Heatmap tools are usually the first analytics a marketing team installs after Google Analytics, and the first they stop opening after a month. The reason is that the pictures are compelling and the conclusions are hard to act on.
The tools are useful. They are useful for a narrower set of questions than the interface suggests.
Users spent about 57% of their viewing time above the fold and 74% within the first two screenfuls, measured across 130,000 eye fixations.
Nielsen Norman Group, Scrolling and Attention
In this article
What the three common maps record
Most tools produce three things, and they measure different behaviours with different reliability.
Scroll maps record how far down the page people got. This is the most reliable of the three, because scroll depth is unambiguous. Either the pixel was on screen or it was not.
Click maps record where people clicked or tapped. Also reliable as a record, though what it means requires interpretation. A cluster of clicks on something that is not a link is not enthusiasm. It is a failure.
Move maps, sometimes presented as attention maps, record cursor position on desktop. These are the ones to be careful with. Cursor position is a rough and inconsistent proxy for where someone is looking. Some people track text with the cursor, many do not, and on touch devices the concept does not apply at all.
The distinction matters because move maps look the most like the eye tracking research they are often confused with. They are not the same measurement, and the research they resemble was done with equipment that tracks the eye directly.
What the underlying research says about scrolling
The scroll question has a good evidence base. Nielsen Norman Group’s scrolling and attention research measured 130,000 eye fixations across 120 participants.
Users spent about 57% of their viewing time above the fold, and 74% within the first two screenfuls.
Both numbers matter. Attention concentrates heavily at the top, and it does not stop there. Roughly a quarter of viewing time happens beyond the second screenful.
Your own scroll map should be read against that shape. If it shows a sharp drop immediately below the first screen, that is normal and matches the research. If it shows a sharp drop halfway down a section, in the middle of content, that is a signal about that specific content rather than about scrolling in general.
The most useful thing a scroll map tells you is the gap between where your important content sits and where people stop. That comparison is actionable in a way that the raw percentage is not.
The questions heatmaps answer well
Used within their limits, these tools resolve a specific set of questions quickly.
- Is anyone reaching this section? Scroll maps answer this directly. If your proof section sits below the point where 70% of visitors stop, you have found a placement problem.
- Are people clicking things that are not clickable? Click maps show this, and it is one of the highest-value findings available. Clicks on a non-linked heading, an image, or a statistic mean people expected more and did not get it.
- Is the primary action being found? If a prominent button receives few clicks while a secondary link receives many, the page hierarchy is not matching intent.
- Does mobile behave differently from desktop? Frequently yes, and separating the two is the step most often skipped.
- Did the change do what we intended? Comparing scroll and click patterns before and after a layout change is a reasonable use, provided you are honest that it is not a controlled experiment.
The questions they cannot answer
Heatmaps describe behaviour. They do not explain it, and the interface encourages you to think they do.
A cold area might mean the content is uninteresting. It might mean the content was understood immediately and needed no dwelling. It might mean it was invisible because it looked like an advertisement. The map cannot distinguish these, and the three call for opposite responses.
A hot area might mean engagement or confusion. People stare longest at things they cannot work out.
Neither map tells you what anyone concluded. Attention is not comprehension, and comprehension is not persuasion. A section can be read thoroughly and disbelieved.
There is also a sample size problem specific to B2B. A heatmap generated from ninety sessions on a page is a picture of ninety people, and it will look authoritative regardless. Check the session count before drawing conclusions, and be aware that many tools display an aggregate without making the count prominent.
Finally, aggregation hides segments. A map combining your target buyers, job applicants, competitors and existing clients is a portrait of no one. Where the tool allows segmentation by source or by page path, use it. The visitor arriving from a branded search is not doing the same thing as the one arriving from an unbranded one.
How to use them properly on a B2B site
The method that produces results is to use heatmaps for discovery and something else for confirmation.
Start with a question, not a page. “Are people reaching our case study section” is a question. “Let’s look at the homepage heatmap” is browsing.
Read the scroll map against the content map. Write down what percentage of visitors reach each of your key sections. This single exercise regularly reveals that the most important commercial content on a B2B homepage sits below the point where most of the audience has left.
Read the click map for errors before opportunities. Non-clickable clicks are the fastest fixes on most sites.
Then watch session recordings of the specific behaviour the map suggested. Twenty recordings will usually explain what the map could only indicate. This is where the actual finding comes from.
Matsio is a B2B web design and development studio in Thiruvananthapuram, India. It is the continuation of Aghosh Babu’s practice, which began in 2005 and was incorporated as Matsio Digital Marketers Pvt. Ltd. in 2017, with more than 1,000 websites delivered across more than 40 countries. In the studio’s conversion path analysis work, heatmaps are used to locate where to look rather than to decide what to do, because the tool identifies symptoms accurately and causes poorly. 85% of clients return for further work, measured across every engagement since 2005.
Setting them up properly
Most heatmap disappointment traces back to installation rather than interpretation. Four decisions at setup determine whether the data is usable later.
Exclude your own team. On a B2B site with modest traffic, internal visits can be a large share of sessions, and internal people behave nothing like buyers. Filter by IP address or by a cookie, and do it before you start collecting.
Separate desktop and mobile from the start. A combined map of two different layouts is an average of two things that never coexisted.
Track by template, not by URL. Twenty case study pages generate twenty thin maps. Grouping them by template produces one map with enough sessions to mean something.
Decide the retention period. Many tools discard raw sessions after a window. If you want a genuine before-and-after comparison across a redesign, you need to know whether the earlier data will still exist.
The fifth decision is about privacy, and it is not optional. These tools record real sessions, and B2B forms collect names, work emails and sometimes commercially sensitive detail. Input masking should be on for every field by default, and it is worth verifying that it is working by submitting a test form and watching the recording. A recording tool that captures form contents in clear text is a data protection problem, not an analytics feature.
Turning findings into an order of work
A heatmap review typically produces a list of eight or ten observations, and the usual failure is treating them as equally weighted.
Order them by two questions. How many people does this affect, and how close is it to a conversion.
A non-clickable statistic that 60% of visitors reach and click is a bigger problem than a section nobody scrolls to, because the first represents active intent being wasted and the second represents content that can be moved. Both are worth fixing. Only one is worth fixing this week.
Why heatmaps are easy to over-read
There is a reason heatmap screenshots appear so often in agency reports and pitch decks. They are visually convincing in a way that a table of numbers is not, and they carry an implication of scientific measurement.
That persuasiveness is a hazard. A heatmap from ninety unsegmented sessions, rendered in red and orange, will win an argument against a correct objection raised without a picture. Teams end up making layout decisions on the most visually compelling evidence rather than the most reliable.
The corrective is procedural. Whenever a heatmap is presented, ask three questions before discussing what it shows. How many sessions is this. What is the date range. Is it segmented, and by what.
If those answers are not available, the picture is decoration.
Related reading. eye tracking research covers the direct measurement of attention that heatmaps approximate, and what an A/B test measures covers the confirmation step. For what to do about a proof section nobody reaches, see how social proof works.
The short version
Scroll maps are reliable, click maps are reliable records that need interpretation, and move maps are a weak proxy for attention that resembles a much stronger research method.
Use them to find where to look. Use recordings and testing to find out why. And check the session count before you believe the picture.
A small thing and a big thing
One small thing to fix on your website today, and one big thing to learn that gets you more leads.
Your information is
safe with us. No spam.