SQUIGGLY^©

History

Over the 34 years of its existence, H K Ball and Associates has assisted a wide variety of companies in making decisions, from what to do about what happened last week, to how in detail things should be done over the next five years. In every case a major amount of effort has been spent trying to figure out what has happened, which has been pieced together from accounting records and from fragments of assessments from various managers.

Decisions, in order to be timely, are made with incomplete information. Were it not that way, no decision would ever be made. Managers, and the consultants that help them, really guess at everything. Giving advice based on incomplete information has made us acutely aware of the deficiencies in management information systems, not in regard to precision, but with management's ability to perceive what the data means. It is humanly impossible to synthesize the stacks of tabular computer printouts now used. With harder times for most companies, a sharper pencil is needed. This system of thinking, called squiggly^©, provides that sharper pencil. All of the above are achieved with the press of a key, or two at most.

What it is

Squiggly^© is a way of thinking that uses both logic and intuition in interpreting historical operating data arranged into line graphs (Figure 1). Line graphs enable a manager, for example, to "see" the impacts of 52 or 365 days of meaningful report data on one piece of paper. Meaningful report data can change from moment to moment, from situation to situation. The visual patterns themselves suggest what items are meaningful. Squiggly^© provides a process for looking at arrays of curves on a screen representing days or weeks or months or years of operations data, arranging them and rearranging them in order to compare trends caused by events (Figure 2). Curve groupings can be changed instantly as the messages they convey dictate. Curves can be added, subtracted, ratioed with each other producing new curves instantly.

The system is designed for key managers to use, not the analysts. For in the largest sense, it is the experienced managers who have the best intuition, and are therefore the best diagnosticians. The process is much like that of a medical doctor looking at an EKG line graph of a heart beat. Once he has made his diagnosis, part based on knowledge of the patient (the company), he then uses his intuition to interpret each little squiggle of the curve (the operating performance over time) to scribble a notation on a piece of the chart, reminding himself and others what the problem is and what should be done. Figure 3: Probe

How It Works

Figure 2 shows what a screen of actual data looks like. It represents 16 curves of overall sales and cost data with 52 weeks in each curve. These can be compared with the same data and time period for a given division in a given year. Vertical event lines indicate time periods for comparing relative performances of each curve .

Juggling the combinations of curves on the screen, analyzing them closely by enlarging them, the manager can finally get a picture, on the screen, of the thing he's trying to get a handle on. He then can send what he sees through Probe in Figure 3, or produce a hardcopy of the page, circle the critical points on the interrelated curves, and scribble a note across the bottom to a subordinate saying something like "Joe, notice the sales and cost performances I circled for Jan-Feb. when we ran that promotion. We may have some answers here. Check it out." Joe can then dig into the numbers in detail and validate the manager's intuitive hunches.

How It Does It

The foundation of the system is the hierarchy, that is the arrangement of client data into a form that can be accessed at random to createany combination of data curves on the screen. It must be random, because the manager uses what is on the screen to tell him what is needed next. Has mind may jump to seemingly unrelated data, from an accounting point of view, to get to a problem. The pre-destined hierarchy, which may include up to 20 years of operating data, can enable an infinite number of choices and comparisons, which is what the mind needs to make its assessments. Any one of hundreds of thousands of curves can be accessed with three or less key strokes, the screen filling in three seconds or less.

One or more curves can be added together to produce a new curve instantly, complete with the event impact percent changes described above. In seconds a cost curve can be divided by a sales curve to produce a new curve with all analysis. A price curve can be multiplied by a volume curve and added or subtracted from several sales curves to produce another curve.

Up to 30 formula curves can be made on one screen. Each formula may include constants for add, subtract, multiply and divide (Figure 4).

The program starts by presenting a group of up to 30 curves on the screen. The manager, upon inspection, may want to compare this grouping with that of another division, or another year of this division. He may want to look at the last seven years of a particular curve for this division. Any curve or group of curves can be compared with any other.

Vertical event lines may be instantly placed in any number anywhere on the screen. The program calculates percentage variations during the event impact period from the average of the curve for the year. These can be compared to the same numbers for other curves to discern difference in impact. The average variations can be shown in dollars as well as percent.

A number appears to the right and above each curve giving the increase/decrease percent of the second six months versus the first six months. This becomes very useful in quickly comparing all the curves on the screen.

Sometimes the curves for weekly periods are hard to interpret because of events in adjacent weeks. Curves with 2,3,4 and five week smoothing can instantly be created. One week through five week smoothing for one curve can appear simultaneously on the screen. The line curves can permit the analysis of old product flow data to find impacts years ago that were never found or understood (Figure 5).

Once the key curves are on the screen, the others can be eliminated and the remaining curves enlarged as in Figure 6. The manager, once satisfied that the screen shows the point he was driving at, sending it off through Probe^©.

How It Makes Money

Because the data is visible as trends, management is able to see impacts of decisions and events weeks, months and often years earlier than with presently available technology. More decisions can be made sooner with less data. Like driving a car, it is possible to stay closer to the center of the road (your plan) if you can make an adjustment to the steering wheel three times per second instead of once every three seconds.

Test studies trying one strategy with one group of company units versus another becomes much more realistic. Studies with curves over time reveal that different operating units have different signatures. More than cyclical patterns, these are little wiggles and shapes that occur only in a certain type of unit. As these are recognized, the experienced human eye can differentiate between signature characteristics and actual operating impacts caused by the study. This has been very useful in examining the impacts of pricing, promotions, changes in line operating procedure in general. They tend to reveal much more explicitly which kind of strategy works in which kind of operating unit (as characterized by its signature).

An interesting phenomena has been discovered in advertising effect where is an initial impact which then disappears totally in a relatively short period, followed by a permanent increase sometimes much later on. In order to locate the phenomena, periods can be varied quickly as in Figure 7. It was only possible to differentiate this effect from the impacts of other stimuli using the signaturing techniques.

When comparing sales in a group of stores where product is introduced against stores in the same market where a product is not introduced, there are always additional variables that cannot be isolated using the signatures technique. A procedure called belief zones (Figure 8) is then used for arranging curve findings in arrays to enable top management to make maximum use of its intuition (knowledge from years of experience) in projecting the future. The arrays are actually tables of data extracted from Squiggly^© curves, data that would nowhere else be available.

One of the most important needs of the manager is to look at totally unrelated patterns side by side, such sales of a product for one store with departmet sales for another store versus overall gross margins for another store. This can be done in seconds using the menu for over one million items shown in Figure 9. HKB has developed this tool based on a long term working knowledge of how people think when trying to decide about things.

Considerable work has been done using subjective data from management, such as a weekly recording of the estimated amount of sales lost by lack of inventory or lack of production capacity.

While not precise, these numbers can be used to show gross margin impact curves subtracted from interest gains on reduced inventory to address the inventory/sales readiness tug-of-war between manufacturing and marketing management.

It is very important to look at the same pictures a lot of different ways, back and forth. With a single click it is possible to switch from sales to gross margin dollars, from dollars to percents, from linear to logarithmic, form small to large, from lots of analysis to a little analysis, from lots of curves to a few curves. Based on experience to date we would expect profit impacts of squiggly^© at 5 to 10 percent of gross margin.

How the System is Installed

H K Ball and Associates has been working on the development of this system since 1979. Progress has been slowed because managers, until recently, have not been exposed to graphic representation of data as a primary means of making decisions. Most managers today think in terms of looking at tables of numbers when making decisions and are uncomfortable depending on graphs. We have therefore developed a training process which accelarates the learning process. While it varies from company to company, it would include the following stages.

Phase I (1-2 months)

An analysis of management decision processes would be made, and a hierarchy would be designed using existing management information systems. An information retrieval system would be designed installed,. Sample curves would be generated. This work would be reviewed with management to validate the hierarchy approach and the overall management information process concept.

Phase II (3-6 months)

Written reports would be submitted by HKB on an agreed periodic basis to include curves, signatures, corresponding data tables and analysis text. These would be used as an integral part of operations with a traditional day-to-day consulting relationship. Installation of the system hardware and software on the client site during the last third of the period. Training of management and other personnel in the use of the system. Working with management on a consulting basis in the implementation of day-to-day system driven decisions.

Phase IV (24 months or longer)

Monitoring of the process, hierarchy modification, day-to-day operating decisions, state-of-the-art system updates.