Startup 101

Learn the Fun(nel)damentals to Boost Your Business

Every product is sold through a leaky funnel. Learn how that funnel works for your company and how to plug the leaks.

• By Patrick Foley
• 06/06/2012

Unless you work for a non-profit, your company sells something; and if your company sells something, it has a sales funnel. The concept of a sales funnel is simple: of all the leads your company pursues, only a fraction of them turn into prospects. Of all those prospects, only a fraction of them turn into customers. If you look at this visually, it looks like a funnel.

Online businesses often call this a conversion funnel, because you're trying to convert visitors to your Web site into sales, but you can't win them all. The places you lose sales are called leaks. If the funnel didn't have leaks, none of the leads would fall out, and you'd convert 100 percent (that would just look like a tube. Nobody talks about a conversion tube). Plugging leaks in your funnel means getting more sales for the same number of visitors.

Let's say you come up with a great design for a "Visual Studio Roars" t-shirt with a picture of a menacing kitten on it. Disregarding any trademark or copyright concerns for the moment, you put up a Web site to sell your masterpiece for $20 a pop. There are just four pages in your site: a product page (same as the home page), an order page (shopping cart), a payment page (perhaps PayPal or Amazon Checkout), and a confirmation page.

You launch your site quickly (before you make the shirts, in fact, since you're all about lean) and tell all your Twitter followers (linking to the product page). Looking at your stats after a couple of days, you discover that 100 people visited your product page. Of those, 10 clicked order. Of those 10, only one actually completed payment, resulting in a 1 percent conversion rate (which is pretty normal).

Plugging the Funnel Leaks
If you want to sell more shirts, you need to get more people into the top of the funnel (more people visiting your page) or try to plug leaks in the middle of the funnel (get more people to order and more of those people to complete payment).

Since you've already exhausted your Twitter followers, you're likely going to have to pay for new visitors (for example, by buying ads), but if you're going to spend that money, you want to make sure that you're getting your money's worth by plugging the obvious leaks. Without changing anything else about the product you're selling, plugging leaks in your funnel increases sales and improves your bottom line. This is true whether you're selling t-shirts or software.

You make changes to your site based on tips you keep reading, like having a clear "call to action" with a nice, big button. Instead of describing the shirt itself, you describe how cool your customers will feel wearing it to their next user group meeting (always focus on the benefits that your customers will experience from using your product more than the features of the product), and you eliminate about half the words you originally had on your site. These steps tend to have a positive effect on your conversion rate.

So you budget $200 for ads, which means you have to sell 10 more shirts just to pay for the ads. You make a decent ad buy, and that $200 brings you 1,000 visitors. However, thanks to your site improvements, now 20 out of every 100 visitors visit the order page, and two of those complete the purchase. That's a 2 percent conversion rate, netting you $400 in sales on 20 orders. Since the t-shirts cost you $5 to make, you just netted $100 in profit for your time and effort.

A/B Testing
This looks promising, so you decide to funnel all your profits back into the business and invest some new money as well. You decide to spend $2,000 to get 10,000 visitors, but at this point, you really want to make sure you're not losing sales, so you decide to start A/B testing (also called split testing).

Most A/B tests share a similar process:

• Make a hypothesis about a change that will improve your conversion rate
• Figure out how to implement the change for a percentage of your visitors
• Measure the difference in conversion rate between the old site and the new site
• If the new change is better, roll it out to the whole site
• Rinse and repeat

You don't even have to understand why a change works or doesn't work; you're simply using data to prove what actually works. It's an optimization process. And according to Noah Kagan, you probably shouldn't start A/B testing at all until you have a sample size of about 10,000. Until you reach that point, focus on measuring everything so that you have a basis for comparison when you do start A/B testing.

For a real-world example of optimizing a software sales funnel, read what Patrick McKenzie did for Fog Creek. Patrick is a genius about this stuff. If you want to get good at conversion optimization, you should read his blog and do what he says. My job is just to get you thinking about it.

Conversion Optimization
Conversion optimization has become an art, science and business in its own right. Brian Massey calls himself the "conversion scientist" (complete with lab coat), which is a great way to think about the process: you're running experiments to improve your sales. Every experiment involves a hypothesis plus a measurement and should result in a go/no-go decision on a change. This is what good marketing professionals do. I've known software engineers to be dismissive of marketing people, but good ones are scientists in their own right.

Companies like Kiss Metrics, HubSpot and Aprimo sell millions of dollars of software to manage and facilitate the process of optimization for marketing professionals. Anything that has a direct impact on every company's bottom line is going to be a big business itself, but you don't have to spend big bucks to start doing this. You can use an open-source framework like Fairly Certain or Split A/B Testing or even code things manually. The concepts aren't hard, but the work itself must be diligent; you have to make sure you're actually measuring what you test.

I've offered a very simplified summary of optimization here (you have to start somewhere!), but as you probably can guess, things get more and more complicated. Returning to the t-shirt example, let's say you get greater sales by promising that your shirt will make customers appear three inches taller. It was just a joke, but after a couple of weeks, you discover that one in four people who purchase your shirt are asking for their money back due to lack of tallness. Clearly, you have to consider your return rate to determine the overall success of changes you make to your site. To do that, you have to track which customers used which versions of your site. Customers using a specific version of your site (and/or who came from a specific lead source) are called a "cohort."

Cohort Analysis
Cohort analysis is really important for subscription software businesses, because the line between sales and the use of your product is somewhat blurred. Every month you have to continue to provide value to your customers or they might leave; you're continuously "selling" to them. Customer retention is affected by how they entered the sales process. Different cohorts might also respond differently to changes in functionality or pricing policies. All those combinations are important to track.

Your funnel might be longer than you think as well. Rob Walling suggests that the first goal of your Web site should probably be to get customer email addresses, not ask for the sale (you ask for the sale in an email campaign). Say a customer doesn't like your first t-shirt, but maybe they'll like a future t-shirt. If you hope to build a t-shirt empire, spend more energy engaging site visitors and getting their email addresses so that you have permission to market other shirts to them in the future. Watch Rob's presentation from Business of Software 2010 for details on this technique.

You might notice that some of the people mentioned here are stud programmers in addition to being great marketers. Someone who can combine coding and marketing is called a "growth hacker," and according to Dan Martell, the discipline is about to explode. Read Andrew Chen's blog to learn how to be a growth hacker yourself.