Definition of Various Metrics

Revenue Metrics

Outward-facing metrics that relate to a company’s outreach to customers. They define how well a company is identifying, communicating to, and selling to new customers.

The Traditional Enterprise Sales funnel is heavily reliant on traditional revenue metrics that are tabulated below. Examples of these traditional metrics include new leads, new qualified leads, expressions of interest, meetings with the correct decision maker, soft-circle sales, and completed sales.

Online and Web-based Sales tend to be reliant on dynamic metrics. Examples of these kinds of metrics are a “controlled vocabulary index”, “real-time sales volume by item category”, “co-occuring sales”, and “co-occurring viewing.”

Traditional Enterprise Sales Funnel

Sales exist on a continuum from personal to impersonal. Traditional enterprise sales exist on the personal end of the spectrum, where traditional (not dynamic) metrics predominate.

Enterprise sales are defined as sales by full-time sales people. Generally, these are large sales of complex equipment or services. The process is very expensive; Eggert’s rule of thumb is that a sales effort should generally not be undertaken unless the sale will result in at least a $250K one-time sale or $100K per year. The sales funnel metrics are:

New leads: a person whose name and contact information you have, who you know works at a company that would potentially be interested in your company’s goods or services.
New qualified leads: in order to be a qualified lead, the sales team needs to establish that the company might be interested in making a purchase within a certain time frame, and that the company could make the purchase if they wanted to. IE, they have a plan to buy your products/services, as well as the budget to make the purchase.
Expressions of interest: an expression of interest occurs when the correct decision maker states that they may be interested in making the purchase from you. The correct decision maker is the person who is able to sign a purchase order. Generally, this is not the same as the original “new lead.”
Meeting the correct decision maker: a meeting with the correct decision maker, which may not take place until after several meetings with other people, who may not have real influence within the client company.
Soft-circle sales: occurs when terms and price have been negotiated, and the correct decision maker has agreed to the sale.
Contract sales: a closed sale, generally with a signed legal contract.

Online Sales

Online sales take place through web or mobile based interfaces, and exist on the other end of the sales spectrum from more traditional enterprise sales. These sales do not involve human interaction at all.

Amazon does an incredible job of exploiting the potential of dynamic revenue metrics. Examples of the types of analytics methods Amazon uses include patterns in its visitors’ click stream data which includes the pattern of clicks, cursor movements, and movements from page to page. This data is used to customize different users’ experience in real-time, based on that particular user’s historical and clickstream data. This clickstream data enables Amazon to rank the item results of a given user query by the probability that the user will purchase each of those items.

In addition to clickstream data, Amazon also appears to maintain what is called a “controlled vocabulary index,” which is a means of correlating a users’ search terms to a given predefined list of high level categories. A system like this would enable Amazon to give users the items they actually want, rather than those the user thinks they want. As an example Eggert presents his own search for information learning theory, which, leveraging the controlled vocabulary index and pattern matching techniques, infers that the pertinent categories for this search include the following: Computers & Technology, Artificial Intelligence & Semantics, Applied Mathematics, and Computer Vision & Pattern Recognition. Note these categories contain terms not included in the initial search. Nonetheless, they are actually pertinent results given the meaning of the search terms.

Once Amazon has identified the appropriate categories, it rank orders the items in those categories by those that are best-selling, and, therefore, most likely to be purchased.

Based on the observations above, we have good evidence that for every item ID (or book ID), Amazon must store a tree of subject categories and subcategories in addition to a record of how each book ranks in Amazon’s recent sales within each category.

Further down the page, there is a section called “Frequently Bought Together” that lists other relevant books, in this case. This indicates that Amazon also maintains a database of “co-occurrence data” that describes how frequently various items are purchased together in the same user session. Which specific books Amazon suggests in addition to the specific one being viewed may be based upon simply rank0ordering the most frequent co-occurring book by sales, or it may be based upon A/B Testing that indicates listing a certain book along with the book being searched results in a higher probability that both will be purchased together.

There is also a section called “What Other Items Do Customers Buy After Viewing This Item?” This indicates that Amazon also builds a co-occurrence database for which users viewed which pages on the site. It is a staggering amount of detail.

Profitability/Efficiency Metrics

Cash Flow versus Profitability

Business Analytics is not Financial Accounting, but it does draw on a few pertinent accounting principles. This small section discusses these principles.

Cash Flow metrics are distinct from Profitability metrics. In fact, they are discussed in different financial reports. Profitability is presented in the Profit & Loss (“P&L”) Statement, whereas Cash Flows are discussed in the Statement of Cash Flows. Unprofitable companies can thrive for years (think Amazon and Tesla). Even massively profitable companies will go bankrupt immediately, if they can’t meet short-term financial obligations (cash needs) quickly enough. This was the situation AIG, the largest insurance company in the world, faced in the Fall of 2008. This can actually be triggered by dramatic sales growth.

Net 60: payment terms indicating a supplier is paid 60 days after it delivers the merchandise.
COD or Cash On Delivery: payment terms indicating a supplier is to be paid immediately upon delivering the merchandise.
Variable Costs: costs that vary with the amount of goods/services produced. Examples might include fuel, hourly wages, and transportation costs.
Depreciation: allocation of capital expenditures over time. An example of a type of depreciation is “straight line depreciation,” where the book value of capital assets decreases by a fixed amount each time period.
Fixed Costs: costs that do not vary with the amount of goods produced. Examples include rent for the building, utilities, insurance, business licenses, and G&A expenses.
G&A expenses or General & Administrative expenses: examples include salaries for bookkeepers and office managers.
Net Earnings or Profit: revenue - variable costs - fixed costs.
Negative Float: a situation where a company must spend cash upon receipt for an order, for which it will not receive payment until a later time.
Accounts Receivable: money that customers owe a company for products already delivered, but which the company has not yet received payment on. This money is booked for purposes of calculating profits and losses, but it does not show up on the cash flow statement because it has not been received.
Aged Receivables: how long it has been that money not yet received has been owed to the company.

Inventory Managements

Inventory management is one of the best ways that operating companies can reduce costs, and maximize profitability, without lowering quality. Inventory is defined as a finished product waiting to be bought. It costs money to have finished items on hand, so the more time a product sits “on the shelf” the less efficient for the manufacturer. There are four reasons for this:

  1. Negative float: the working capital associated with “fronting” the cost of manufacturing something carries an interest expense.
  2. Fixed cost of storage: it costs money to heat, light, guard, and pay rent on warehouses and retail stores. The amount of this fixed cost that must be allocated to a given product is proportional to the time on the shelf.
  3. Wastage: inventory items lose value the longer they are in inventory. Extreme examples include bread and restaurant food. Other examples include anything with an expiration date, or hotels and airline tickets that are a complete loss if not sold.
  4. Obsolescence: fashions may have changed, or technology may have improved.

The metric that describes how much inventory is on hand is called “average days inventory,” or “days inventory,” for short.

Days inventory is not usually reported by public companies, but it can be estimated by dividing the “annual cost of goods sold” by the “year end inventory.” In the case of Walmart in year-end of 2014, those numbers were $358.1B and $44.9B, respectively. The estimated days inventory turnover for the year is therefore approximately 44.9/358.1*365 or 46 days. Walmart therefore sells all of its inventory on average within a 46-day period.

Companies that have embraced data analytics will actually track this information at the level of individual SKUs, which is more a dynamic metric than the traditional metric described above.

Hotel Rooms and Airline Tickets

Two interesting “wasting assets” are airline tickets and hotel rooms. The cost associated with each of these products are almost entirely sunk, or fixed, costs. In the case of the former, variable costs are limited to some very marginal increase in fuel usage and a soda and some peanuts. In the case of the latter it is some fraction of a housekeeper’s time, the cost of supplying clean linens, and some shampoos and soaps.

Examining the occupancy rates for hotels will reveal that there are patterns that vary by day of the week, as well as week of the year. In particular, Wednesday tends to be a high-occupancy day of the week, and September, October, March, and April tend to be high-occupancy months of the year. Given that demand varies like this, a variable pricing strategy is appropriate. More should be charged for Wednesdays in general, and especially more for the busy months. Less should be charged for Friday evenings, which tend not to be a major travel day, especially those following major holidays.

Rack rate: the undiscounted, publicly-available, “listed” price for a hotel room.
Floor rate: the theoretical, break-even price for a hotel room.
Promotional rate: a rate between the rack and floor rates that may be offered to repeat customers with some brand loyalty. As an example, loyal customers may be given an offer to rent a room at 80% of the list price, if the hotel believes that there is a less than 80% likelihood that the room will rent at list price before it expires.

Of these, only the floor rate remains constant. The rack and promotional rates vary as a function of day of week and week of year, among other factors.

Risk Metrics

Leverage Risk

In the US, before the financial crisis of 2008, banks were allowed to borrow $33 for every $1 of equity they held. These lenient regulations made it possible for banks to leverage themselves very highly. As a thought exercise, consider a bank with $10M in equity. They can borrow $330M, which they can then loan out. Assume they borrow at 2% and lend at 3%. The gross profit they could potentially earn on that $330M is therefore $3.3M, a 33% return on their $10M in equity. Unfortunately, levered that highly, only a few of those loans needed to be in default before the bank becomes insolvent.

Reputational Risk

Reputational risks are risks that can damage the perception of your brand, possibly impacting your ability to sell in the future. This is an area of particular vulnerability for restaurants and grocery stores, whose food products always carry some small probability of making their customers sick.

Costco is an example of a company with excellent risk mitigation capabilities. In order to obtain a Costco membership, you must provide various types of contact information to Costco, which they can then use to contact you in the event of a food recall. Since Costco also stores a record of every purchase made by every customer at the SKU level, if a manufacturer recalls a product, Costco has the means to rapidly identify and contact the list of customers who are at risk of being made sick.

This content is taken from my notes on the Coursera course “Business Metrics for Data-Driven Companies.” It is sponsored by Duke University and the course content is presented by Professor Daniel Egger.