Forecasting Sales: how does a Franchisee make a realistic sales revenue estimate?

Many franchisees are opening their first business (often in retail), and are risking all their hard work, house, life savings, right up to their kid’s toys to afford the money to begin their new venture. They will have done a business plan, normally with their Franchisor, and they think all looks positive.

The question Peter Buckingham, Managing Director of Spectrum Analysis , a business services provider for mapping, demographics, and area planning to enable businesses to make better decisions, poses to these franchisees is how does one realistically make a sales revenue estimate for a new business?

In many cases the Franchisor will offer to provide some information such as the demographics of the area, and maybe a listing of phone numbers of other Franchisees to talk to. The final word from the Franchisor is along the lines of, “do your own due diligence, as I cannot say what sales revenues you will generate.”

In many cases this is the truth and following some legal cases over the last few years, most Franchisors are not prepared to give any concrete sales forecasts, as they run the risk of having these held up against them in court. So where does that leave the Franchisee?

Forecast P & L

In most businesses people think in terms of forecast profit, and they rely on this to live, pay back their debts and basically secure their future. The simplest way of looking at a forecast P & L is as follows: 

    Sales Revenues – very hard to predict

Less Costs 

    Cost of Goods Sold (factor of sales revenue) 
    Rent – fairly predictable cost 
    Wages – fairly predictable cost (with some variable component) 
    Other (power, phone, all fairly small) 

    = PROFIT

Whilst most of the costs are fairly predictable, and the cost of goods sold is a factor of the total revenue, the sales revenue is the most unpredictable factor in a simple Forecast P & L.

If any of the other factors are out by a small percentage, there is a relatively small effect on the PROFIT (up or down).

If the Sales Revenue is out by a fair percentage, say +/- 50%, then the effect on the bottom line is either fantastic or catastrophic.

So how one does a good sales revenue estimate for the future is the “$64,000 question”. 

    1. Research – the analog approach

The simplest way to make a sales revenue estimate is with the analog approach. This means think of a future store or business, and look for similar stores or businesses in similar situations. Then try and find out what their sales revenues are, and use that as the basis for forecasting.

An example may be going into a food court of a shopping centre, and the franchisor has 10 other stores in similar situations. Shopping centres are normally measured with a few main factors: 

• GLAR – Gross Leasable Area Retail – or how many square meters the centre is. 
• MAT – Moving Annual Turnover – total amount of dollars transacted in a 12 month period. 
• seats in the food court 
• number of cinema screens; and 
• several other factors.

Take, for example, a shopping centre with a GLAR of 70,000m2 and a MAT of $300M, with 600 seats in the food court, and no theatre screens. One should then find out if similar shopping centres exist and if the Franchisor has sites in them. Normally one will not find exactly the same analog, but looking for say two variables that are similar to the shopping centre being considered is a good start. The first 2 Mr Buckingham would suggest are GLAR and MAT.

In shopping centre terms, centres are grouped using the Property Council Definitions by their GLAR as follows, and these are a good start to compare apples to apples: 

• Super Regional Centres 85,000 + Square meters 
• Major Regional Centres 50,000 – 85,000 square meters 
• Regional 30,000 – 50,000 square meters 
• Sub Regional 10,000 – 30,000 square meters 
• Neighbourhood < 10,000 square meters

The Franchisor will suggest doing due diligence, and one of Mr Buckingham’s suggestions is to talk with Franchisees in similar types of centres, not only about the franchise system, but also about their sales expectations.

From a Franchisor’s view, they should be able to do quite a deal to give themselves comfort in approving a store, whether they are at liberty to pass that on is up to them.

If a Franchisee has a network of stores, they can learn from their sales history. A Franchisor should have their sales, and be able to bring together other information to explain why the good stores are good, and why the poor stores are acting that way. In some cases it may be the Franchisee, however in other cases, Superman could not run it at a profit. 

    2. More statistical methods

At some point potential Franchisees may reach their practical limit with the logic above. However a Franchisor or a Master Franchisee with multiple stores can do it better.

Regression modelling is a proven statistical technique that looks at which individual factors contribute positively or negatively to a store’s performance.

Once the strongest individual factors are identified, these can be combined by statistically trained people to create a prediction model.

As an example: In a gift shop chain in shopping strips, it may have been concluded from simple regression that the main positive and negative variables for sales are: 

1. Number of households within 3 kms radius is a positive 
2. Higher household income is a positive 
3. Size of shop is a positive 
4. Number of retail shops in the strip is a positive 
5. Number of Coles or Woolworths supermarkets in the strip is positive 
6. Section in strip – Busiest, middle, quietest section - rated (3,2,1) is positive 
7. More gift shops in the shopping strip is a negative.

If there is a large network of 40 + stores, there may now be enough for a multiple regression model which may look something like:

Monthly Sales Revenue     =     2.55 X Number of Households 
                                                       + 1.28 X Average Household Income within 3 kms 
                                                       + 256 X sq m of store 
                                                       + 25 X Number of shops in the strip 
                                                       + 2,567 X Number of Coles + Woolworth Supermarkets 
                                                       + 5,400 X rating of section of strip 
                                                       - 3,233 X number of other gift shops in the strip

This will deliver a sales revenue prediction, tested by looking at each individual store, and how closely the sales predictions are to the actual sales.

The closer they come to the diagonal line, the better the prediction model.

As a potential Franchisee, it is comforting to know a Franchisor uses some statistical process before they approve the site where capital is being placed. Whilst they may not be prepared to show their internal workings and calculations, confidence in the process they use should help in making good decisions for both parties.

In summary, sales revenue estimates are the number a Franchisee will be building their business plans around, and the better the logic and the process that generates this, the higher the chances of the business meeting expectations.

As a Franchisee one cannot access all the resources their Franchisor has, but they can use an analog process to at least give themself some logic in what sales revenues should generate in a new business venture.

Spectrum Analysis Australia specialises in assisting clients with decisions relating to store and site location using various scientific and statistical techniques.