business

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 1/10

4. The Early Warning System

“Forewarned, forearmed; to be prepared is half the victory.”

—Miguel de Cervantes, author of Don Quixote.

Situational awareness is a term used to define the need for heightened awareness of one’s surroundings when entering an area of uncertainty or danger. In a business context, it is the ability to identify, process, and com- prehend on an ongoing basis the critical elements of information about what can and is occurring. The purpose of this chapter is to raise situational awareness on a continuing basis among companies concerning our ever- present companions: failure and underperformance. Usually, managers have a heightened sense of awareness during the creation and initial stages of a new endeavor, but as soon as that is over, their focus naturally shifts to other matters, thereby reducing that heightened awareness.

Chapter 1, “Failure & Stagnation,” revealed how in any major business en- deavor, whether it be the launch of a new business, product, or a campaign, there can be literally thousands of different options to choose from, which makes finding the optimal combination very difficult. As a result, businesses need to monitor those signals that provide the earliest possible warnings about emerging problems. If you are in a business­to­business (B2B) indus- try and selling equipment with a sales cycle measured in months, you may find yourself in irreversible trouble if you wait for revenue metrics. Tradi- tional outcome metrics, such as revenue or market share, are obviously nec- essary, but these should not be the key drivers of an early warning system (EWS). Instead, you need to identify other variables that provide more ad- vanced warning.

B A C KG R O U N D

In our daily lives, we see the widespread use of EWSs that provide advanced notice to prevent negative outcomes. Gauges on our cars are constantly measuring the engine temperature, oil, and gas levels. This might appear to be stating the obvious, given that we also have business dashboards that capture metrics like the number of orders, revenue, profitability, and so on. Although that is true, the majority of metrics used in business dashboards are lagging (e.g., revenue) instead of leading indicators (e.g., awareness). In addition, not all leading indicators are equally good. Some are critical suc- cess drivers, whereas others just add to the clutter. To finish with the car analogy, we tend to have on business dashboards gauges with lagging indi- cators, such as the check engine light, which appear when something has al- ready malfunctioned. Some business areas, such as web analytics, make good use of leading indicators, including them as key performance indica­ tors (KPIs), but they rarely separate them out from lagging indicators, pri- oritize, classify them as short term versus long term, or forecast what num- bers they expected, all of which are important components of the EWS.

Finance has been using a type of early warning mechanism for years, which is the Z­score. This score was introduced in 1968 by Edward Altman, a New York University finance professor, to predict the likelihood that a publicly traded manufacturing company might find itself in a state of bankruptcy 12 to 24 months before that event (whether the company actually takes the le- gal step of declaring bankruptcy is, of course, another matter). Altman sub- sequently introduced a Z­score plus and double­Z prime versions for use on privately held and nonmanufacturing companies. In tests over a 30-plus year period, the Z-score was found to be 80 percent to 90 percent accurate

in predicting bankruptcies one year prior to the event. Bond rating agen-

cies like Moody’s use a similar methodology, weighing many different fac- tors; however, some of the factors are subjective (e.g., management exper- tise). While subjective factors may be necessary, the fact that many of the companies being rated, such as Countrywide Mortgage, Lehman Brothers,

1

P R E V 3. The Business Failure Audit and the Domain Transfer of Root Cause Analysis

⏮ N E X T

5. Blind Spots and Traps ⏭

�

 Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode and E�ects Analysis, and an Early Warning System

Find answers on the fly, or master something new. Subscribe today. See pricing options.

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 2/10

Bear Stearns, and AGI were also major customers posed a problematic con- flict of interest. The Z-score, on the other hand, inherently avoids the bias and incestuous relationships demonstrated by many credit rating agencies during the run-up to the 2007–2008 crash.

The key takeaways of the Z-score formula for our purposes are the follow- ing: the use of different weights (e.g., EBIT or the C variable is more impor- tant than the Ratio of Working Capital / Total Assets or the A variable); a limited number of variables (more than two but fewer than six); and, lead- ing (predictive) indicators. Mainly for informational purposes rather than applicability to the EWS proposed here, Altman’s Z-score formula consists

of the following five variables and weights:

Z= (1.2) A + (1.4) B + (3.3) C + (0.6) D + E

A= Working Capital / Total Assets

B= Retained Earnings / Total Assets

C= Earnings Before Interest and Taxes / Total Assets

D= Market Value Equity / Total Liabilities

E= Sales / Total Assets

C R E AT I N G A Z-S CO R E M E T R I C F O R OT H E R A R E A S O F B U S I N E S S

Can you create a universal Z-score type of formula to predict the likelihood of failure for new product introductions, company acquisitions, mergers, marketing, or sales campaigns? It’s possible but highly unlikely for several reasons. Altman was able to create this universal score due to the availabili- ty of data from publicly traded companies that are required by law to pro- vide a predetermined set of financial variables and do so on at least a quar- terly and annual basis. Altman then compiled this data and compared com- panies that had gone bankrupt versus those that were successful. Through this process, he was able to use statistical analysis to identify key predictive variables (mainly ratios), assign differing weights to each, and arrive at a score that closely correlated with the likelihood of bankruptcy.

Unfortunately, there is no such publicly available data or standardized vari- ables for products or campaigns that fail compared to those that are suc- cessful. Moreover, good luck trying to obtain detailed data from private or even public companies, especially when it comes to their failures. There are mathematical and simulation models available to predict product or market failure or underperformance. The purely mathematical models (static) are not widely used by industry because of their special data requirements, lack of customization, and accuracy rate of less than 70 percent. Market simula- tion software has a much better track record, with accuracy rates closer to 85 percent to 90 percent and are more customizable, but this type software is used primarily for major product introductions and in categories such as packaged goods (e.g., food products), consumer services (e.g., banking, trav- el), healthcare, and consumer durables (e.g., apparel, electronics). There must also be a substantial historical body of data on the category so that the simulation can be calibrated correctly for the initiative being forecasted. These types of software are also proprietary, so there is a cost associated with it, typically in the tens of thousands of dollar range, depending on the modules needed and complexity of the business or venture. Moreover, large companies often outsource customer acquisition campaigns to agencies; as a result, the campaign itself is not being monitored for failure or underper- formance in real time. Therefore, although this software plays an important role in reducing failure in new product launches, it does not currently fulfill the role of an EWS, especially in the campaign area, because it focuses on lagging indicators and does not weigh or score the variances as an EWS should.

The main reason the Z-score is still widely used after 40 years by financial practitioners is its high accuracy rate and the fact that it can be calculated in minutes using a spreadsheet. Until that level of cost and simplicity is achieved, a one-size-fits-all tool will remain the domain of agencies and For- tune 500 companies and even then used mainly for forecasting new product introductions in certain categories and not as the EWS being proposed here.

An additional issue preventing the creation of a one-size-fits-all Z-score is that advertising, marketing, new product launches, and sales are incredibly heterogeneous. There are product extensions, launches in stable and low- competition environments versus those in volatile and high-competition in- dustries, short versus long sales cycles, different stages in the life cycle, and so on. There is also a wide range of platforms available for your campaigns, such as radio, TV, PPC, social media, personal sales, and innumerable com- binations of these, different levels of expenditures, and timeframes. Finally, every industry has its unique set of key drivers. Some are heavily impacted by commodity prices, others by repeat business, while others require ongo- ing customer acquisition and multiple combinations of these. In other

2

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 3/10

words, the level of complexity and heterogeneity makes a simple and univer- sal EWS nearly impossible.

Given the difficulty in creating some type of simple universal Z-score for products and especially for campaigns, a highly customizable and easy-to- build (using a spreadsheet) alternative is proposed and based on the same underlying principles as the Z-score.

T H E O P T I O N O F B U I L D I N G A M O R E S O P H I S T I C AT E D E W S

Companies with the expertise and historical data can create a more sophisti- cated model that either builds on existing forecasting software or create their own from scratch. However, that level of complexity and cost is neither the focus nor the objective of this book. The focus here is on solutions for the “everyday manager and company.” Domain transfers will never be wide- ly adopted when cost and complexity are a significant factor.

In the final analysis, a sophisticated statistical EWS is still not a replace- ment for a causal forecast. Both approaches are actually complementary rather than mutually exclusive. The statistical model eliminates subjectivity and biases that a causal forecast may contain. However, the causal forecast has advantages not found in the statistical model, such as how it can be built it in a few hours in a spreadsheet; the thinking process that goes into build- ing a causal forecast is as valuable as the EWS itself; and by having to list out the assumptions and key leading drivers, those responsible for the prod- uct or campaign are forced to be more in tune with the performance and ef- fectiveness of the different platforms used and decisions made. An addition- al benefit of the causal over a sophisticated statistical model is that when something goes wrong, the business professional in charge can easily and instantly determine where the problem resides and proceed to fix it.

Case in point: If you were given a typical (non-causal) forecast predicting that a new B2B product launch would generate sales of 80 units per month during the first year, this forecast might not set off any alarm bells. Howev- er, if you were shown a causal forecast, in a few minutes you could see that it assumed a conversion rate of 35 percent, which might be four times higher than what your company had ever managed to achieve in the past.

C R E AT I N G T H E E W S A N D I TS F O U N D AT I O N , T H E C A U S A L F O R E C A S T

The foundation for an EWS is to start by creating and fine-tuning a simple cause-and-effect type of forecasting method. There are many types of causal forecasts. Large companies use forecasting software that incorporates dozens of internal and external variables, including economic indicators such as changes in the cost of living, the unemployment rate, and so on. If you have the resources to leverage such a forecast to create an EWS, that’s great, but fortunately, this is not necessary. All you need is a spreadsheet and a few hours to think through what initiatives generate revenue for your business. One driver is obviously repeat business, which might account for 40 percent of your revenues, but what about the rest? In the causal forecast, you detail all the different drivers that bring in customers, such as word of mouth, direct mail, digital campaigns, salespeople, and many more.

The casual forecast basically consists of assumptions, leading indicators (some are key revenue drivers, whereas others are foundations for success such as your distribution reach), “connector” variables, lagging indicators, and the extensive use of variance analysis between your actual results and forecasted goals. Figure 4.1 provides a high-level overview of the process for creating an EWS.

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 4/10

Figure 4.1 Overview of the creation of a causal forecast

Step 1: The Assumptions

The assumptions are one of the most critical parts of both the causal fore- cast and EWS. Any business initiative, be it a financial investment, company acquisition, sales, marketing, or advertising effort, is based on certain un- derlying facts and evidence that drove the decision to pursue that particular initiative (e.g., launch a product or conduct a campaign). Very often, these assumptions are implicit, not detailed, and assumed to be true. For exam- ple, when proposing the acquisition of a company, we may make assump- tions that are benign and probably factual while others lay the groundwork for failure. We might implicitly (perhaps correctly) assume, for example, that the economy will not go into a deep recession during the payback peri- od, given that many economists and indicators show a healthy economic outlook for the next five years; or we may (incorrectly in this case) implicitly assume that the 12-month forecast provided by the seller is realistic, despite not having dissected the forecast and assumptions.

The assumptions should be the first part or section in any causal or other type of forecast and thus EWS. Assumptions require critical thinking and thoroughness because they can prevent serious blind spots and are impor- tant for troubleshooting purposes should your actual results deviate from your initial forecast. The list of assumptions should also include a detailed explanation of the basis upon which they are made and/or what evidence was used to make them. Table 4.1 shows an example of how to list and detail assumptions. Item 2, for example, states that the conversion rate is based on an industry benchmark. Any generic statement like this should be chal- lenged to see how applicable it is to your particular situation.

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 5/10

Table 4.1 Example of Assumptions in a Causal Forecast

Unfortunately, assumptions are often incorrectly estimated, based on “gut” feeling, and more often than not completely left out. An additional benefit of requiring this level of detail is that the manager responsible for the initiative will do a more thorough fact-checking job before sharing it with manage- ment. Finally, and if during a post-failure audit it is revealed how the fact checking was haphazard or an outright misrepresentation, that should be cause for corrective action, given that laziness or dishonesty are hard-to-fix root cause problems. For new products, the awareness level and distribution reach if applicable (Number of Outlets You Are In / Total Number of Outlets Available for Your Category) should always be included in the assumptions section.

Leading Versus Lagging Indicators (The Foundation for Steps 2–4)

Lagging indicators (e.g., revenue) are those that follow an event(s), the re- sult at the end of a given timeframe. Leading indicators (e.g., orders booked) are those that signal future outcomes and those that feed directly into the performance of an outcome (lagging). Performance management, economics, finance, digital, and direct marketing have all been using a com- bination of leading (forward-looking) and lagging (rear-looking) indicators for many years. Lagging indicators for economists are outcomes like the un- employment rate or a recession.

For economic leading indicators, one independent research association cre- ated The Conference Board Leading Economic Index for the United States,

shown in Table 4.2. As with the Z-score, each leading variable has been

assigned a weight. However, given the complexity of an economy like ours, they used ten variables. This index has proven successful in predicting re- cessions, although it has given some false positives as well, predicting im- pending recessions that never materialized. In defense of this index, when the economy is starting to show signs of a recession, the Federal Reserve of- ten starts taking actions that defer or prevent the forecasted recession.

Source: The Conference Board Leading Economic Index

Table 4.2 The Conference Board Leading Economic Index

One other type of indicator is called the coincident. It usually occurs at the same time as a leading or lagging indicator (e.g., “company profits” is a lag- ging indicator, and “average employee bonuses” would be a “coincident” in- dicator). When a company has sophisticated resources at its disposal, coin- cident indicators can be useful to pinpoint the dates when peaks or dips oc- cur in the business cycle.

3

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 6/10

The EWS focuses mainly on leading indicators—which are the cornerstone of the EWS—and whose relevance must in turn be validated by lagging indi- cators and the “connector” assumptions that join the two types of indicators. For example, if you identified “orders booked” as a key leading indicator, this would need to be validated by a lagging indicator (e.g., revenue). How- ever, what if “orders booked” was tracking the same as during a previous pe- riod and yet the company’s revenue had actually declined? In this situation, there might be other leading indicators to better explain the decline, or you might have omitted other leading indicators, or perhaps made a faulty as- sumption in regards to the “average order size.”

Step 2: Identifying All the Available Data

One approach to make sure that you do not leave anything out is to identify all the data variables you capture or could capture and then proceed to cate- gorize and tag them as leading, lagging, or coincident, and then also as short term or long term. You may find that an increase in returns or complaints and a decline in unique website visitors are the earliest and leading indica- tors of the subsequent revenue decline. Chapter 5, “Blind Spots and Traps,” provides some additional details on how to make sure that you are not miss- ing out on relevant data points.

Step 3: Best Practices When Selecting and Categorizing Leading and Lagging Indicators

• Short time frames: Using monthly or quarterly data defeats the raison d’être of the EWS. Make sure that your causal forecast is broken down into daily or weekly time frames. Also resist the temptation to use averages. In- stead, mimic any seasonal or day-of-the-week patterns your business ex- hibits. For example, if 70 percent of your orders occur on the weekend or during the last three months of the year, it would undermine the effective- ness of the EWS if you allocated them evenly throughout the weeks or months of the year.

• Customers: Separating new from existing customers is important be- cause some of the leading indicators and connectors will be different. For example, in some companies, repeat business may account for 75 percent of annual revenues (e.g., a supermarket), whereas in other businesses (e.g., an appliance manufacturer), repeat customers may only account for 25 percent of annual revenues. Another benefit in separating new from existing cus- tomers is that it will save you time when troubleshooting a decline in perfor- mance and revenues.

• Heterogeneity: Because causal forecasts are incredibly heterogeneous, the leading indicators will be based on additional factors driven by your dis- tribution model (online only versus a combination of brick-and-mortar and online versus indirect only through retailers), the average consumer deci- sion-making purchase cycle (days, weeks, or months), and the stage in your product life cycle (as you move into each stage, you should adjust your lead- ing indicators and assumption connectors).

• Promotional drivers: When creating a causal forecast, the primary but not exclusive focus is on “promotional” type leading indicators (e.g., ads, email, PPC, personal selling, search engine optimization [SEO], or catalogs) that generate revenue (lagging indicator). However, some leading indicators will not be promotional, such as awareness (do prospects even know your product or brand exists), distribution reach (e.g., in how many retailers is your product available), or on customer complaints. Table 4.3 provides some examples of both promotional and non-promotional leading indica- tors. Note that there are other non-promotional drivers such as price and distribution, but these will not be used when creating the EWS. Instead, they will be used for troubleshooting purposes if and when a variance occurs between your forecast and the actual results.

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 7/10

Table 4.3 Examples of Different Types of Leading Indicators

• Special leading indicators: Some leading indicators are either difficult or expensive to capture on an ongoing basis, such as awareness or brand re- call. Short of large, statistically representative surveys or making assump- tions based on the reach and frequency of your advertising (e.g., gross rating points), which is usually the domain of large companies, these types of lead- ing indicators are not easy to track on a continuous basis. Table 4.3 shows a few examples of these special indicators.

• Using social media listening platforms to capture leading indi­ cator sentiment: For some of these difficult-to-quantify leading indica- tors, some businesses with a strong digital customer engagement level might be able to aggregate several variables using a social media analytics platform to come up with a plausible “awareness index” (Volume of Product Mentions / Total Category Mentions, along with some positive versus nega- tive sentiment weighing). Another option is to consider using the keyword search term query volume tools provided by Google and Bing, which you can then use to compare to brands in your category (e.g., if the three top brands in the category have a combined total of 100,000 monthly searches and Brand A has 20%, Brand B 50%, and Brand C 30%, this can be used as a rough indicator of consumer preference). A cautionary note if using a social media listening platform: Many conversations are private (e.g., many Face- book posts), so make sure that any heavily weighted source such as Twitter (which usually has the largest volume of publicly available data that these tools are allowed to access) is a statistically valid representation of your tar- get market. If not, you could be making assumptions based on Twitter men- tions by users who might represent less than 2 percent of your target market.

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 8/10

Example of a Social Media Analytics Platform Used to Capture a Lead-Example of a Social Media Analytics Platform Used to Capture a Lead-

ing Indicatoring Indicator

Figure 4.2 shows how Schick had a significant increase in Facebook fans during a certain period; additional re­ search revealed that Schick was giving away product by bundling it with other online product offers through a part­ nership with key online retailers and a sports website. So while the increase in Schick Facebook fans was modest compared with Gillette’s base, this could act as a leading competitive indicator that could interfere with Gillette’s promotional campaign. The social media listening plat­ form may also capture customer complaints and reveal that your sampling campaign is not performing as expected.

(Source: MutualMind)

Figure 4.2 Example of a social media ana­ lytics platform and leading indicators

Lagging indicators are the easiest to identify. These are the outcomes through which your company defines success. Examples are metrics such as the number of orders, revenue, profitability, market share, return on invest- ment, lifetime value, net promoter score, and so on. The most important consideration when selecting the lagging indicators is to make sure that you have picked the leading indicators that are best translating and driving your lagging indicator.

InsightInsight

For companies that sell directly and only online, the lines between leading and lagging indicators may be blurred, given that a sale may occur on the same day as the lead­ ing indicator (unique visitors and the order). However, with additional brainstorming and research, you can find other leading indicators. Perhaps a downward trend in your website’s organic search page rank for high­convert­ ing keywords or average PPC ad rank position can presage future declines in website traffic and orders.

Step 4: Adding the “Connector” Assumptions

Connectors are those values, often expressed as percentages, that translate promotional leading indicators into lagging indicators as shown in Table 4.4. These connector variables can encompass a wide range of variables but are usually metrics such as click-through rates, conversion rates, average or- der sizes, repeat purchase rate, and so on.

Table 4.4 Example of Leading, Connector, and Lagging Variables

Step 5: Entering Leading, Lagging, and Connectors into a Spreadsheet

The next steps are all outlined in greater detail in the Appendix with a sam- ple spreadsheet in Table A.1, that shows where the leading, connector, and lagging indicators are entered and also where, for each indicator and con- nector assumption, “actual” numbers are entered next to the forecasted values.

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 9/10

Step 6: Calculating the Variance

The “actuals” are the results (numbers) that come in after you complete your different campaign initiatives. Most variance calculations are pretty straightforward (e.g., the [Actual results-Forecasted results]/Forecasted re- sults) as the example in Table 4.5 shows. Moreover, when the “actuals” are greater than the forecasted amount, that is usually a good thing, and you have beat your forecast. However, there are a few situations or cases where you are tracking a ranking (where ad position 1 is better than position 3); how to deal with these cases is explained in the Appendix, in Table A.2.

Table 4.5 Example of a Typical Variance Calculation

Step 7: Calculating a Weighed Scored

This step requires calculating weights for the lagging indicator so that you can prioritize your early warning system. Table A.1 in the Appendix shows how the weights are calculated (the number of new customer orders fore- casted for a given initiative divided by the total number of orders forecast- ed). Each driver will have its own individually weighted score, which can be negative or positive; with a cumulative score for all the lagging indicators. The objective of the weighed score is so that you can focus on the higher negative values in your EWS to see what needs to be prioritized.

Step 8: The EWS Dashboard

Table 4.6 shows what the EWS section of your business dashboard might look like. These measures are all entirely customizable, with some compa- nies perhaps wanting to show only negative variances to reduce the clutter. Details of how these numbers were obtained are explained in more detail in the Appendix in Table A.1.

Table 4.6 EWS Dashboard

2018/4/7 4. The Early Warning System - Breaking Failure: How to Break the Cycle of Business Failure and Underperformance Using Root Cause, Failure Mode a…

https://www.safaribooksonline.com/library/view/breaking-failure-how/9780134387000/ch04.html#ch04lev2sec10 10/10

Recommended / Playlists / History / Topics / Tutorials / Settings / Get the App / Sign Out © 2018 Safari. Terms of Service / Privacy Policy

Step 9: Troubleshooting: When the EWS Shows Underperformance

When multiple negative variances appear, focus on the higher EWS scores because those are the ones that will impact the lagging indicators (e.g., rev- enue) the most. In Table 4.7, we are only highlighting two important nega- tive indicators for the sake of brevity. Without running a regression, there is probably some correlation between some of the assumptions and the nega- tive outcomes. For example, the decline in “average ad rank position” is probably a contributor to the negative leading indicator “PPC” and its poor conversion rate. When an ad rank drops (it becomes less visible to most prospects), you can safely assume based on experience and historical data that the click-through rate and conversion rate will also decrease. The Ap- pendix shows a detailed troubleshooting tree in Table A.5.

Table 4.7 Negative Variances for Troubleshooting

Because the EWS provides such granular detail of where the problem re- sides, conducting a root cause analysis (RCA) will take a lot less time now. In this example, for the PPC negative variance, the drop in ad rank position could be due to many causes, some of which would be easy to identify (e.g., a lower-than-suggested bid), whereas others may be less obvious (e.g., poor quality of the written ads).

Once you find a failure cause, look back at your controls and detection mea- sures in the FMEA for this initiative and add or tighten any preventive con- trol or detection measures.

The EWS is both a learning and preventive tool. The learning component resides in that it helps business owners or managers carefully think through their key revenue and other leading indicator drivers. These are usually not broken down and evaluated in any systematic manner by managers. Only good things can happen when the manager has a good pulse on what is and isn’t working and what drivers matter the most. The preventive component of the EWS is more obvious because it alerts early on about the underperfor- mance so that prompt corrective action can be taken. The benefits are no different than with the early detection of diseases or conditions like diabetes or high blood pressure versus finding out you have them while at the emer- gency room.

P R E V 3. The Business Failure Audit and the Domain Transfer of Root Cause Analysis

⏮ N E X T

5. Blind Spots and Traps ⏭