When we first went about designing RevaHealth.com we had a very specific user flow in mind. Visitors to our site would arrive through a search term like ‘Dentists in Rathfarnham’. They would land on a specific search results page displaying dentists in Rathfarnham and then would narrow their search by location, treatment, etc. Then they would look at individual clinics’ profiles and compare them until they converted by contacting their chosen clinic.
Landing to Conversion Flow
But the best laid plans of mice of men and all that… Over time an increasing number of our visitors started landing not on our search results pages but directly on the profiles of individual clinics, arriving as a result of very long tail searches.
The problem we faced was that while our search results pages were designed specifically to be landing pages, our clinics’ profile pages were not. As a result of this our profile pages were suffering from a much higher bounce rate than our search results pages. For specific clinic types it was as much as 20% in the difference, and it averaged 10%.
Why Do Our Search Results Pages Work As Landing Pages?
If you search for ‘Dentists in Rathfarnahm’ in Google this is the result you will get:
Dentists in Rathfarnham
If you click on this search result you should have a clear expectation of the type of page it will bring you to. In most cases the user will be expecting a page from a 3rd party site featuring several or all dentists in Rathfarnham, and this is the page we bring them to:
Rathfarnham Search Results
When the user lands on this page it is immediately clear that they have landed on a relevant page and therefore they don’t bounce.
Why Didn’t Our Clinic Profile Pages Work As Landing Pages?
If you search for the Rathfarnham Dental Practice in Google you get the following search result:
Rathfarnham Dental Practice
If you click on this link you are clearly expecting to find information about the Rathfarnham Dental Practice. Now look at the page we brought you to:
Rathfarnham Dental Practice Profile
The most prominent branding on the page is RevaHealth.com, not Rathfarnham Dental Practice, and the relevant information is a third of the way down the page. Many users were landing on the page and hitting the back button before they discovered that we had the information they wanted.
So What Did We Change?
We dynamically alerted the page that the user was landing on it directly from Google, Yahoo, etc, or alternatively that they were navigating to it internally from RevaHealth.com.
When users were landing on the page we moved the RevaHealth.com logo and search boxes into the right hand gutter. We promoted the clinic name into the main place on the page so that it didn’t get lost in the content. We also added a stock photograph.
Updated Rathfarnham Dental Practice Profile
We intended for these changes to have to two effects:
- Move the relevant information further up the page
- Better meet the visitors’ expectations that they were landing on a page specifically about the clinic
What Were The Results?
The results of the experiment were far better than any of us anticipated:
- The bounce rate on profile landing pages dropped by 6%
- The average number of pages viewed by visitors who entered the site through profile landing pages went up by 19%
- The average time spent on the site went up by just over 20%
- The conversion rate increased by 14%
- Advertising revenue decreased by 2%. We suspect this is because we were counting users who landed on the site and immediately clicked on advertising as having bounced. In addition, the advertising in the right hand gutter has been pushed below the fold.
Concerns
We do still have some concerns about this approach, mostly centring around the design of the site changing when a visitor navigates away from the clinic’s profile to the remainder of the site.
Let us know if you have any success or failure stories about increasing conversion and reducing bounce rates on your sites.
Nice results Caelen. Quick question: Can the fact that you are dynamically presenting the “google user” with a different page from the one presented to the “reva health user” be considered cloaking or not acceptable by Google’s policies? I remember reading something not long time ago but can’t with the bookmark right now.
Hi Facundo, I was just talking to Caelen about this yesterday. We think we’re OK, i.e. we’re not cloaking. Here’s a couple of quotes from Matt Cutts blog:
“Cloaking is serving different content to users than to search engines.”
“… when a Google user clicks on a search result at Google, they should always see the same page that Googlebot saw.”
Source: http://www.mattcutts.com/blog/a-quick-word-about-cloaking/
The redesigned landing page is what we serve to users landing from a Google search and also to Googlebot, so there should be no problem. We still serve the old design to people who come to the page while browsing within RevaHealth.com, but even this is the same content just styled slightly differently.
I don’t see us having any problems with this approach but if anyone has a different opinion I’d love to hear it.
Good point Phillip, I think you are right, seems that you would be presenting the bot and the searcher the same stuff.
If this is pure CSS implementation then you’re likely fine, but looking at the screens I’m not sure how you’d do that with pure CSS. It’s very unlikely to cause issues, but Google does use some automated routines to identify cloaking AFAIK, and serving different content based on referrer/user-agent can be dangerous.
Nice post, and an interesting idea. What happens when users move to other pages and the branding changes though? Does this cause confusion I wonder? What was conversion like for users who viewed other pages after the initial profile page?
I’d consider seperating PPC site content from normal visitor content a basic but hugely important step for evaluating the PPC channel. It also allows split testing more manageable as PPC traffic is very different from direct.
The great thing is that any improvement can then be used as feedback for the main site content.
@Barry Hand
Reva Health were analysing traffic from organic search, as far as I can see in searches, they do not do PPC.Are you suggesting they should start and use it to analyse the content?
@Richard it was a concern when users navigate around the site, however the performs seem to indicate that it didn’t matter. I guess the screen size, colours, fonts, etc. are sufficient to maintain the user experience.
The page changes are done in Javascript, however there is no different content. They are purely positioning changes and as @facundo says we serve a user coming from Google or indeed anywhere else other than ourselves the same thing style and all as we do to the bot. Interested to hear if you still think this is an issue.
@kararynka we do PPC on a campaign basis. We also do it for keyword research and as Barry mentioned specifically to refine pages.
Great post and some really interesting insights. Like Richard said, i would of been interested to hear how the page layout changing effects conversion rates for users who land on the page, view another page and then return. Interesting you see no difference.
Is there a reason you don’t use the search landing page for all users weather they are internal or external (from search engines) if the metrics have improved on the back of this ?
@searchbrat This is organic traffic so we can’t control the landing page. In these example there is specific text on the profile page that the user has search that isn’t on the search pages.
Great post Caelan – we at http://www.whoseview.ie are constantly looking for ways to reduce bounce rates.
[...] King of RevaHealth published an excellent case study on Tuesday outlining how they reduced the bounce rate on key landing pages by 6% by tweaking the layout of the content. RevaHealth is a business directory specialising in [...]
We’d love to see this uploaded to A/B Tests with all the relevant data shared in our testing community. Give us a look – http://www.abtests.com/ and also follow us on twitter @abtests
Thanks for uploading the test results that you can find here on A/B Tests – http://bit.ly/c4JWBy
Very informative post Caelen. Well done on a successful outcome.
I would be curious to know whether any particular element contributed most to reducing bounce rate. Did you perform a straight A/B test comparing the original page with the redesigned page, or was it possible to perform multivariate tests measuring the impact of individual elements e.g. adding the stock photo?
One significant difference I see between your “Dentists in Rathfarnham” page and the “Rathfarnham Dental Practice” page is the prominent position of advertising in the latter. Might this have been a contributing factor in higher bounce rates? It would be interesting to know how the original page minus this advertising would perform.
Good thinking Paul.
@Caelen – I’m sure you’d be able to set this up yourself, but if you wanted to build an MVT and need any help ping me – happy to help to see if everyone can derive more information from this test.
Rgds
Richard
@Paul – it was a straight A/B test, which is what we generally do these days. Well spotting on the advertising and you are correct the advertising on our free clinics does increase bounce in comparison to our search page. We have already tested this figure but I don’t have the results to hand.
@Richard many thanks for the offer. We will get in touch if we can free up some bandwidth
[...] were also very interested in the blog posted from Caelen King over at Revahealth.com titled “Improving landing pages to reduce bounce rates”. Like Revahealth.com, we are finding that the majority of our daily visitors are entering WhoseView [...]
[...] Another fine post from WhatClinic who keep sharing their experiences. This one on bounce rates. [...]