Full Article: Convergence Techniques

Here is a link to a full version .pdf of the Convergence Techniques article.

You can also find prior articles on my website.

Cheers... Rob

Convergence Techniques

Convergence Techniques

There are countless ways to build a convergence process, but I suggest they all follow the general model I presented. Whether a simple single pass with basic criteria, an artful process where the steps and stages may not even be clear or definable, or a scientific process with dozens of passes and a high degree of measurement and rigor, it’s important to be intentional and conscious during convergence. To better understand some of the options, it may be helpful to briefly discuss a few of the basic kinds of test and criteria that can be used. Here are a few suggestions for options in convergence. I have attempted to roughly sort them based on rigor and complexity with the most simple and easiest, but maybe least robust, methods first.

Simple audit. This may be the most basic form of reducing ideas and probably one to use early in the process rather than later. It’s a simple audit of all of the ideas and opportunities that emerged from the divergence process. This is done by a single individual, maybe the project leader or client lead. Using the initial design criteria or a simple set of requirements, a single pass or sort divides the ideas into two or three buckets. I like to create labels (something like awesome, average, and ridiculous work). This can save a lot of effort later when more costly or rigorous tests would better be run on smaller sets of ideas. It cuts off the low hanging fruit, moves forward the obvious choices, and can be a simple gut check of the process. I like to keep the process fluid and open to moving and idea from one bucket to another later.

Group consensus. In some situations, it may be better to give this task to a group, especially if the process is contentious in any way. I have used long-standing committees, decision making bodies, or specially appointed selection groups to perform this task. The process is much the same as the simple audit, but the results must be achieved through consensus or voting if necessary. The group process could be done in steps or passes at the list of ideas. Initially, the group needs to understand the task at hand and what success looks like. Then it must agree on the criteria and the number of buckets or sort piles. And finally, it can sort. The last step is endorsing or agreeing on what resulted. Larger groups could function in subgroups and compare the groups’ results as a twist. Simple audit and group consensus are often used methods.

Bi-dimensional comparisons. This sounds more complicated that it is, but a good technique to make choices from a mid-sized list of options is to plot them based on two key criteria. Often I like to use effort and impact along the X and Y axes of a grid then compare the quadrants. The axes can be rough scales from 1 to 10 or more specifically measured quantities if these measurements have been taken. It’s fun to label the quadrants too, and a low effort and high impact idea may pass through the filter most easily, high effort and low impact ideas may be eliminated, and the two remaining quadrants may require further thought or discussion. The method can be applied along with a simple audit or group consensus to help build a more powerful test. In reality, any two criteria can be put along the axes. Here are some combinations I have used in the past: cost-value, simplicity-impact, usability-functionality, function-form, outcomes-duration, or delight-cost.

Rating systems. A more rigorous testing method is to apply rating scales or systems. There is an entire discipline for testing and measurement and I won’t go into the methods here; know that you can learn about these if you need to. Simply put, a set of measures can be established for the key testing criteria and assessments of each idea are generated. For example, a 10 item rating scale might be developed using statements of agreement like I think this idea has a great chance of success and a group of individuals evaluate each idea. Basic statistics can be calculated on all of the items across all of the ideas and comparison methods can yield information about where cut points should be drawn, which ideas move forward in convergence and which ones do not. Different kinds of raters can also rate, like customers, users, managers, the public, and so on to add variety and dimensionally to the statistical analyses. As you can see, the rigor, precision, time, and cost of the testing is increasing.

Large-scale participatory events. This may be a less common method, but it is one that I’ve used to considerable success in my work with organizations. When considering significant strategies or organizational changes, getting broad input and buy-in during the convergence process can make the difference between acceptance and rejection of an idea. For example, I have staged day-long or multiple day events where a large number of early ideas are considered, built out, filtered, and rebuilt successive times with the result of reducing the number of choices in consideration and simultaneously adding detail to ideas that move forward. For example, working with one client over a multi-day event, we went from 25 opportunities that had some detail, down to 14 big ideas with a basic plan and cost structure for each, down to 7 key strategies with implementation plans. These stages can happen in a continuous event or be more separated in time. The key is that a large portion of (or all of) an organization or user group is there to witness and participate in the convergence.

Failure scenarios. For a convergence process where there is less empirical and historical data and evidence, we can use the scenario process to help give insight to which ideas should move forward and which ones should not. One method I like is called a failure scenario, and the converse success scenario. Basically, the first step is to develop stories about or situations in which ideas could fail or succeed and then sort the expected results of implementing an idea into one of many possible future outcomes. A final step is to review outcomes of the scenario sort to see if it makes sense. Scenario planning is sometimes more art than science, like comparing political or economic actions in large complicated systems. But there are examples where failure scenarios can be more quantitative such as materials testing, like a plastic, ceramic, or metal for application in a vehicle.

Voice of the customer. Consumer or user input is both a divergence and convergence technique. For divergence, we are looking for new ideas or builds on existing ideas from users. For convergence we are looking for what appeals, wows, or makes sense from the people who will be most impacted by the implementation of an idea. An example here is the software development or selection process when design choices can be selected, eliminated, or altered based on usability testing from a group of users. A second example might be consumer products like food or cosmetics, where the success of an item in a hyper competitive marketplace depends on subtle perceptions of a large number of consumers. The focus group is a specific example of this technique.

System constraint testing. Like a set of hurdles, technical or regulatory systems can set constraints of hopeful solutions. Sometimes, but not always, using these constraints are helpful ways to narrow down potential solutions, the downside is that these very same constraints can stifle innovation. Here the criteria can be written as requirements and tests can be done on possible solutions. If they do not meet requirements, they can be eliminated. An example here is the software development or selection process or regulatory changes in a governmental system.

Standardized tests. When we are dealing with a very large number of choices and when the decision process gets repeated over and over, we may be able to use standardized testing to help with convergence. Standardized tests assume that there is a distribution at play, a mathematical formula that repeats that can be applied and exploited for selection. For example, we know that most human traits follow what is called the normal distribution. Here there is an average and variation around that average that is commonly known. Height is an example, there is an average height for adult males, maybe around 5’10”, and a certain portion like two-thirds of adult males are between 5’6” and 6’4”. Very few adult males are shorter than 4’10” and taller than 6’10”. Tests can be constructed to determine a score for quality that follows a distribution and cut points can be developed that serve as filters (like the SAT for university admissions I mentioned earlier). The advantage here is that we can learn from history what traits and qualities have succeeded before. Standardized testing systems are complex and expensive filters.

Cost modeling. Typically a later stage convergence method, cost modeling allows for a fiscal analysis for competing ideas of costs and revenues as they are known at the time or could be estimated. Key outcomes like net revenue and return on investment can be calculated and compared across ideas. Ideas with costs that exceed resources may be eliminated for consideration. Cost modeling requires ideas that have a lot of detail built into them. This method may be a final filter in the process, say if three good options remain and the most revenue positive of the three would be selected for implementation. The method requires special analytical skills and significant data about production, markets, supply chains, and other financial metrics.

Business modeling. I offered up Osterwalder’s business model canvas as a divergence technique where intentional abstract models are built to explore and explain how value can be created through focused effort. The canvas allows you to explore both revenue and expenditure sides of the value propositions that organizations offer to customers. I have also used canvases in the convergence process. Perhaps less data driven than cost modeling, what the canvas may lack in precision it offers significant strategic breadth. I’ve found that groups like to build business models for ideas and the canvas is easy to work with. Early in the convergence process, many business models can be quickly generated and compared. Later in the process, as detail is added, more stringent criteria can be applied and the best value propositions selected for survival in convergence and lesser value propositions put aside.

Full feasibility study. A full feasibility study is a costly and complicated endeavor and it should be relegated to the last stages of the convergence process. Feasibility studies are also reserved for the most complicated ideas and strategies. I would hardly commission a feasibility study for my supermarket yogurt selection, but I would consider it before splitting a corporation’s key strategic business unit in two. The construction of feasibility studies is worth an article, or book, of its own, so I’ll leave you with little detail here. Here are a few pointers I’ve learned. Sometimes a feasibility study is completed on the one remaining best idea that resulted from the convergence process, a separate stage prior to implementation. While it may be possible to do a feasibility study quickly, in my experience it takes months or longer, so keep this in mind when planning the convergence process. And finally, there are many interchangeable components in studying feasibility, so look to keep the entire process as streamlined as you can while still getting the results you need.

Taking the prototyping route. Early in the convergence process, I like to challenge myself with this question: to prototype or not to prototype. The reason is that it changes how convergence goes. Several of the design models I presented are built to favor prototyping over a convergence-implementation pairing (see Ambrose/Harris, IDEA, and Plattner). I can go either way on this, but I like to be intentional about the decision and I do like to combine them. There is significant power in taking the final handful of choices and options from the convergence process and subject them through further development in the prototyping process. Here ideas are refined, developed, and adapted in further tests, but differently than in the convergence phase. Prototyping can be messy, require relentless iteration, and may combine any number of methods I’ve presented here for convergence. There are also some key differences – more on this next month.

In conclusion. There are endless testing protocols for new ideas, products, and solutions across industries. For example, the development and bringing to market drugs and pharmaceuticals requires lengthy trials and testing. Delivering a final bridge to cross a river to a city client has equally high stakes but different final criteria based on sound design from years of trial and error and computer modeling.

Most applications in organizational strategy do not get the same treatment that you find in product development or the pharma industries. They are more expeditionary. Regardless of the application, however, convergence takes a large number of choices and narrows it down until you left with something you can do something with, something that can be acted upon that brings you closer to meeting your original need.

The rigor you apply along way, the number of tests and their accuracy, the breadth and scope of testing, and the costs required depends on the stakes involved in the solution and the potential negative impacts of making bad choices. I hope you take some of my ideas and suggestions and apply them to your own convergence. Please stay in touch with any stories of success or learning.

---

Robert Brodnick, Ph.D. Founder, Brodnick Consulting Group robert@brodnick.consulting 530.798.4082

A Simple Convergence Model

A Simple Convergence Model

Early in the design stage, we expend considerable time and effort and learning about the desired outcomes of the entire design process. While we may not know the specifics of the destination, we do know a bit about where we think we would like to end up. One of the outputs of the design stage should be a thoughtful set of initial criteria to help guide the process and develop an image of what the overall outcomes may be like. Through successive discovery and divergence, the world of possibilities opens up, and as intended they should provide far more opportunities than we could or should pursue. Convergence allows us to make the best choices moving forward as resources are added to the refined design moving toward implementation.

Step 1: prepare. The pattern of convergence is reducing and combining ideas to build progressively valuable and implementable solutions. Early in convergence, the choices are easy. During the preparation stage, focus should be given initially to developing processes to help make easy early choices. It’s time to revise the initial criteria from the design brief or plan. Do these initial criteria make sense at this point? How should they be adjusted? Also, consider what has changed in the design process. What critical assumptions must be made going forward? How do they help refine choices? At least one decision pass or phase through criteria-testing-selection must be made. On the most simple end, it can be done in a group discussion; on the more complicated, a test could involve a large number of participants, data collection, and sophisticated analyses with statistical decision models. At either end of the spectrum, be intentional about each phase. In most case, I’ve found that multiple decision passes or phases need to be made to narrow options down far enough. When establishing these processes, ask questions about the number of stages, the rigor of each (i.e. data collection, metrics, inputs, decision making), and the inputs and outputs (from how many to how many). Sketching this out during preparation is helpful, if not required activity. One sketch might look like this.

Steps 2-4: criteria-test-filter-select-repeat. It can be hard to know exactly how much filtering and reduction can be done at any one pass. There are many standards and metrics for innovation and product development pipelines and several rules of thumb. My research suggests that these metrics and average rates vary considerably by industry with some convergence funnels or pipelines beginning with thousands of ideas and opportunities being narrowed down only a handful for implementation with a rate far less than 1%. At the less selective end of an innovation process we may have 100 or so good ideas that lead to 15 that could be implemented representing a 15% success rate for ideas.

When convergence process are repeated often or use precise tests, you can develop expectations for how the process will go. The following graphic shows where metrics or expected counts can be applied to a three pass convergence process:

1. How large is the entire convergence process? Count the ideas that resulted from the divergence process and exist at the beginning of convergence. This can help you understand the resource required for testing relative to the benefits you get through precision in the process.
   
2. How selective will the first filter be? Determine the percent reduction of the number of ideas or as a success rate. You can change the criteria for testing to move the rates up or down.
   
3. How many ideas or opportunities should go through the second set of tests and filters? In some cases you may want to really limit the number of ideas that pass through and keep the success rate low. For example, if testing is costly or may take a long time. In other cases, you may want more ideas to pass through to be sure too many early limitations don’t result in too few at the end.
   
4. How selective will the second filter be? Depending on the nature of the tests and processes used to understand the viability of an idea, the final test and filter may need to yield a polished idea ready for implementation. If the convergence process is focused on a product or service designed to go to market, some form of customer feedback should be included.
   
5. How many opportunities need to emerge from the entire convergence process? If you know that you need the three best options for final comparison prior to implementation, you can design the last stages of testing to deliver a specific number by changing criteria or the testing and filtering process.
   
6. Finally, a rate of reduction or success for the entire process can be determined by comparing the number of ideas at the end to the number at the beginning.

1 # of opportunities at the beginning of convergence

2 % reduction by the first filter

3 # of opportunities in the second pass

4 % reduction by the second filter

5 # of opportunities at the end

6 % reduction by the entire process

Step 5: conclude. Whether a science or art, understanding how ideas flow through the convergence funnel is important. The convergence stage ends when you have either viable early-stage prototypes or a design even more evolved and ready to implement. Sometimes things can go awry and convergence can yield nothing viable or ideas could be reduced too far. In this case, we may go back and forth between divergence and convergence if none of the choices appear to be working or back to the tests and filters and modify the criteria to produce more at the end. When convergence is followed by prototyping, as I suggest in my 6-stage design model, the final filter should be tuned to ensure an early stage prototype emerges from the process, which can be further refined through prototyping (more on this in the next article).

Convergence and Selection Techniques

Convergence and Selection Techniques

The partner process to ideation and divergence is convergence and selection. After having generated a large number of ideas, few try to implement them all. It is a good idea to pick the best ones and consider the implications of implementation. While it might be possible to look at the list of ideas from a single brainstorming session and pick the one or two you like the best, I suggest that it’s bad practice and recommend are more robust set of techniques to make choices, especially when the stakes are high. Over the following pages I will explain and define convergence generally, give a broadly-applicable model for making choices following ideation, and outline a handful (or more) of different techniques.

Convergence Defined

Convergence means coming together in the most general sense, yet quite a range of definitions exist with more specific disciplinary applications in computing theory, economics, literature, mathematics, and popular culture. Most online definitions and dictionaries are useless here as “the act of converging” is the most popular definition (who lets them get away with that anyway?). We do better with the word converge which means to gradually change so as to become similar or develop something in common. While that helps a little, it’s not fully there.

What about selection – definitions yield “the action or fact of carefully choosing someone or something as being the best or most suitable” – more on target. Adding the terms evaluative (making informed decisions and preparing for action) and critical thinking (making clear, reasoned judgements to reach a conclusion) to the mix now gives a good footing. When I compare my research on divergence versus convergence, it seems we humans, at least in our organizations over the last 100 years, have a tendency toward analyzing and judging more so than creating and wandering in thought.

Well, I do like the graphic that shows a coming together, a narrowing options, and the number of possibilities that are in play being reduced from the many to the few that are the best choices. To focus on four stages in the design process for the purposes of strategy crafting, the convergence stage becomes more obvious in this context.

Convergent, or selective, thinking has several characteristics and all of the methods have the outcome of decreasing the number of options being considered. During selection processes, individuals seek to select only the best ideas, they test dreams against reality, and apply judgement. Convergence and selection techniques uses criteria and filters and is more practical than the dreamily divergent stage. Let’s look at a bit of the history of convergence and selection techniques and then what’s happened more recently.

Of tests and filters. Given the pervasiveness of convergent thinking and the broad distribution across many fields and applications, a fair history would encompass a book in itself. There are innumerable methods and techniques for evaluation, but fewer for selection, the narrowing down from the many to the few. Without a full discourse, I have identified the key characteristics of the process – basically it boils down to sequential tests and filters guided by useful criteria.

An understanding of the concept of a criterion is useful at this point. Criteria (plural) or criterion, is basically a standard on which a decision is based. Definitions like “something that is used as a reason for making a judgment or decision” and “a standard, rule, or test on which a judgment or decision can be based” are common. In terms of the design process, criteria help us articulate our assumptions about the process, draw boundaries around possible results, and give an objective approach to eliminating choices that are not satisfactory. So, criteria are standards by which some decision can be made. For example, a simple criterion may be the height requirement on the roller coaster, “if you’re not as high as this (48 inches) you can’t ride”. Tests and criteria pair well and can combine easily to make filters. Another example here is the SAT, “if your score is above 1300, welcome to our university” and “if not, perhaps start at the local community college”. Here a general test of aptitude and prior learning is used as a filter for potential students based on research of who has succeeded in the past. The combination of tests, filters, and criteria forms a core foundation for the convergence process.

The result of applying criteria and narrowing down your choices or options establishes a filter. Most broadly, a filter is as something that has the effect of holding back elements or modifying the appearance of something. Other definitions like “move slowly or in small quantities or numbers through something or in a specified direction” and “a device that prevents some kinds of light, sound, electronic noises, etc., from passing through” are found online. Criteria and filters are often applied and executed using a test, or a procedure intended to establish the quality, performance, or reliability of something, especially before it is taken into widespread use. This definition fits our convergence stage of the design process well.

Making a choice can be a simple matter. When presented with the perhaps 200 or so cups of yogurt in the grocery store, I seem to be able to navigate the convergence process without too much thinking or delay. I have three criteria: taste, flavor, and price. I know that certain brands meet my taste expectations and others do not. I like some flavors better than others and I make a final decision on price. A really good sale on my second favorite brand may sway me to buy it. This example is a simple, single-filter approach with three criteria. I have a certain number of options and I narrow down my choices to a final selection in a single process.

Perhaps the most commonly used approach is a single test and single filter. For example, a collection of well developed ideas are put forth to evaluation and the results measured and ranked. Often, it’s the best performing model that wins, get a few tweaks to improve glaring weaknesses, then readied for implementation. While there may be situations where this works just fine, my experience suggests we need at least two or three rounds of filtering (and often idea combining) to adequately make the necessary refinements. There are several disciplines where this was put into practice ahead of others. Sometimes the single filter approach does not even have criteria developed ahead of time, you just take the best of the bunch. More sophisticated selection processes (like the SAT example above) use pre-defined criteria.

For more important decisions, I may take two or three passes, like purchasing a car or a home. There will likely be a larger number of criteria and they will likely be applied in more than one stage. Perhaps some online research based on my needs, test driving the best options, talking with others about their experiences over time with the car, and finally the lovely negotiations with the sales team at the dealership.

We see early evidence of the test-filter approach arising out of engineering, especially chemical engineering, in the 1940s in the form of the new product development cycle, or phase-gate methodology. The phase were scoping, definition, development, testing, and launch and each had a corresponding gate. The alternating approach from phase to gate was sometimes expanded to seven or eight or reduced to three of four depending on the needs of the situation. Generally rigor increased as the product got closer to launch.

Interestingly, when I began my study of these kinds of processes, I found them also referred to as stage-gate. The interesting point is that in my recent research, the stage-gate name has been trademarked and the process made much more specific. Some innovation and product development experts still use stage-gate name but I imagine this will slowly fall from favor since the name is now “owned”. In any case the seven stages and gates can be loosely found as: idea generation – screening – development and testing – feasibility and analysis – market testing – implementation – commercialization.

What’s happened lately? To be fair, the history of the human (and other) species is probably an organic example of prototyping, unknowingly at first, then more intentionally. From the development of the first hand tools, to the evolution of shelter, to the refinement of my much adored Grenache wines or Bechamel sauce, the iterative nature of “what works” through prototyping is a fundamental as evolution itself. What has gotten better are the specific techniques, perspectives, and training that goes along with it.

The last 20 years have seen a real growth of design and innovation techniques and with it a refinement of the convergence and selection processes. I will present again the six design models and highlight the areas of each model that apply convergence.

General Model	Ambrose/Harris	IDEO	Ogilvie	Plattner	Brodnick
define the problem	define	discovery	what is?	understand	collaborative design & direction setting
	research	interpretation		observe	discovery, research, & assessment
explore, create, refine			what if?	point of view
	ideate			ideate	divergent thinking, & ideation
		ideation
					convergent thinking, filtering, & selecting
			what wows?
	prototype	experimentation		prototype	prototyping & piloting
implement preferred solution	select
	implement	evolution	what works?	test	implementation, tracking, & adjusting course
	learn

Across these models, the lines between convergence and prototyping blurs. Definition and ideation clearly following the preparation and discovery stages, but convergence, prototyping, and implementation is more fuzzy. Some, notably IDEO’s model move from ideation to experimentation and evolution with a less clear convergence process. Other’s like Ambrose & Harris and Plattner have a clear testing and learning stage at the end, much like IDEO’s evolution. My guess is that the models have been shaped by the experiences and industries in which they were developed.

In some instances, the refinement process could be in the hands of an artist; think about the shaping of glass and the convergence on the final form of the piece from the many options that emerged from the heating and blowing process. Or internally as the designer creates and recreates the new font or logo. While the designs may be tested more publically later, the first filters are those of the trained artist. In more open design, opinion and preference shape early choices, like in IDEO’s empathy forward approach or Ogilvie’s “wow” factor. Other models use prototyping as the primary convergence method. I like to suggest we have all of these possibilities as tools in our toolbox and we can pick and choose what’s best for our application. In strategy and organization change projects, I’ve found that blending intuition with opinion and art with metrics is helpful if not required. For the big change projects, we need to pass through financial and resource feasibility in any case.

Allow me to continue with offering a simple convergence model and follow up with a handful of techniques you can consider and try. In the next section of the article, I will focus purely on convergence and selection and offer a generalized process that could actually be used in conjunction with any of the models – five clear steps with the option of repeating until the desired outcome appears.

Full Article: Ideation and Divergence Techniques

Here is a link to a full version .pdf of the Ideation and Divergence Techniques article.

You can also find prior articles on my website.

Cheers... Rob