How to measure, understand and improve engagement with your website

Engagement is one of those words that might make your eyes glaze over. It’s not the most engaging or precise of words, is it. 

It’s very hard to nail down a definition. In fact, many people will have vastly differing opinions.

In the new post-Panda world, Engagement is everything. It’s so important, I’ve accorded it a capital E which I’d usually only do for Everton, or ExpressionEngine. It’s that important.

We need to understand it better by measuring it correctly and regularly, and we need to act on our findings.

But first, a real-world example:

A client of ours came to us with an issue in October 2014. Around the 24^th of October, traffic had dropped around 20%, but many of the main rankings they were tracking had stayed the same. How could this have happened?

We discovered that a portion of their website had been downgraded in terms of rankings. A very specific type of content on their site was performing very badly in terms of engagement – in other words, people didn’t like it. They didn’t engage with it.

Time on page varied from 1 second to 10 seconds, and the bounce rate was nudging 90%, whereas, on other types of content, these metrics were much better.

They worked hard at improving that content – they redesigned the page template to make it more readable, they added videos and slides to make it more sticky, and they reviewed all of the content in line with the keywords they wanted to rank for. The result? A significant recovery.

This case is instructive – and it tells us a lot about how Panda. I’ll get back to that, but first of all, a little explanation about Panda, and what we know (or don’t know).

You need to love Panda

Even if Google Panda acts like this, you need to love it.

Google Panda is all about content. Some fools say it’s all about on-page SEO, and they’re partly right but they don’t know it. It’s all about content, and how people engage with it.

There are no hard metrics to say what good engagement looks like because every industry is different.

Panda is essentially machine learning. It pits you against your competition in whatever niche you’re in and understands which websites people like, and which websites people don’t. It pits you against them in terms of quality – but measures that quality by how people are using the site.

It is partly influenced by Google’s human raters, too. They’re real human beings who go through a set of questions when looking at your site – questions like “Would I trust them with my credit card?”… (well, would they?).

That’s a million miles away from the original presumption that the websites with the most links will outrank everyone else (although as I’ll discuss in another blog post, this is probably still the case).

So if you’re a software reseller and the average for all software resellers is 2.3 pages per visit, then consider that a decent benchmark to exceed. It’s never that simple, but view it like this:

If people see your website in the search results, click on it and bounce back immediately to look at the next website in those same search results - you deserve to lose that ranking position.

So rather than looking at Panda and trying to analyse it mechanically, think of Panda as a mindset. It’s an evolving beast, learning all the time, trying to sort out the high performers from the low performers, alongside other filters and algorithms that measure other things such as links, on-page SEO, indexability, etc.

The impact can be disastrous: part or all of your site can be downgraded in the search engine rankings simply because people don’t like it.

Now you love Pandas, love data

There are some easy metrics to understand:

• Time on page - how long a user spends on this page on average

• Time on site - how long a user spends on this site on average

• Pages per visit - how many pages a user visits on the site on average

• Bounce rate - what proportion of people view this landing page and leave immediately after just one page view

• Exit rate - what proportion of people exit the site from this page

For less clarity, have a look at Google’s explanation of bounce rates and exit rates. It’s not written for human beings, so don’t worry if you don’t get it.

In short, a bounce is where someone lands on a page for the first time in a session and leaves.

Your metrics might be good or bad – the truth is you don’t really know what your industry benchmark is, although similarweb.com have a nifty tool which might help you understand how you’re doing against your competitors. 

They use a network of millions of workstations (probably yours, if you visit similarweb.com) and they take that data to understand engagement with websites. If you run a PR company and you’re not using it, you might not be running a PR company for long.

Try out our Macro

What you can do right now is get down and dirty with your data. Find the pages that are underperforming and the pages that are overperforming. 

We’ve created a handy Excel Macro for you, which does the job for you, you can download it here.

Here’s what you have to do…

1. Go to Google Analytics -> Behaviour -> Site Content -> All Pages
   
2. Increase the number of rows to 5,000
   
3. Export to CSV
   
4. Open up the CSV file and Copy all
5. Open the CLD Panda Macro
6. Paste all into the tab that handily says “Copy-paste your data in here”
7. Run Macro

Give the monkeys some time to process the data, and what you’ll get is this:

1. A tab with your pages sorted by lowest time on page
2. A tab with your pages sorted by highest bounce rate
3. A tab with your pages sorted by highest exit rate
4. A tab with an assessment of each page’s risk level

The spreadsheet allows you to cross-reference your data… so for instance, if you find a page that has a low time on page (say, 10 seconds), but the bounce rate and the exit rate are both low, then it probably doesn’t matter.

Therefore, in the “Pages at risk” tab, it will be listed as LOW.

If you have a page that is high up on all three tabs (time, bounce, exit), then naturally, the risk for that page is going to be HIGH.

Understanding & Interpreting the data

I love data, but most people don’t.

So you have to interpret this data and bring it into the real world. Numbers are just numbers, you have to tell the story and make it actionable.

In my mind, Panda is doing just this with your data. 

Let’s not forget, that with millions of Google Chrome users, this data is being sent to Google constantly, and if they’re not using it in the algorithm, they’ll be looking for a way of doing so. It makes sense.

Panda is analysing your data and sorting it, and saying to itself “you know what, this site has some good stuff, but there are about 20 pages here that people hate…” and it’s creating a checklist:

• Does it have a low time on site?
• What’s the bounce rate like?
• What’s the exit rate?
• Are people scrolling down the page? (heatmapping would be good, eh)
• Do people go to other pages from here?
• How does this compare to similar sites?
• … and so on

Your job is to pinpoint those pages with the data at your fingertips, and do something about it. The macro gives you those priorities, and helps you sort that data to say “here are 10 pages at risk, what can I do to improve those metrics”. It might be better content, or more content (“thin copy” is particularly bad)... but it might be something else.

How to improve engagement on your website

I’ve always said that “Good Design is Good SEO”, and I’ve often been proved wrong.

Thanks. For. That. Google.

It’s true that not every visitor cares about good design, so long as they can do what they need to do on a website.

Maybe “Good UX is Good SEO” is the right expression, although some websites have a horrible UX and they still rank. The key is that users are able to do what they intended to do.

And as we’ve seen, we can measure that, so therefore Google can measure that.

Here are some guidelines…

Design in your content

We did this for Animas – we took an old website that had blocks of text are people weren’t reading, and we redesigned the content so that it was more readable. 

That meant strips of text, in nicely designed boxes, with images. More bullet points. Bigger images.

That meant better headers, a nicer font, and a template for writers to follow when modifying content… designers and copywriters working together.

The results were impressive – an immediate leap in pages per visit, and an immediate leap in time on site. Add to that a reduction in the bounce rate, and you’ve got a recipe for success.

Design your content in – don’t expect designers to work without content, because they’re designing the words as much as they’re designing the layout.

Make your site quick

Page load speed, or time to the first byte, is hugely important. Use Google’s PageSpeed Insights page to assess how Google sees your site, and make the suggested improvements. 

By going through this one by one with a developer and getting them sorted, you will make a massive difference. 

Slow sites don’t rank as well as fast sites – generally speaking. Again, you know the metrics – impatient people won’t hang around so the bounce rate and exit rate will rise on your key pages. 

And if you’re afraid of mobile speed, then you’re right to be afraid. Speed is everything, so consider what goes in - and what goes out.

Get better at writing

Your writing style affects your bounce rate.

Your grammar & your style both affect how people perceive you, and how people engage with the rest of your content. If you write in long sentences, with long paragraphs, then people will either skip to the next bit that’s readable or give up.

If you can’t distinguish between your and you’re, then you’ve immediately damaged your brand, and you’ve lost a visitor.

Usually me.

So, I’d recommend you improve your writing. Take some classes, maybe – but most of all, write a lot. The only way to improve your writing is to keep doing it, and lots of it.

Be clear about the next steps

Back to that Animas site – one of the reasons we hid the navigation and introduced four paths from the home page was that we knew what our users wanted to do. We’d carried out the market research and we knew our user personas.

So we gave them exactly what they wanted. No confusing drop-down menu system, just a way of navigating for our key personas, so that they get to the pages they need to see first.

A lot of B2B websites get hung up on internal jargon. They’ll populate their menu with “Solutions, Services & Products” as if that’s actually what people want. Who really knows whether they want a solution, a service or a product? 

Be clear about where you want people to go… they’ll thank you for it.

Make your Calls to Action contextual and easy to find

We have to assume that Google knows what people are doing and whether their journey through your website has been satisfactory or not.

A natural conclusion would be that they don’t go back to the search results.

Job done.

Calls to action are often underestimated by some websites. There’s a contact button, which is always on the navigation, and some might see that as enough. It’s not.

Make calls to action relevant to the user journey. You wouldn’t thrust a pop-up in someone’s face the minute they enter the website, would you? Well, some would…

You wouldn’t offer a download to someone who’s looking through your “About Us” pages.

The user journey has to end naturally, with the right conclusion for the right person, in the right place, so map them out. Make them clear, unobtrusive but obvious, and make them relevant to the page.

Satisfy different user types

You can’t keep everyone happy. Well, you can try…

For instance, I personally never watch videos. I get frustrated waiting for someone to get to the point when I can read it in far less time.

However, plenty of people feel the opposite way, so interspersing text with different ways of ‘consuming’ content is a good way to go.

You can throw in slides, embedded videos, Infographics, and Canva graphics to instantly break up the text, make it more engaging, and your metrics will start to move in the right direction.

Turning around or averting a Panda penalty

We talk about Panda as if it’s a big hairy monster waiting to destroy our websites. But what if we looked positively at it – not just averting a penalty, but trying to give it content that people consume.

Firstly, analyse your content and those engagement metrics. Look for a benchmark and look at the content that is underperforming for you. 

Conversely, look at the content that is over-performing, because this is stuff you’ve already done that people like. If you can replicate that method with the under-performing content – you might be on to a winner.

Look at design and the user experience, and match up your users’ intent with what you want from their visit.

And when you look at engagement in this light – it makes so much more sense. Don’t over-engineer the Panda filter, it’s always going to change… but understand what you can understand, and act upon what you can act.