Skip to main content

“How do I increase my conversion rates?”

Oy vey.

(For those unfamiliar with the phrase, that’s Yiddish for a nice, big eye roll.)

I get asked this question a lot when I’m at conferences chatting with eCommerce store owners.

Truthfully, It’s not fair for me to give ‘em the yiddish eye roll. Increasing conversion rates is the most critical revenue driver a store owner can act on. It’s not the question that bugs me; it’s the fact that I know my answer will be underwhelming to them.

When asked the question, “How do I increase my conversion rates?”, my answer is:

“Use visitor research to understand the absolute deepest reasons why your customers buy your products. If you don’t feel a bit awkward doing this research, you haven’t gone deep enough.”

When I give this answer, half of the people give me a look like, “hmm… that’s nice” and walk away, but the other half say, “Good lord, tell me exactly how to do this.”. <- These are my people.

So, let’s get tactical. Here’s how to get profitable, conversion boosting insights using visitor polls.

What are visitor polls?

A visitor poll is a tiny widget that asks a visitor a question you’ve come up with at a time you’ve chosen. Visitor polls are the easiest way to collect qualitative research at scale.

“Eww… so it’s like a pop-up?”

Yea, it sort of is. Reeeeelax, your customers won’t leave your site running and screaming just because you’ve asked them a question, I promise.

Our favorite tool for launching visitor polls is Hotjar. Sure, there are others, but we’ve found Hotjar to be the best option.

What types of questions should I ask?

Our goal with visitor polls is not to know whether customers prefer “this” or “that,” we want to know why.

The best type of questions to ask are open-ended questions.

Multiple choice or yes/no questions can work for specific research, but for deep research stick with open-ended questions.

Don’t make your questions easy to say “no” to.

There’s a rule that comedians have that goes like this:

Always say “yes.” Better yet, say “yes, and…”

If you’re doing an improv sketch with a comedy partner and you ask, “Are you heading to the grocery store?” and they respond with “No,” that’s it. They’ve killed the flow of the sketch before it even started.

We want to craft our visitor poll questions so that we make it easy for our visitors to say “Yes, and…”

Here’s an example of a bad visitor poll question:

“Are you having any difficulty navigating our website today?”

When you ask this question, 80% of people will answer with: “no.” And of the 20% that say “yes,” they won’t add any comments. It’s too easy to say “no” to.

Don’t do this.

Let’s reshape that question:

“What’s the most frustrating thing about navigating our website?”

Ya see, by tweaking this question we’ve optimized it for better responses. You can’t merely say “no” to this question.

Also, if you noticed, I changed the word “difficulty” to “frustrating.” Not an accident. When you ask if something is difficult, you’re putting it on them, as if it’s their lack of intelligence that’s the problem. Switching this to “frustration,” we’ve lifted the responsibility from them and put it on us, where it belongs.

Show the polls where it makes sense

When you’re building polls, you can choose what pages they run on and precisely when to show them. Timing and placement are critical, for instance, don’t show a poll question on the home page that asks, “What products would you like to see us add?”. Chances are they haven’t seen most of your products; they’re only on the home page.

Also, only show a poll once. Don’t drive visitors crazy by continually showing them a poll question while they’re trying to shop your website.

Gather enough results

The great thing about visitor poll data is that you don’t need a massive sample size do start finding significant insights. It’s not like A/B testing, thank god.

I like to look for about 100 responses before I close out the poll and review the results. More is fine, but 100 will do.

Finding insights from your poll data (w/ examples)

Alrighty, so here’s where I get oddly specific about this process. Why? Because it’s where most companies fumble the ball. It’s easy to do this research, take a peek at the results, feel good about what you’ve done, and shove it in a drawer. I see this all the time.

We have to follow through and finish strong here.

So, here’s my exact process for getting (and acting on) the insights from your research:

  1. Kill some trees. Print out the poll results on actual paper. Use recycled paper if you need, but this is essential.
  2. Go outside. Ok, “outside” doesn’t truly have to be outside, but go somewhere out of your office where you won’t be interrupted. Head to a coffee shop, take a walk, sit in your car, who cares, just get away from Chad who can’t wait to tell you how lit his CrossFit class was.
  3. Find “high-emotion” words. Use a pen or highlighter to scan for “high-emotion” words. What are high-emotion words? I’ll give you a short list here, but it’s a bit of a “you’ll know them when you see them” thing. Examples: love, hate, always, never, scared, worried, excited, cheap, pumped, stoked, etc.
  4. Look for trends. You’ll be surprised looking at 100 or so responses how quickly patterns will develop. Categorize the trends that you see, and highlight the best examples in each category.

Here’s an example:

Let’s put ourselves in the shoes of Jill. Jill runs Coffee Crazies, a direct to consumer coffee company that I just made up. Jill ran a visitor poll on her product pages that asked the question, “What’s preventing you from adding this product to your cart?”

Damn, Jill. That’s a good question.

Jill scanned the results, and these five responses were repeating trends.

  • “I’m worried it’ll be too bitter tasting.”
  • “How much coffee does this package make?”
  • “It this ground or whole beans.”
  • “I hate being jittery, how much caffeine is in this?”
  • “What if I don’t like it, can I return it?”

At first, Jill was a little pissed. She thought, “I’ve already answered these questions in the product description, what more can I do?” But she then remembered that most website visitors act like really lazy drunks. She came up with a plan to implement these insights and make it easier for customers to solve these objections.

Jill added a “benefits-based” bulleted list above the add to cart that answered the questions from above. Here’s how it read:

  • Smooth, robust flavor. Never burnt or bitter.
  • This package makes 75 cups of fresh coffee. (That’s $0.17 per cup)
  • Pre-ground beans, ready to brew.
  • 95mg of caffeine (Enough to get you going, minus the jitters)
  • Just want to try it? We have 4oz. Sample packs here.

Run these polls quarterly

I don’t advocate having visitor polls on your site all the time, but they’re essential to revisit on a quarterly basis to see how the information has changed. Plus, once you start running these polls, it’ll become addicting. When you hear someone in your office say, “I’m pretty sure like 95% of our customers don’t care about that!” you’ll say “Oh really, let’s find out.”

Josh Frank

Author Josh Frank

Founder & Head of Optimization

More posts by Josh Frank