase by X amount over the next 12 months. If you take this approach, carry out market research to assess the conditions and establish what the potential growth rates are.   

If there are market research reports modelling the growth of your wider market at 5% compound annual growth rate and there are no other major factors, like major product launches, then this is a proxy you can use (note, we’ve not used the phrase ‘good proxy’, but it is the only one). So your sales might be 1,050 each month, 12,600 for the year, plus you might increase your prices by 5%, so revenue increases to £12.6million. In statistical terms, this is straight-line forecasting. 

If you can, allow for seasonality, particularly around major holidays. For example, some parts of Europe shut down during the month of August, and decision-makers may not be around to make purchasing decisions. If this is a major market, assume this will have an effect. It’s also important to note that your buying times may be long and complex, so seasonal changes may not take effect until the following January. 

Statistical forecasting using existing time series data

Rather than taking last year’s sales, plucking a number out of the air and scaling up, you can use a variety of statistical sales forecasting techniques. If you have multiple years of time series data of sales, revenues and profits or a business analyst on the team, this is more realistic. 

However, past performance is not always a good indicator of future performance, and you can’t predict the future no matter what the forecast. Airlines, for example, use sophisticated sales forecasting methods to ensure as much accuracy as possible, but how do you think airlines felt about their forecasting when air travel all but came to a halt in March 2020?
 
The common statistical modelling techniques include methods such as moving averages, the Box-Jenkins method/Auto Regressive Integration Moving Average (ARIMA), and simple and multiple linear regression. Moving averages and linear regression can use excel, Box-Jenkins/ARIMA (a favourite of scientists and engineers) uses more specialist software, although you can write your own excel macros.
 
These statistical calculations review the pattern of sales over previous years and predict what might happen next. This approach tends to be more accurate than straight-line forecasting and is the basis that many B2B businesses use for their sales forecasting.
 
There are also some effective tools that model cause and effect. These can be handy for marketers looking to correlate lead generation activities with revenue uplifts. In B2B, there is often a time lag between marketing investment and activity and resulting sales, as the sales process for complex solutions can be lengthy. But it is possible. If you have three to four key trade shows for your sector, and major deals are usually done at these events, you can associate the events with the sales statistically. It is also possible to use the same techniques for above the line advertising and other activities that are not as directly measurable.

Modelling sales for new solutions

Statistical models of future sales using time series data only work if you have existing sales data. So what do you do for new products and solutions? Or how do you predict sales in a new market? 
 
For most ‘brand new’ products and solutions, you can use existing sales data as most new products and B2B solutions are not genuinely new. Though you should factor in changes, like logistics or location, it’s worth noting that, if heavy lifting is your business, the heavy lifting needs of one petrochemical plant operator are much the same as another 1,000 miles away. And if you’re launching a new excavator, you will have data of how your current excavators perform so you can make reasonable assumptions.  

But what do you do when you have something genuinely new or are in a completely new market? Well, you can still build models of your sales and revenues using a combination of actual data and reasonable assumptions. In fact, there are plenty of corporates and consultancies that do just that, and accurately. 

Modelling your market, step by step

Modelling your market means examining multiple types of data and market information to build a picture of your typical customer and their needs. You, or your business/employer, has almost certainly developed the solution in response to a perceived need to solve a problem or provide a benefit, so you already have an ‘ideal’ customer in mind who needs that benefit.
 
You have created a solution that delivers benefits, so use a benefit segmentation approach to model your market size and buying patterns. Start by asking: who benefits? What persona within what type of organisation? How many of them are there? How much of the product will be required per organisation to provide those benefits, or how often might the solution be required, if it is a service? When is the benefit needed, and how often? 

Do some market research to help inform your model. You will need to establish: 

1.    How often the target persona / organisation needs the benefit – a petrochemical plant might need a heavy lifting solution once a month, a small business lender might need a plug-in for their app only once but also require ongoing maintenance and upgrades. Let’s say you have a novel fastening solution that joins steel pipes so they last twice as long in harsh environments. There are around 450 nuclear reactors in the world, controlled by about 150 various types of organisation with a clearly defined supply chain. You can work out roughly how many metres of steel piping are likely to be in these reactors, and how many welds, which gives you a potential total market size. How many competitors are there, and how attractive is your proposition likely to be if your customers do a cost benefit analysis?

2.    How many personas there are and how many organisations – you know that your fastening solution is a compelling proposition for the buyers of steel pipes in nuclear power stations. The buyers will be in the supply chain, which you can model. You can ask some of them how likely they would be to switch from the existing cheaper solution to your more expensive one. How often do new plants get built and how often are pipes replaced? This data is out there or can be estimated using reasonable assumptions grounded in fact.

3.    How much they will pay for it – this will include all the different components and ongoing costs, if any. 

With an understanding of the market size, the possible penetration of your unique new solution, how often it’s needed and a price, after lots of iterations, additional research, reasonable assumptions, adjustments for seasonality and response by market incumbents who have substitutes delivering the same benefits, you have your sales forecast. 
 
There are few needs that require a completely new and unique benefit in the form of products, services and solutions to meet them. Even many innovative digital technologies are satisfying needs we already had, for communication or entertainment, just in a different way or amount.
 
But there are some genuinely new solutions, and you can still forecast the need for these based on your understanding of the benefits that the new solution, product or service provides, which, using the techniques above, will enable some form of sales forecast. This, in turn, provides cash flow, profitability and return on investment data.  

A combined approach

Ideally, a combination of modelling and statistical forecasting, with an element of judgement based on experience, delivers the most accurate sales forecasting solution. 

Use time series data to statistically model future sales based on past sales data, either directly if your solution is the same, or a direct replacement, or indirectly if you have a realistic comparator. Then model your market, starting with benefit and needs analysis. Once you have the statistical forecast and your new market model, use judgement and the judgement of colleagues and consultants, to reality check it. 

Remember that what we label as ‘gut instinct’ is in fact the human machine’s ability to recognise complex patterns. It is OK to factor in gut instinct, alongside all the other data you have assembled. 
 
And finally, recall the caveat at the start of this article – past market performance is not evidence of future performance. Your sales can go down, as might have occurred at the start of the pandemic and during lockdown, or they can increase, as occurred for many organisations during lockdown. You should factor in market shocks, not just for sales forecasting, but also for your wider organisation. That’s just basic risk management.   

Read more in our series of blogs on B2B marketing.

If you need support to write engaging, audience-driven content for your B2B brand or agency, contact Copestone today.