3 Ways to Accelerate Your Learning Process

 

So you have a few exciting new ideas that could really create some breakout growth. Congratulations! But which one(s) is (are) really worth betting on? How can you get to market fast enough to not only beat the competition but drive a wedge into the market? And can you keep these winning ideas under wraps until you're ready to announce them with great fanfare to garner millions in free public relations?

And you thought coming up with the great ideas was the hard part!

Concept testing — the winnowing of a broad list of ideas to a single homerun concept with life and the pledge of profitability — traditionally has been used to reduce the uncertainty surrounding the risk/reward decision underlying investment in new products or programs. Unfortunately, it also has had a long cycle time. Often single attributes of a product offer or messaging strategy undergo expensive and time-consuming trials through endless focus groups, yawning quantitative research, and (not so) controlled market testing. All the while, CEOs drum their fingers on boardroom tables, concerned that the "big idea" will not produce a return on the dollars or time invested, let alone any competitive advantage.

In just the past few years, technology and mathematical science have combined to offer us quicker, quieter (in the stealth sense), less costly, and more accurate ways to test new concepts and explore dozens or even hundreds of discrete value propositions simultaneously. Better, faster, cheaper. Choose all three.

Thanks to e-mail and the Web, construction and delivery of marketing research studies have experienced hyperinnovation. Infinitely more complex surveys don't require infinitely longer interviews, dramatically expanded sample sizes, or months and months of cross-tabbing. Instead, prospective customers see concepts specifically tailored for their demographic, life stage, or behavioral appeal.

Today we can re-engineer our innovation processes without neglecting precious information-gathering steps. Enhanced modeling and analysis in this pretest phase mean in-market tests will more fully explore and deliver on end-user expectations, not just throw something up against the wall to see if it sticks.

Here are just a few of the new techniques to deliver on the promise:

1. Testing Multiple Variables 
If you've been doing marketing for 10 years or more, chances are you have followed a "test and control" method of testing a single variable element in your marketing mix or message countless times, holding all of the other elements as constants in a control group or market. This approach, while sound, is today a quaint relic of the era of slow computing.

In the complex world we live in, the opportunities and risks are often buried deep within a cocktail of variables well shaken. And as they did when you were in your young bar-hopping days, these cocktails can poison your marketing body and kill organizational brain cells, sending you gingerly groping for the analgesic effect of simple solutions.

Unfortunately, this isolated-variable approach, while soothing, makes it practically impossible to determine if the right variable is being evaluated. Testing one-by-one, you might never find the one that's really draining resources or the one that stands to deliver the biggest gains. If you have 10 factors that might influence an outcome, either alone or in combination with others, you have to run 1,024 tests. With 20 variables, there are more than 1 million scenarios to cover.

 

Enter science and experimental design, in which many variables are altered at once. Using multivariable testing, you can see how different changes (not a single change) in your "cocktail" affect outcomes. Working backwards from the preferred outcomes, you recognize which ingredients need to be adjusted or even discarded to strengthen your offering or mend your potential marketing ills before they are exposed through a product or service launch. Almost invariably, says Kieron Dey, technical director at Knoxville, Tenn.-based QualPro Inc., you will find that as few as 25% of the variables help propel your marketing campaign, product, or service to success. Of the rest, 55% make no difference, and about 20% make things worse.

Now you can exercise your drive for simplicity and hone your focus on just those elements that give you enviable innovation health.

You can also apply optimization here to find the ideal innovation recipe. Like multivariable testing, optimization opens your eyes to components of your marketing efforts you previously thought unimportant. And optimization knows few bounds. You can set and reset parameters, change the test variables, and explore the relative promise of any number of approaches.

To extract new life from a campaign already in market, you can run multivariable testing on data already in your database, slice it and dice it a million different ways, and come up with new ideas to re-engage customers. Multivariable testing finds previously undiscovered triggers to process improvement in data.

A quick scan of historical data for patterns gets the ball rolling. That part of the process determines the variables to take into multivariable testing. Now you can merge internal variables, such as budgets, manpower, and technology resources, and external variables, such as regulation, supply chains, and market economics.

Multivariable testing takes the finger-crossing guesswork out of the product or service innovation process and replaces it with efficient experimentation and fact-based problem solving. It doesn't need to be run with supercomputing power or tricked-out technology, but it will require an experienced scientific eye on the experimental design.

2. Modeling Discrete Choices
When your head is full of ideas and you don't know which to cling to and which to shake loose, you can test them all — many more than you had time or money to test using traditional methods — through Web-based discrete choice work. Choice options modeling helps you sort through a myriad of options, from product features to distribution channels to price elasticity — all in a single Web-based survey. This fairly recent innovation can help narrow a broad spectrum of options to a more useable subset, swelling with potential for success.

In discrete choice modeling, consumers are asked to make nested, multiple-level decisions based on an immediate reaction to a set of options (which of the following savings plans — A, B, or C — would you choose to save for your child's college education?). This survey method differs from conjoint analysis because survey participants aren't asked which they prefer or how they'd rank the options. Instead, they are asked to react, to choose, and to move on in a deliberate, sequential questioning that always presents the choice in the context of the expected behavior change.

The method was developed by Daniel McFadden, who won a Nobel Prize for his work, on the theory that consumers make everyday decisions without analysis but often almost instinctively, choosing from a host of competing products or services.

The Web makes it easy to not only get a broad sample base through e-mail lists, but to test hundreds of different options in a compressed time period. Just a handful of those hundreds can statistically uncover strengths or weaknesses. (See John Cripps's article Accelerating Your Best Ideas to Market for a more detailed description of this approach.)

 

As with multivariable testing, the options that emerge victorious through discrete choice then proceed to an optimization phase. Here, you find the ideal attribute bundle can be derived given some hard-and-fast parameters like production limits — we can't make any more than 10,000 widgets in a year's time — and simulated to optimize budget efficiency, profitability, or time to launch. The outcomes achieve order among the other variables like marketing expense, price point, and how many widgets consumers will buy at different price points and determine how big a product line should be, how it should be marketed, and what riches it will secure for you.

The power of discrete options modeling lies in its ability to help sort quickly and inexpensively through a laundry list of potential solutions to find the short list worthy of further exploration. Used properly, it speeds the innovation cycle dramatically and reserves precious resources for targeted applications where they are likely to do the most good.

3. Testing Individual Executions 
So now that you've isolated the features/functions of the ideal new product or service, you can assemble an online panel once again, this time to present a range of visual, audio, and text communications to the target prospects and find the best combination of message and execution. You can even speed research and enhance learning by showing consumers a simulated new product or campaign and soliciting feedback about its many lifelike attributes.

Say, for instance, you want to test four new brand positions for an existing product in a competitive market, crowded by 11 similar offerings. You can design four different branding booklets, presenting one to each of four randomly selected, equivalent customer groups. Pages 1-11 of the 12 pages in each booklet carry the same text and graphics, the same messaging, representing your competitors' campaigns. Page 12 introduces a unique, new brand position — your own.

Reactions to the positions will generate statistically significant differences, answering not so much the why of consumer choice but cutting to the quick and telling you what will make your customers buy your product. (Relationship managers may balk, but this what matters more than the why because in accelerating the move to market, we focus more on predicting behaviors than on understanding them. The why matters only when the what eludes your creative team, and the techniques presented here aim to determine the what through reliable math and science.)

This technique, known as partial differential testing, does what some of us have long claimed is impossible. It disaggregates an aggregated marketing effort, setting your test executions against competitors' executions, medium by medium. And in some cases, this process unseats in-market testing altogether because it delivers a clear winner. An in-market test that might consume four to six months of the precious innovation lifecycle now takes but two or three nights online. In some categories, for some products (wireless, for example), that's an entire innovation cycle gained.

This same approach can be used to test your ad agency's three different proposed television commercials online, as video clips. Give each of three broadband-enabled groups two competitor commercials to view and one of your new executions. Ask not which ad is funniest or most heartwarming but solicit a preference based on likelihood to purchase.

You can count on partial differential testing to deliver sound direction on nearly any given critical issue — any critical issue, that is, but price. The technique does not work terribly well with suggesting appropriate pricing. But you can fill this shortcoming with, yes, optimization again. Optimization, like each of the other recommendations here, will save you time, reduce your risk — its results are more reliable than those of more heavily used, old-school tools — and get to work with ease.

Tying It All Together
How might these tools work together in an actual new product development process? Let's take a look at a hypothetical example.

A skateboard company known for its custom, high-end, professional-quality boards introduced a product intended for the mass consumer market a year ago. The board's unique feature, sparking wheels, solidified sales in the months after the product launch into the winter holiday season.

 

But with the return of warm weather, the company's expectations of additional sales did not materialize. Retailers shared reports of kids claiming the product didn't work as advertised, and without the buzz of viral marketing in the boarding community, the goal of increasing penetration into the general consumer marketplace, attracting especially beginners to the sport and seeing them through their maturing interest, looked less achievable with every passing month.

Looking at the elements of the sparking-wheel board and the marketing surrounding it, product tests proved the sparking wheels worked consistently. Television ads showed boards in action, wheels sparking. But in the multivariable testing phase, where data on users was examined for correlation to satisfaction, the problems surfaced.

Commercials for the sparking-wheel board showed not the product's target users of 6- to 8-year-olds, but teens with considerable skill and significantly more weight than a younger audience. Turns out beginning boarders, especially 40-pound beginners, had neither the weight nor the speed necessary to get the wheels to spark. The product wasn't engineered to match the market, and the ad campaign was highlighting the problem. What's more, the complaints by parents to specialty retailers didn't win any new channels, but lost some. Worse, boys who received the boards and couldn't get them to spark lost interest in the sport and future board purchases.

The company took these findings to families with young boys with online discrete choice modeling, offering a similar board (with a spark trigger on the underside of the board that rubbed the wheels when in motion), modified for boys 30 to 60 pounds. They presented different product configurations (colors, sizes, shapes, etc.) at a variety of price points, with alternative promotional packages (knee pads and elbow pads, coupons for helmets, etc.), and availability through either the existing channels or expanded channels (including the Web, mass merchandisers, sporting goods stores, specialty toy shops, and national toy retailers).

 

With the clear high-potential winners coming through the discrete choice work, the company took to the Internet again, this time with partial differential testing. One new campaign featuring boys, fully suited in protective pads, riding skateboards in a residential neighborhood, was tested against another featuring a company spokesman and skateboarding pro teaching a skateboarding safety course in which boys on the sparking-wheel boards stood out in performance and safety. Each was alternatively produced alone or as an ad co-op'd with new channel partners.

The entire concept testing occurred quickly enough to get new product back on store shelves in time for the next holiday season, giving the company (and the retailers) greater confidence and another shot at a critical growth market that they might otherwise have missed.