Skip to content
Mike Lieberman, CEO and Chief Revenue ScientistFri, Mar 3, 2017 11 min read

How To Conduct An Inbound Marketing Scientific Experiment

Knowing How To Test Inbound Marketing Tactics Produces Better Results

Inbound marketing experimentsIf you’re doing marketing correctly, all of your conversations should have an element of data, metrics, numbers and quantitative analysis associated with them. There are no more questions like, “Did it work? Is it working?” You know, beyond a shadow of a doubt. The numbers show you. The science of inbound marketing ensures you know if your marketing is either working as expected, working but less than expected or not working at all.

But knowing what’s working is just the first step. The real secret to inbound marketing results, demand generation and frankly any type of marketing is being able to make improvements over time. The methodology for making constructive and productive improvements is not random adjustments in the hope that the numbers improve but rather a scientific method of testing.

Here’s how we conduct scientific experiments for us and for our clients.

Selection Based On Observation

There’s a ton going on when you’re looking at sales and marketing activities. Where do you start? This usually comes down to goals and priorities. What’s most important to you and your company? Is getting more people to your website the most important goal? Is it getting people already coming to your website to convert into leads? Or is it turning leads into new customers?

For this article, let’s agree that getting visitors to convert into leads is the most important objective. Now you drill into this area to find smaller data points. For example, what website pages are getting the most visitors? This is a great place to look for improved conversion performance. Once you identify those pages, look at those pages. What is the visitor behavior on those pages? Perhaps select the most-visited page. Are people scrolling all the way down? What are they clicking on? What type of page is this? Is it a page at the top of the funnel, middle of the funnel or bottom of the funnel? Your site has these types of pages, right?

What offers are on this page? Do you have one, two or more? All of this investigating is going to help you create the hypothesis that you’ll end up testing.

Hypothesis Setting

Speaking of hypothesis, you’re ready for that now. You found the page and you want to run an experiment on it. You need a hypothesis to test. For example, if we put a different offer on this page, we can increase the conversion rate. Or if we add an additional offer, we can increase the conversion rate. Or if we move the offer, we can increase the conversion rate. Or if we rename the same offer, we can increase the conversion rate. Get it?

Pick one of these and establish what you think is going to happen if you make the change you’re planning on applying. This becomes your hypothesis. For example, the page is converting 20% of all visitors into leads, but I’m not sure the offer is perfect for this page. So, if we swap out one offer for a better one, we expect the conversion rate to improve from 20% to 30% over the next two weeks.   

Improvement Goals

Inbound marketing data dashboardSince every aspect of marketing is performance based and results driven, you have to use data and improvement goals to marry up with your hypothesis. It’s not good enough to say that you expect to increase conversion rates; you have to attempt to hit a certain improvement objective.

Your improvement goals can span across a number of experiments. You might have a goal to increase the conversion rate from 20% to 30%, but it might take you five experiments to accomplish that. Improving by 2% each time is reasonable. Once you get to a new high, that becomes the benchmark by which you’re trying to improve. Some of your experiments will produce lower results. In that case, go back to the benchmark configuration and start iterating again.

Variable Adjustments

If you’re like most marketers and some scientists, you might be tempted to change more than one element or variable on each experiment. Don’t! That might seem like some smart shortcutting, but it’s going to seriously limit what you learn. Going back to our example, if you change an offer and change a form and the conversion rate increases, which change contributed to the increase? You don’t know.

Typically, I like to build the chronology of the tests I want to run. That keeps me from changing more than one variable per test. I’ll line up a series of experiments or variables and then just start working them down, one at a time, keeping an eye on the performance data. Its important to keep track of notes on what you’re doing. It can get complex if you have a couple of experiments running at the same time and then you have a series of variable adjustments you want to make over time.

Timing

A lot of people ask me how long their experiments need to run for before they can evaluate the data. Great question, but its usually more about the data than the length of time. Once you have enough data points to make a good decision on the test, you can end the test and move on to the next test.

Let’s look at our page conversion example. If that page only had 10 visitors, I’d let the test continue. If it had 100 visitors, I might consider stopping it, but if it had a few hundred or even 1,000 visitors, that would probably be plenty of data to feel comfortable that the sample size was large enough to be telling.

There are more scientific and statistically accurate ways to measure this, but I don’t think you need to have a PhD in statistics to run marketing experiments. The bottom line is if your data sample is low, your confidence in the results might be lower than if the sample size is high.

Cycling These Experiments

Experiments-resized-600.jpgIt should be obvious that you could be running hundreds of experiments right now across all of your marketing and sales efforts. The faster you move through your experiments, the better your results. The more concurrent experiments you can run, the better your results.

However, I know time is always a constraint in almost every client scenario, which brings us to prioritization. You’re going to want to deploy a prioritization model that helps you decide which experiments to run and which ones to wait on. The best prioritization methodology to consider is this: Start your experiments with the ones that will produce the biggest potential lift for the least amount of effort. Get those in the queue first and then start stacking up experiments that might produce big results but take more effort or produce lower results and take a small effort.

This is going to help you focus on those experiments with the biggest potential to produce results.

No matter what type of marketing youre practicing, you need data on the performance of that marketing. Even if you’re doing direct mail and cold calling, you need all the results-driven data on those practices. How many calls are you making, how many appointments are you getting, what’s the conversion rate, how many of those appointments are turning into proposals and how many of those are converting into sales? You want every single conversion rate up and down that funnel.

Today, 100% of what we do and 100% of what you do must be quantifiable and measurable. Once you have the benchmark performance data, run a series of experiments to improve those metrics. If we change the script, how does the conversion rate from call to qualified lead improve? If we change the headline on the landing page, how does the conversion rate improve? If we shorten our contract from 10 pages to two pages, how does our contract to customer conversion rate improve? It’s all about the numbers.

Square 2 Marketing – Innovating Marketing And Sales To Match Today’s Buyer Behavior!

avatar

Mike Lieberman, CEO and Chief Revenue Scientist

Mike is the CEO and Chief Revenue Scientist at Square 2. He is passionate about helping people turn their ordinary businesses into businesses people talk about. For more than 25 years, Mike has been working hand-in-hand with CEOs and marketing and sales executives to help them create strategic revenue growth plans, compelling marketing strategies and remarkable sales processes that shorten the sales cycle and increase close rates.

COMMENTS