Using analytics to validate startup ideas for products?

Today we look at using predictive analytics to validate startup ideas. This is part of our continuing Ask Acculation series; scroll down to find out how to get your analytics or start-up questions answered here. A version of this post was originally published elsewhere.

“How do I validate my B2B idea? I have considered joining an accelerator, building a website prototype, or asking entrepreneur friends, but are there other ways?

This is essentially one of the main goals of predictive analytics (what we do) when applied to business data: be able to evaluate your idea using fast, inexpensive tests, than transform that data to how the market might behave under “real” conditions.

First off, the premise of the question is intriguing. We’ve seen social media memes that have gone viral “there is no B2B or B2C. There are only people” superimposed over some sort of feel-good graphic. And the memes have their staunch defends. If you’re dealing with a B2B (Business-2-Business) product, you need to sell to people inside a company, and you’re selling to people when you have a B2C (Business-2-Consumer) product.

This is a fine sentiment. If your photo with this catchphrase gets heavily retweeted, all the more power to you. But, we’re an analytics company, and this is blog is about making better decisions through data. Viral catch-phrases aside, there are significant differences in the way B2B and B2C products behave in the market, and this shows up in the data.

These differences arise because B2B and B2C customers make decisions in a different way. Businesses (especially the more successful ones) try to make purchasing decisions by committee, or at least by spreadsheet. (They might be trying to make data-driven decisions.) Consumers, on the other hand, often don’t have time to analyze or over-analyze their purchases, so might just grab something on a shelf because it looks cool and they are in a hurry.

These different behaviors clearly show up in the datasets. The clearest way to convince yourself of this is the “Bass Diffusion Model” by marketing research professor Frank Bass. It’s quite good at modeling market saturation effects, and it’s used to figure out when to stop selling a product because everyone in your target market has essentially already bought it. Bass diffusion constants for many products have been published and are well-known. The model itself predicts B2B and B2C products will have very different coefficients, and the published coefficients confirm this. Any readers in a marketing department at a company that sells both B2B and B2C products can easily confirm that B2B and B2C products really do behave different in marketplace by fitting products to the Bass curves. The two coefficients, in turn, are inspired from Bass’ fundament insight that businesses and consumers really do make purchase decisions in a different way. So, the next time you see the “There is no B2B or B2C” viral meme, you have the tools to know better.

Although B2B products do behave differently than B2C products, an analytics expert will know how to tweak most of their models so that they’re applicable to both. Once you accept that, many of the same lean startup techniques used to validate B2C products can also be used to validate B2B products.

The original question pointed out the example of a website prototype as a way of evaluating ideas (presumably by looking at website traffic and marketing response curves).

Lean start-up techniques like building website prototypes generally also work well for B2B businesses (although you may find it harder to recruit test users due to it being harder to market the website to businesses).

Here are some other techniques that can potentially be used together with analytics techniques:

1.  Test marketing is another common technique that works for both B2B and B2C products. An expert in predictive marketing analytics can then sometimes take the results of the test marketing and transform that data to get a rough sense of how the product would fair under a much larger marketing campaign carried out by a professional ad agency. (This transformation takes into account things like marketing spend, type of product or industry, ad creative quality, and the touchpoints used in the original test campaign versus the final campaign.) This can then give you a rough quantitative sense of how the market perceives your product’s quality, which can then be used to guide further planning and investments. And, yes, the marketing expert will known how to tweak their analysis to compensate for the B2B nature of the product (or even how to convert data from a B2C test campaign into how a B2B final product might fare). Market saturation models (like the Bass model we already discussed) and marketing strategies and analysis are one of the few ways B2B differs from B2C. Beyond that, however, there are few other systemic differences between B2B and B2C.
2. Some market research companies use techniques specific to B2C. However, there is no reason why classic market research techniques such as surveys cannot apply equally well to B2B as they do to B2C. Of course, if your B2B survey can only accept responses from CEOs, it may be more expensive to administer than a B2C survey that can accept responses from anyone. A survey is a more scientific extension of the questioner’s “ask entrepreneurial friends” approach
3. The original question mentioned incubators. If your product idea or company are accepted to one, these also will help you evaluate your business concept, often in a data-directed way. Incubators often have their own individual formulas for evaluating concepts. This may include combining classic market research from #2 with estimates of market size and competitive analysis. They may combine these data points with expert evaluation of the idea by entrepreneurial panels.
4. Although crowdsource platforms like quirky tend to focus on B2C products, there is no reason why you cannot adopt similar techniques yourself for B2B products. For example, you can post your ideas to social media and seek crowdsourced feedback. There are social media sites that cater to the start-up community, so you could try posting your idea to these (Obviously, this may also attract competitors and copycats. You can consider this a cost of doing business in exchange for getting valuable insights off these sites.)
5. There are firms that have analytics products for specific B2B verticals. For example, if your B2B idea involves, say, licensing out an existing patent for example, there are websites and software that can attempt to (roughly) quantify its value. The advantage of using software (as opposed to asking individual human experts) is that predictive analytics methodologies are supposed to be consistent and unbiased, so that you can, in theory, estimate the error in its predictions by looking at the software’s past track record in making similar predictions.Of course, buyer beware. There are plenty of sites that promise to gauge the value of your Twitter, for example. We’ve looked at some of these, and many of these sites seem to produce highly dubious numbers if not complete nonsense. Another issue is that it easy to make predictions in “hindsight”, something you need to make sure is not happening when you are establishing the track-record for one of these analytics engines.Loyal readers of our blog will recall our earlier article on predictive analytics models for Hollywood movie script or a pop songs. So, if you happen to be trying to sell one of these to a major corporation, you already have a specialized analytics engine for your specific product. The same is true for many other industry verticals.

Have a question about start-ups, business or analytics? Send your questions to [email protected] or tweet them to us at @Acculation. We’ll respond to some of the submissions here in this column.

Related posts:

Social Progress Index and Big Data Analytics: government by computer? Big Data Analytics: Articles, Movies, Songs Robo-written by Computer? Watson from IBM: Why semantic text tech helps analytics Data science and limits on human lifespan (or … vampire heart rates)