This project feels short, but it will actually take some time.

It’s one of the highest-leverage things we do after we launch a new landing page, and it can drastically improve conversion. Especially when you don’t have massive amounts of data to work with.

Phase 1: Set up and install HotJar

Hopefully, you did this when you set up conversion tracking and have hundreds of user sessions recorded.

If this hasn’t happened yet, follow the instructions from Conversion Tracking > Project: Install Hotjar now.

Phase 2: Filter by audience

Make a copy of this template.

Head to HotJar and click the recordings section. Start looking through the user recordings you’ve been gathering.

First, make sure you only look at traffic from countries you care about. (Scroll down in the “filters” section on the right.)

Then, break down your recordings by the different acquisition channels. To do this, you can filter down by text that shows up in the landing page URL. This is where your UTM tags come in handy.

For example, if you wanted to track everyone who came from a Facebook ad, you could filter like this:

Phase 3: Find patterns

Look for common patterns that:

• Cause people to leave your site or have a bad experience
• Pop up across at least 3 recordings.

Without 3 recordings to base your guess on, there’s a high chance you’re seeing an anomaly: a robot, someone who mistakenly came to the page, etc. We want to zero in on the legitimate sessions your page gets.

Specifically, look for parts of your page that people don’t engage with. For example, maybe they don’t scroll all the way to the bottom and never see your last feature sections, maybe they don’t realize a button is a button, etc.

Remove those sections. Or, change your page to compel visitors to scroll to them.

But before you jump to those conclusions…

Phase 4: Validate your guess

Look at more recordings — see if the pattern holds up across 5-10 recordings for a similar type of audience. If it does, there’s a serious issue with your site that you need to fix.

Phase 5: Break down by other criteria

If you’ve set up your UTM tags correctly, you can also break down by audience; for example, freelance photographers may use your site differently than founders.

• At minimum, break down by:
• Source
• Campaign (audience)
• Device (desktop, tablet, mobile)

Phase 6: Look through at least 100 recordings

When we launch a new campaign at Demand Curve, we look through 100 recordings. Minimum.

Otherwise, we can't easily figure out whether our landing pages (and other pages) work the way we want. Our guesses are often proven wrong.

You need to put in this level of personal effort to really get what’s going on. Data will only tell so much of a story. Though it is important.

Phase 7: Implement changes

Write down your interventions in a bullet-point list and prioritize them. Meet with your product team about anything that would seriously affect conversion.