The web site visitor is always right – even if she doesn’t know it.
According to the guest on our second podcast, Dr. David Darmanin, founder and CEO of Hotjar, your customers and website visitors tell you where and how to change your landing page to boost conversion. Sometimes directly; more often indirectly. But they will tell you. You just need to know where to look and how to analyze what you see.
With over a decade’s experience in conversion optimization and user experience; David offers his insights, both theoretical and practical, on translating user behavior data into actionable decisions where and how to change your page.
Here’s a bit of what David shared during this podcast.
Starting Out Strong
So much of landing page optimization is reactive. I wanted to get David’s thoughts on how to build a stronger landing page right from the start. He said the most common mistake companies make is focusing on what they want to sell, rather than what the visitors want and what will affect their decision to act or not. And he offers his solution.
David: If I’m building a new page from scratch, what I should do, is to use my beloved tool, which is the survey, which is extremely effective. Existing customers or users are a huge asset to leverage. You start to get an idea of what’s actually motivating customers, what nearly stopped them. These are the things that you really want to build your landing page around. One really nice golden nugget that comes out of this process is that you start to see patterns in the actual words that your users actually used. That is, that’s truly invaluable.
In the podcast, David suggests specific questions to ask and shares how his company has used this kind of survey feedback to make changes on landing pages.
[tweetthis twitter_handles=”@ElisaKapha”]David shares how Hotjar has used survey feedback to make changes on landing pages.[/tweetthis]
Why Multiple Choice Questions Don’t Work
While David loves the survey, both for building and refining a landing page, he’s less enthusiastic about multiple choice questions. I had noticed on the Hotjar website that they make the recommendation to use open-ended questions in the webpage survey, instead of a quantitative question format. I asked David what motivates this recommendation.
David: The moment you’re [using a multiple choice question], you’re already assuming that you know the answer, that you’re just giving them the choice of which answer it should be. Whereas, the biggest wins are going to come from things that you’re not yet aware of.
If you are doing a poll and saying, “Why did you not sign up? Was it money? Was it this or was it that?” that’s awesome because it gives you a nice graph and it takes you 5 seconds to see the results. Typically, whatever is easy is not very effective. Instead, it’s much more effective to say, “What stopped you from signing up.” Unfortunately, you need to spend some time processing the results, but that’s awesome because, as we say, you see, you leave it much more open ended to them and then you can actually see the actual words, terminology they use.
Later on the podcast, David explains the process Hotjar uses to bring some quantification to their analysis of the qualitative responses. It does take more effort than processing a multiple choice survey, but David is adamant the results are worth the time.
The Great A/B Debate: Incremental or Go Big?
I’ve had the chance to speak with many landing page optimization experts. And frankly, I spend a large amount of time reading about it as well. I’m always curious about their thoughts on the A/B debate – do you need to test a single change at a time, or make major changes if you want to see a major lift in conversion rates? David’s perspective? Whether you’re making a big change or a small change – be bold.
David: I think anyone who’s giving you advice on whether you should do big changes or small changes, I think it’s completely off topic when you think about it. It’s like telling a scientist how to understand or uncover a new learning by telling them how long or small the experiment should be. To me, it’s more about based on [why] are you making the change. That’s the big question. What’s driving the change, then you should be doing a mix of both. Now, having said that, I’m a big exponent of bold changes. Bold changes don’t necessarily have to be big.
So what does David mean by “bold”?
[tweetthis twitter_handles=”@daviddarminin”] I’m a big exponent of bold changes.[/tweetthis]
David: A bold change could be completely changing your headline to be a completely different angle and it takes you 5 minutes to do the test. Again, even in this case, it comes back to what’s driving your decision to test. Typically, from my experience, it’s so easy to get caught in these random ideas which are coming up within the organization. You can’t do a landing page optimization via democracy, kind of. It needs to be on proper research, of understanding how is the page being used.
Listen to the podcast (above) to hear David share some case studies of changes they’ve made to landing pages based on user feedback, including his interesting take on how to allocate your landing page’s real estate.
Data Sampling: It Works because We’re Predictable
Hotjar uses a data sampling technique to creating its heat maps since that’s a more cost effective approach. Not being a data scientist myself, I was curious how the tool selects its data sample in a way that ensures the sample is representative.
David: Just to clarify, we never stop collecting data from one particular visitor. The sampling is seen on a visitor level. If I have a shop and I have 10,000 visitors coming in to that shop, what we’re doing is we’re stopping 1 out 10 of them and asking them questions or observing how they behave. That type of analysis or sampling-based analysis is very representative of the total population.
Now, we give guidance as to what the sample rate should be. Most of the sites that do use Hotjar, the sample rates that we do provide on the different plans are more than enough, but we also offer guidance to sites that have much more traffic where their sampling needs might be higher and we can accommodate them by increasing the sample rate.
But how do they select which 1 out of 10 visitors to record?
David: It’s equally distributed over time. That’s key in our sampling algorithm. Week after week, month after month and across different geographies, time of the day, [human behavior] rarely changes that much. That was one of the motivating reasons to do Hotjar the way we did it as the patterns are so clear and so obvious.
Hotjar has one more unique aspect to its data sampling approach, which is using a report-driven approach to selecting the pages recorded, rather than recording the entire site. David explains the value of this approach in detail in the podcast, and how it relates to understanding the traffic into your landing page.
I wrapped things up asking David about where Hotjar is heading. The company recently celebrated its one year anniversary, and already has a user community 40,000 strong.
David: We’re currently growing the team, which is exciting. We’re focused much more on stability right now because, also, there are some really exciting, new features and new technology that we’re planning to build. At the same time, we’re not taking our eyes off the most important thing, which is customer service. We provide support to our free users as well. The amount of amazing feedback we get from them, we feel that’s extremely valuable.
Listen to the podcast to hear more from David, including his discussion on the eight types of pattern problems they see time and again via the heat maps, particularly how to correct distraction and information patterns.
Show Notes
- David Darmanin on Twitter
- Hotjar company site – You can access Hotjar’s road map for the future of the tool and company if you sign up for a free account.
- Influence: The Psychology of Persuasion, by Robert Cialdini
Elisa Silverman is a B2B content writer with a background in law and technology, who’s spent a career helping diverse groups of people communicate well with each other.
