Rich posted the full research paper last week, but as not everyone wants to read the full 30 pages, we decided to continue posting excepts here. We still encourage comments as this will be a living document for us, and we will expand in the future. Here is Part Four:

Understanding Potential Losses

Earlier we deliberately decoupled potential losses from risk impact, even though loss is clearly the result of a risk incident. Since this is a business justification model rather than a risk management model, it allows us to account for major types of potential loss that are the result of multiple types of risk and simplifies our overall analysis. We will highlight the major loss categories associated with data security and, as with our risk estimates, break them out into quantitative and qualitative categories. These loss categories can be directly correlated back to risk estimates, and it may make sense to walk through that exercise at times, but as we complete our business justification you’ll see why it isn’t normally necessary.

If data is stolen in a security breach, will it cost you a million dollars? A single dollar? Will you even notice? Under “Data Loss Models”, we introduced a method for estimating values of the data your company possess to underscore what is at stake. Now we will provide a technique for estimating costs to the business in the event of a loss. We look at some types of loss and their impacts. Some of these have hard costs that can be estimated with a reasonable degree of accuracy. Others are more nebulous, so assigning monetary values doesn’t make sense. But don’t forget that although we may not be able to fully quantify such losses, we cannot afford to ignore them them, because unquantifiable costs can be just as damaging.

Quantified vs. Qualified Losses

As we discussed with noisy threats, it is much easier to justify security spending based on quantifiable threats with a clear impact on productivity and efficiency. With data security, quantification is often the rare exception, and real losses typically combine quantified and qualified elements. For example, a data breach at your company may not be evident until long after the fact. You don’t lose access to the data, and you might not suffer direct losses. But if the incident becomes public, you could then face regulatory and notification costs. Stolen customer lists and pricing sheets, stolen financial plans, and stolen source code can all reduce competitive advantage and impact sales — or not, depending on who stole what. Data stolen from your company may be used to commit fraud, but the fraud itself might be committed elsewhere. Customer information used in identity theft causes your customers major hassles, and if they discover your firm was the source of the information, you may face fines and legal battles over liability. As these can account for a majority of total costs, despite the difficulty in obtaining an estimate of the impact, we must still account for the potential loss to justify spending to prevent or reduce it.

We offer two approaches to combining quantified and qualified potential losses. In the first, you walk through each potential loss category and either assign an estimated monetary value, or rate it on our 1-5 scale. This method is faster, but doesn’t help correlate the potential loss with your tolerance. In the second method, you create a chart like the one below, where all potential losses are rated on a 1-5 scale, with either value ranges (for quantitative loss) in the cells, or qualitative statements describing the level of loss. This method takes longer, since you need to identify five measurement points for each loss category, but allows you to more easily compare potential losses against your tolerance, and identify security investments to bring the potential losses (or their likelihood) down to an acceptable level.

Potential Loss Categories

Here are our recommended assessment categories for potential loss, categorized by quantifiable vs. only qualifiable:

Quantifiable potential data security losses:

• Notification Costs: CA 1386 and associated state mandates to inform customers in the event of a data breach. Notification costs can be estimated in advance, and include contact with customers, as well as any credit monitoring services to identify fraudulent events. The cost is roughly linear with the total number of records compromised.
• Compliance Costs: Most companies are subject to federal regulations or industry codes they must adhere to. Loss of data and data integrity issues are generally violations. HIPAA, GLBA, SOX, and others include data verification requirements and fines for failure to comply.
• Investigation & Remediation Costs: An investigation into how the data was compromised, and the associated costs to remediate the relevant security weaknesses, have a measurable cost to the organization.
• Contracts/SLAs: Service level agreements about quality or timeliness of services are common, as are confidentiality agreements. Businesses that provide data services rely upon the completeness, accuracy, and availability of data; falling short in any one area may violate SLAs and/or subject the company to fines or loss of revenues.
• Credit: Loss of data and compromise of IT systems are both viewed as indications of investment risk by the financial community. The resulting impact on interest rates and availability of funds may affect profitability.
• Future Business & Accreditation: Data loss, compliance failures, or compliance penalties may impair ability to bid on contracts or even participate in certain ventures due to loss of accreditation. This can be a permanent or temporary loss, but the effects are tangible. Note that future business is also a qualitative loss — here we refer to definitive measurements, such as exclusions from business markets, as opposed to potential losses due to customer loyalty/concern.
• Continuity of Business: Denial of Service impairs customer service and interferes with business. These are often measurable for transaction-based businesses.

Qualifiable Potential Data Security Losses

• Reputation Damage: The reputation of a company affects its value in a number of ways. New customers often seek out firms they know and trust. Investors are likely to buy stock from companies which are trustworthy and operate effectively. Risks to reputation affect both, but it’s generally impossible to attribute an impact to a single event because other events, non-risk factors, and general market forces all feed into customer behavior.
• Customer Loyalty: How the data loss is perceived by customers has an effect on customer and brand loyalty. If the loss of the data is viewed as preventable, and the inconvenience or financial cost to customers is high, some customers will stop doing business with the company.
• Loss of Sales: Your customer contact information and pricing sheets in the hands of your competitor provide ample data for targeted sales campaigns. Any successes come at your expense.
• Competitive Advantage: R&D expenditure to create a new and competitive product can be devalued if that research, source code, process, or ingredient list is stolen; but since you aren’t blocked from still bringing the product to market, the lost benefit is not fully quantifiable.
• Future Business: You cannot accurately predict lost future business, unless you restrict it to market/ecosystem/contract exclusion as mentioned above. We’ve seen single breach disclosures put a company out of business, while other companies see sales growth despite major public breaches.

Exponential loss growth

While a single incident might result in minimal losses, a string of ongoing incidents is likely to exponentially increase losses- especially in qualitative areas such as reputation damage and lost future business. Despite what most of the surveys claim, there is very little evidence of correlation between single data breaches and lost business, or even stock price. For example, TJX suffered one of the largest data breaches in history, but sales increased steadily through the incident. It’s clear customers either didn’t pay attention, or felt that the security controls implemented after the incident made TJX safer to shop. But if TJX suffered an ongoing string of data breaches over a period of months, at some point there would be material loss of business.

When providing a business justification for security spending, you do not need to account for every single aspect of loss. Nor do you need to even show that the majority of data value is at risk. Instead, you need to understand and be able to show that valuable data is at risk, and examine the potential benefits of security and the reduction of loss in relation to cost of the investment. Simple monetary damages may be small, but the potential for loss can still considerable. If it can be shown that mitigating the risk vector associated with data theft also accomplishes operational goals, the argument is even stronger. If the investment also accounts for compliance controls, or makes a business process more efficient, the effort may pay for itself.