Modeling Sales When Planning a Start Up

This is the second of three articles on how to create a sales forecast when writing a business plan for a start up company.

In the previous article we discussed the activities the create revenue. In this article, we will survey some of the techniques you can use to create estimates of units you expect to sell in your start up company.

Many dozens of techniques exist to model your anticipated sales. Each is based on unique assumptions and each produces quite different projections. We can categorize sales modeling techniques into three rough groups: top-down, bottom-up, and wild-ass guesses.

Top-down techniques:

  • Look at the overall target market and allow you to predict sales by estimating market penetration.
  • Usually result in highly optimistic estimates.
  • No educated third party will believe your top-down estimates, nor should they. However, they are helpful in that they provide you with an absolute ceiling of reasonableness when you perform your bottom-up estimations.

Bottom-up techniques:

  • Ask you to determine how you are going to actually find leads (or how they are going to find you),
  • Ask you to define the efficiency of the “engine” that converts those leads into customers.
  • Ask you to model the viral and organic growth engines of the company.
  • Once your company launches, you will be modeling the effectiveness of your company’s revenue this way, so it is a good idea to plan the company from the start using such a model.

Doing both top-down and bottom-up is a good eye-opening experience. I recommend that you use at least one top-down and at least one bottom-up technique.

Wild ass guesses (WAG) are really not techniques per se. They are, well, just wild ass guesses. They:

  • Ask you to estimate how many units you will sell (or customers you will have) each month (or year).
  • Educated third parties are even less likely to believe WAGs than top-down technique based estimates. However, a WAG may be all you have when you create your first financial plan. I highly recommend that you replace the WAG with something more reasonable before you show your financial plan to anybody else!

The following sections describe various ways to model sales. Each section describes a technique, when you should consider using it, its advantages and disadvantages, and provides a list of relevant business assumptions that you must record in each case. The business assumptions can be used to drive your predicted revenue during the planning stage; they also become the hypotheses you will be validating while conducting experiments [RIE11].

  1. “Sales by Market Penetration” Sales Model (Top-Down)

For each product and each market, estimate the percentage of that market you think you can reasonably capture each month (or year). Note that the term market is not particularly specific. If you choose to define it using its broadest possible interpretation (e.g., the population of Gen-X-ers in the United States, or the number of small software companies in New York), then you should expect your predicted target penetrations to be quite small. On the other hand, if you define it to be very specific and limit it to a very narrow vertical (e.g., Gen-X-ers who are members of public golf courses in Colorado, or software project managers in Fortune 500 companies), then you should expect your predic­ted target penetrations to be much larger. Assumptions you want to record when using this sales modeling approach (for each product and each market) are:

  • Market size. Current number of customers in the market, or alternatively, the number of products that could be absorbed by custo­mers in the market. Whichever one you select as your definition of market size, you must remain consistent throughout the rest of the model.
  • Average order size: If market size was specified as numbers of cus­tomers, then “average order size,” is the average number of units you expect each customer to purchase per month (or year); if market size was specified as numbers of products that could be absorbed, you may ignore this assumption.
  • Annual rate of growth of market. At what rate do you expect the market to grow each year? If you expect the market to shrink by 10% per year, record this assumption as minus 10%.
  • Market penetration. What percent of the market do you expect to have as your customers by month (or year)? This will be an array of assump­tions, one for every product in every market during every month (or year)

2. “Sales by Marketing and Sales People Effort” Sales Model (Bottom-Up)

One way to think about how your business will attract customers is for the business to have employees who proactively seek leads, and convert them into customers. This approach is most applicable to companies using direct sales. You will model sales by stating which employee types are ones that drive sales, and how many sales of which pro­ducts can be generated by each employee. Such a company often becomes successful when it creates a repeat­able sales model, i.e., when it can predict it would increase its revenue by 25% by increasing its marketing and sales staff by 25%. However, this ignores the long-term financial implications, i.e., whether the profit derived from products sold is greater than the cost of having marketing and sales staff on the payroll. This will be corrected in later techniques. The assumptions you need to define (for each product and each market) for this sales model are:

  • Which marketing/sales people? Of the various personnel the company is employing, which specific one(s) drive sales?
  • Ratio of employees to units sold. If one salesperson can close, say, 5 customer deals per month and the average customer purchases 4 units, then this ratio should be recorded as 20 (i.e., 5 × 4). The interpre­tation is that the company must hire one salesperson for every 20 products it wishes to sell. Notice this ratio aggregates all levels of the aforementioned sales funnel.
  • Ramp up. On average, how months elapse between when an employee is hired and s/he is fully up to speed?

3.  “Sales by Marketing and Sales Dollars Spent” Sales Model (Bottom-Up)

Perhaps your sales are not tied to salespeople but to some other proactive marketing and sales effort like rent (which gives you visibility to passersby), direct mails, phone calls, sales calls, product demonstrations, trade shows, maga­zine advertisements, Google AdWords™, or whatever. Performing these activities creates leads. Following up on these leads results in a certain percentage of them being converted through the sales funnel into sales. You could model your sales by stating which expenses are the ones that drive sales, and how many sales of which products can be generated by each dollar spent. A company has a repeatable sales model when it can predict it will sell X units when it spends $Y in a specific type of marketing or sales effort. The assumptions you need to define (for each product and each market) are:

  • Which marketing and sales activity? Of the many marketing and sales expenses that the company is incurring, which specific one(s) drive sales?
  • Ratio of dollars spent to units sold. If spending $10,000 on the specific activity creates, say, 200 leads, and you expect to convert 20% of them into prospects, and 50% of those into paying customers, and the average customer purchases 4 units, then this ratio should be recorded as 125, i.e.,

$10,000 / (200 × 20% × 50% × 4).

            The interpretation is that the company must spend $125 in order to sell one product.

  • Sales cycle. On average, how months elapse between when marketing dollars are spent and resultant revenues are realized?

4.  “New Customers by Marketing and Sales People Effort” Sales Model (Bottom-Up)

The previous two techniques allow you to predict unit sales of products from marketing and sales efforts. This (and the following) technique allow you to predict the acquisition and retention of customers from marketing and sales efforts. The assumptions you need to define (for each market) are:

  • Which marketing/sales people? Of the various personnel being employed, which specific one(s) drive sales?
  • Ratio of employees to new customers acquired. If one salesperson can acquire 20 leads and convert them into 4 prospects each month, and 50% of them into paying customers, then this ratio should be recorded as 2. Note: an alternative to recording this as a ratio is to record it as a customer acquisition cost (CAC). The CAC would equal the employee’s monthly base salary divided by the number of customers s/he could create each month [BES10].
  • Ramp up. On average, how months elapse between when an employee is hired and s/he is fully up to speed?
  • Average order size. What is the average monthly recurring revenue attributed to each customer? This will be an array of average order sizes, one for each year of the model.
  • Annual retention rate. What percent of existing customers remain customers after 12 months?
  • Viral coefficient. How many new customers will each existing cus­to­mer attract?
  • Viral cycle. How many days elapse between each new generation of customer referrals, i.e., how long does it take (on average) for customers to “spread the word” and get their friends to become customers?

5. “New Customers by Marketing and Sales Dollars Spent” Sales Model (Bottom-Up)

Similar to the previous model, but this one allows you to predict how many new customers you acquire as a function of how many marketing and sales dollars you spend rather than employees you hire. A company has a repeatable sales model when it can predict it will acquire X new customers with an expected average order size of $Z when it spends $Y in a specific type of marketing or sales effort. $Y/X is called Customer Acquisition Cost (CAC) [BES10]. A company has a sustainable engine of growth when the profit derived from a customer’s lifetime (called CLV, customer lifetime value) exceeds CAC. The assumptions you need to define (for each market) are:

  • Which marketing and sales activity? Of the many marketing and sales expenses being incurred, which specific one(s) drive the creation of new customers?
  • Customer acquisition cost (CAC). If spending $10,000 on the specific activity creates, say, 200 leads, and you expect to convert 20% of them into prospects, and 50% of those into paying customers, then CAC should be recorded as $500, i.e.,

$10,000 / (200 × 20% × 50%).

  • Average order size. What is the average monthly recurring revenue attributed to each customer? This will be an array of average order sizes, one for each year.
  • Sales cycle. On average, how months elapse between when marketing dollars are spent and the resultant customers are acquired?
  • Annual retention rate. What percent of existing customers remain customers after 12 months?
  • Viral coefficient. How many new customers will each existing cus­to­mer attract?
  • Viral cycle. How many days elapse between each new generation of customer referrals, i.e., how long does it take (on average) for customers to “spread the word” and get their friends to become customers?

6. “Raw Materials Available” Sales Model (Bottom-Up)

If demand for your product is unquenchable and the only limitation on your ability to sell is availability of raw materials to produce product, then the above sales models will not work. In such cases, use this model. Assumptions you need to define (for each product and each market) are:

  • Which raw material is the key driver of sales? Of the various raw materials, which specific one is scarce?
  • Availability of this raw material by month.

7. “Manufactured Product Available” Sales Model (Bottom-Up)

If demand for your product is unquenchable, raw materials are abundant, and the only dam­per on sales is how fast your manufacturing process can produce product, then this is the sales model for you. Assumptions you need to define (for each product and each market) are:

  • Units available by month. How many units can you manufac­ture during this parti­cular month? This will be an array of unit counts, one for every product during every month.

8. “Sales by Year” Sales Model (Wild Ass Guess)

When you first start planning a company, you may have no idea how large your market is (and thus cannot perform a top down analysis) and have no idea how you are going to sell your product (and thus cannot perform a bottom up analysis). When this is the case, you have no choice but to simply make guesses of how many products you will sell or customers you will attract. However, in no case should you ever show revenue projections based on such wild ass guesses to anybody else; they are solely for your own consumption.

In this “Sales by Year” sales model, you predict how many units of each product you will sell in each market during each year. It is ideal when trying to obtain a very rough idea of whether or not your company makes business sense. Because it is so rough, it is not effective at any later stage, especially when trying to convince others to join your team or invest in your company.

  • Units sold. How many units do you expect to sell in this market during each year? Because this sales model is used only for gross calculations, the assumption will be made that sales are spread evenly across all months of the year. If a finer spread is necessary, use the earlier sales modeling approaches. This will be an array of unit counts, one for every product in every market during every year.

9. “Sales by Month” Sales Model (Wild Ass Guess)

In the “Sales by Month” sales model, you simply predict how many units of each product you will sell during each month. It might apply to two kinds of businesses: (1) businesses that are less proactive with respective to sales, so that customers are expected to come to the company as the result of walking by a store front and you have no plan to proactively seek customers, or (2) businesses that expect to be proactive but have not yet developed a repeatable sales model and thus cannot predict with any level of accuracy the ratio of employees or expenses to sales. The biggest disadvantage is that it requires entry of a great many numbers. Because you might use different sales techniques in different markets and you might introduce different products at different times to different markets, you should define the following assumptions separately for every product × market combination:

  • Units sold. How many units do you expect to sell during each month? This will be an array of numbers, one for every product in every market during every month.

10. “Sales by Annual Growth” Sales Model (Wild Ass Guess)

This approach allows you to fine tune monthly predictions for sales of a product during its first year of launch, and then give rougher estimates for subsequent years by just stating an annual percentage increase. It is only slightly more refined than the previous method; not particularly good when trying to convince others to join your team or invest in your company. To use this model, specify these assumptions for each product in each market:

  • Introduction year. In what year will you introduce this product to this market?
  • Units sold per month during first year. How many units do you expect to sell during each month of the first year of its introduction? This will be an array of unit counts, one for each product in each market during the first 12 months of the product’s introduction.
  • Annual increase for each subsequent year. The first annual increase will be used to calculate the product’s second year of revenue in that market. Its rate of increase will be based on sales that occurred during the last month of the 1st year, i.e., last month will be considered to be the running rate of sales for the 2nd year of ramp up.

11. “New Customers by Month” Sales Model (Wild Ass Guess)

Use this approach when customers become a source of recurring revenue once they have been acquired. For example, if you are selling (a) annual service agreements, (b) software access using a Software as a Service (SaaS) model, (c) membership-based products or services, (d) subscriptions, or even (e) utilities. This approach could even work for some (bricks-and-mortar or bits-and-clicks) retail stores with strong custo­mer loyalty programs because once customers are acquired, the company might be able to reliably forecast return visits and recur­ring purchases. For all such businesses, you should define these assumptions for each market:

  • New customers. How many new customers do you expect to obtain during each month? This will be an array of new customer data, one every market during every month.
  • Average order size. What is the average monthly recurring revenue attributed to each customer? This will be an array of average order sizes, one for each year.
  • Annual retention rate. What percent of customers will remain customers after 12 months?
  • Viral coefficient. How many new customers will each existing cus­to­mer attract?
  • Viral cycle. How many days elapse between each new generation of customer referrals, i.e., how long does it take (on average) for customers to “spread the word” and get friends to become customers?

[BES10] www.bvp.com/sites/default/files/bessemer_top_10_laws_ecommerce_oct2010.pdf.

[RIE11] Ries, E., The Lean Startup, New York: Crown Business, 2011.

This article is extracted from my book published by Scrub Oak Press titled Will Your New Start Up Make Money?. Buy it at http://www.amazon.com/Will-Your-Start-Make-Money-ebook/dp/B00JOOZQNE.