In my travels around the world the most frequently asked question is: "What's your favorite web analytics report? "
A close second is: "How can I improve my web pages with high bounce / low conversion rates?"
Or "I have done all I can and I don't know how else to improve my webpage, ideas?"
If you think about it for a moment it is not a very hard question.
I believe the insights for improvement exist at the intersection of customer intent and the webpage's purpose.
Let me explain.
The Customer Intent – Webpage Purpose Gap
There is a very simple reason many websites and web pages have a very high bounce rate, and in turn very low conversions…
There is no connection between why the customer came to the page and what the page exists for.
This could be someone typing in vegetarian shoes into Bing and landing on your web page for swim suits (as happened to me recently).
This could be me visiting www.couponcabin.com and clicking on a $10 off a $35 coupon link to Snapfish and landing on a page for "great new gifts" or, the other day, landing on a page that said "Get a free deck of cards". What! Where's the scent?
Never let your campaigns write chq's that your website can't cash.
Fix that, your outcomes (revenue, leads, donations…) will improve.
The second type of problem is a lot more common…
There is some overlap between what your customers want and what your web page exists to do. But the overlap is not very much, only the most dedicated (say your mom) will put up with the pain required to complete the task.
I land on your site to buy QuickBooks Simple Start but you have it very well hidden because you want me to buy the $400 QuickBooks Premier product.
This could be an email campaign you sent me and the landing page does not have a one click checkout link, though it does have a ton of irrelevant content.
This is every site with a painful flash intro, this is pretty much every site ever created by every big CPG company, this is Propel Water's website where the only reason for your existence is to be impressed by a slow site with a dancing water bunny that hops!
What rarely happens, and what we should all aspire for, is this…
Not only is there a large overlap between Customer Intent and Webpage Purpose, the company's own objectives are subservient to customer needs.
That's how you get to nirvana. That's how you get to low bounce rates. That's how you get to higher conversion. That's how you get to the kind of magic with your website visitors that will make your competitors green with jealousy!
Here's an example, it is extremely rare for me to do a Google search and get a result from Amazon (PPC or SEO) that is not magnificently relevant.
It is rare for me to shop at www.crutchfield.com and not feel that the entire company exists to make me happy. From recommendations that start with Budget Friendly first (rather than most expensive) to humongous product pictures to all kinds of shopping guarantees to bonus stickers and custom helpful manuals that are in the box when I get the product.
They understand my intent (worry free quick shopping to install process) and they have done their best to have web pages whose purpose is to meet that intent.
Moral Of The Story
If you want to have high performing web pages make sure that you :
1. Have a clear understanding of what the purpose of that page is and
2a. a clearer understanding of what drove customers to the page and
2b. what they want to accomplish to ensure that
3. #1 and #2 are in alignment.
Glory will be yours!
Tips For Improving Web Pages (Understanding Customer Intent)
We know what needs to get done, right? I think so.
With each tip below my hope is to share with you how I try to glean customer intent (2a and 2b above) so that I can improve the pages (accomplish #3 above).
Uno: Insights from Sources (URL's, Keywords, Campaigns)
One of the obvious sources for understanding customer intent is to use the sources that drove traffic to your website. In your analytics tool this is all available in one nice window where links are just waiting to be clicked!
You are looking for the Entrance Paths, Sources (referring websites / url's) and Keywords.
To the left of that screenshot, from Google Analytics, you'll also see a segmentation drill down that you can click to see Medium, Campaign, Ad Content etc, all great ways to understand intent for Visitors that arrived via campaigns.
This concatenated screenshot shows the analysis that I end up doing…
In this case the high bounce rate is now easier to understand.
9 out 10 keywords referring traffic to the website are not about the purpose of this page. This page is about web analytics career planning, and only keyword number 6 is remotely related to that. No wonder people bounce.
Now I have several actions I can take. I can either do better SEO so it ranks for the right words. I can add this line to the top of the page: "Hey if you are here to learn about Avinash or about the Occam's Razor then go here and there".
For my other websites I also click on the referring url's and go back and visit those sites and see what they are writing about this page when they link to it. They are saying "get discounts on iPods" and linking to my site. But my webpage is about the ZuneHD! Well I can contact that site and get that fixed or add a promotion on my Zune page for the iPod discounts. Both will help meet customer intent.
Ditto with your campaigns, see what campaigns drive people to the site and what promises you made on those campaigns (content, discounts, calls to action) and make sure the web page reflects those promises.
If you do the first part well then this is how your webpage entrance keywords report should look…
Every single referring keyword is a perfect match for the content on the site. It reflects my hard work with SEO and a perfect match with customer intent.
Dos: Insights from Mis-matched Calls to Action
I think this is the biggest miss when it comes to why webpages stink. The customer wants to do x on the page and you are pushing y.
Take a look at this example from www.frys.com (click on the image for a higher resolution, you know you want to!)…
What is the call to action on this page?
There are three layers of tabs on top, a dysfunctional left navigation (still with lots of choices), a link on top that says Disable Menus (I clicked, nothing happens, hmm…), two sets of searches, DVD deals, Blueray something, category links, ….. lots more.
I know this is a category page, but what's the call to action Fry's wants me to follow? How does that reflect what I might want to do as a customer?
How about some clarity?
Another example. What is the call to action here…
OMG!
The job of this page seems to be to get me to attend MONEYFest. That's the most important thing for this company because that's the only call to action that stands out.
But should it?
Here's one final example to hammer this concept home.
I just searched for Color Laserjet Printer in Google and clicked on three ppc ads.
The Dell ad takes me to a page that asks me to choose if I want Home & Home Office Printers or Small Business Printers. What? I just want a printer. I also don't want to be conned into a expensive price, yet I feel one of those two links will do that. Why should I have to put up with this simply because Dell's business is organized into two divisions? Dell's revenue, analytics tools and number of analytics people is not an issue – all quite large. So why not land campaigns on pages where I get what I want, a printer / netbook / music player with no prices. As a customer I am satisfied, then when I click Show Me Price make me choose Small Biz or Home. Why not?
[ Oh and for the record click on Small Biz, offers were $10 cheaper than Home and Home Office. How mean!]
The HP ad takes me to a page where half of the page is taken up with the menu, dancing promo, best deals of the week link (which sells computers) and at the bottom of the page, almost below the fold is every printer they sell in all categories. I just want a color laserjet printer.
The Xerox ad takes me here:
No crappy menus. No crappy promotions. No home or small business choices. Just a printer. Just a color laserjet printer.
Perfect match between customer intent, content and call to action.
And it compares it to the direct competitor and tells me that with HP I would not only pay more ($749 to $1,299) but the HP printer would cost me $320 more to operate!
I see you think this is all too ecommerce centric. Ok look at www.flickr.com.
When I land on the site I see three links above and beyond all else. Your Photostream. Upload Photos & Video. Your Contacts. That is what I want to do 90% of the time on that page/site. For the other 10% of the times the other calls to action are there, unobtrusive and yet always there.
Look at your web pages. Identify what are the one or two jobs they are supposed to do. Eliminate every thing else. Focus your calls to action.
Tres: Insights from Website Visitors
Why guess how to improve your webpage? Why not just ask them? You know, them. The customers? : )
For site level surveys free onexit survey tools like 4Q from iPerceptions are a good choice.
But for for feedback from pages I prefer specialist page level surveys like the one from Kamplye.
The survey invite sits nicely at one corner of the page and provides localized feedback from visitors about that particular page. What they liked, what they did not, what could be done better.
[The screen-shot above is from my buddy Brian Clifton's excellent analytics blog: Measuring Success.]
Page level surveys won't get too many responses, and are more likely to contain negative responses. But both of those things are quite ok, and you do want all the negative responses. I know that because you have a strong ego!
You can also easily build one on your own (embrace your IT person and get her/him a case of red bull). The benefit is that you can deeply customize it.
Here's the page level survey from the Turbotax Support website… on the left of every page is a floating Yes or No box, and depending on which one you click on you get a short custom survey…
So nice! Try it on this page: What If TurboTax Is Updated after I File?
Both the answers (Yes or No) and the open text VOC ("Let us know how we can improve") will be perfect places to glean clues as to how you can improve your page to deliver against customer intent.
While both these things are easy to do, you should expect to assign atleast part of an Analysis Ninja's time to go through the data and find insights. I know that seems obvious, but I do want to reiterate that.
Cuatro: Insights from Site Overlay
Another excellent way to get into the heads of your customer is to step out of your MS Excel world of rows, columns and pivots. Again! :)
I have always been, and remain, a fan of the Site Overlay report. What better way to infer what customers might have wanted than to look at a visual distribution of visitor clicks on a live page?
I have created this blog with the express purpose of getting people to follow me on twitter (not!!) and by looking at the click density I can see that only 1.5% of the people are doing that. How terrible!
Site Overlay does not always work in Google Analytics so if you are using GA on your site you'll have to look for alternatives.
My love for ClickTracks has been quite clear since day one of this blog, one of the reasons is the site overlay report in CT, it is wonderful, it just works. More wonderfully in CT the right "frame" shows all the key metrics for the page that provide key context, and the bottom "frame" shows Traffic From and Traffic To which is very helpful.
Another tool that is quite good (though it does not have the two contexts mentioned for CT) is CrazyEgg. You can get the heatmap view which is quite good in helping you understand the difference between web page purpose and customer intent as identified by the website visitor clicks (if they did not hit Add To Cart then what are they clicking? why?)
The thing I absolutely adore about CrazyEgg though is the confetti view…
To the best of my knowledge it is unique amongst web Analytics tool, and it is super insightful.
As you can see for a given time period it shows me the click density (clusters of dots), a la heat map. That is cool but not very useful (remember all data in aggregate is crap!).
What is delightful is that it shows the clusters of clicks by top 15 referrers (segmentation baby!).
So, for example, I can see that very few people from amazon and analytics.blogspot.com care to search on my blog or click on links to my podcasts and videos (why?).
Visitors that come from Google and the Direct traffic click a lot on internal site search. Why? What are they looking for? Segment!.
And Visitors from amazon, grokdotcom and my book's site click a lot on the About link.
I can get even greater detail by hovering my computer mouse over one of the dots, which shows more details about that particular visitor (see bottom of the above image, someone who came on keyword "avinash kaushik" took 30 seconds to click through, and hopefully, buy my book, yea!!).
As you do this for your own website you are starting to not just understand the overall clicks (heatmap) but you are actually starting to understand segments / clusters of visitors and what they want and how it is differentiated. What is the job you want your webpage to do, what customers actually do.
They have cheaper plans as well, but most expensive plan for CrazyEgg is $99 per month. That is such a cheap price to pay for the kinds of insights you'll get. If you are running a website with more than 10,000 visitors I don't know why you would not pay this. It might actually be a crime!
[ Note: I am not affiliated in any way with ClickTracks or CrazyEgg, once you install and use them you'll see why I am so fond of both.]
Cinco: Insights from Experimentation & Testing
Did you think I would forget this one? Not likely! My slogan is: Experiment or die!
Seriously though, I can't think of a better way to improve your web pages then to ask your customers what tasks they want to complete and then come up with ideas for how to make your pages better and testing them.
Let me make the point about the power of testing by showcasing a test Christian Watson, from True Games Interactive, wrote about recently.
This was the original landing page…
Pretty darn cool right?
It evokes passion, it is sexy cool, and it does not hurt at all that the calls to action are very clear, you can't miss the Download & Play Now and Sign Up For A Free Account links.
The conversions were good but Christian writes that Marketing (!!) wanted to try something new. To quote him:
The goal of the second landing page was to brighten the page up a little, move more content up above the fold, and remove any non-conversion-related links from the main content area.
Here's "version B" of the page:
Not too shabby, right?
Result?
Again Christian:
I thought we achieved this pretty well; however, the landing page performed very slightly worse than the original.
Before reading the result my first impression was that the second page, smaller and more direct with the video as more prominent, would do better.
Not really said the people who matter: Potential Customers.
At this point most people give up.
Not the people at True Games!
They went back and tried a version that not everyone was totally psyched about…
Dramatically different.
Slightly risky to go with just one call to action (Join Now For Free). No text (About and Features information both gone).
Results?
Once again Christian:
I'll be honest; I wasn't a fan of this design at all. The white background felt too stark to me and it removed the content which highlighted the key features of the game.
The great thing about using an A/B testing tool to optimize your designs is that you get a definitive answer as to which works better. The only thing you have to invest is the time spent developing the variations.
It took less than a day of A/B testing against the original design to show that this new version dramatically outperformed it. It's a good job I didn't listen to myself.
That in a nutshell is the power of testing. Trying different ideas to simplify the purpose of the page, trying things you don't think will work, all to make sure the page's purpose is aligned with what the customers want.
Do a lot of this.
Oh and if you are impressed with what they did go sign up for a free trial of Warrior Epic!
Seis: Insights from First Impressions
This is more of a bonus tip, something I have started to leverage recently.
There are a huge number of wonderful sites that have made traditional user centric design principles more accessible. One such site is Fivesecondtest.
The idea is simple.
You have a page. You are not sure how well it works. Or you are sure the page is God's gift to humanity, yet no one else seems to believe you (especially not your website Visitors :).
Take a image of your page. Go to fivesecondtest.com and upload it. Send the resulting link to people (tweet it or send it to your friends or send it to a focus group / panel you have or send it to your co-workers or post it on a forum where your future customers might exist).
Here is the process they'll go through:
At the end of the process you get a sweet report that lists the first impressions of people who looked at your pages.
For free!
Fivesecondtest will not be your be all and end all for improving your web pages, but for me it has been a great source of very rapid feedback that then provokes discussions with the Usability, Design, UI, IA, Analytics teams.
The output will be ideas for tests, dumb things we should stop doing, key calls to action that no one notices that we should make more prominent etc etc.
The Epilogue
Six simple ideas that you can execute on tomorrow, most for free and just one that costs any money.
There is one cost that few people are willing to bear, the cost of actually doing this analytical / listening work.
I hate to end on this note but many of us believe that just by implementing Omniture or WebTrends or a tool we'll have God herself whisper the insights to us. We tend to be frustrated when tables and rows don't scream out things we should fix.
It is very difficult for many of us to close the gap between webpage purpose and customer intent because we are unwilling or unable to put in the blood, sweat and tears required.
I hope you'll choose otherwise. Ideas are cheap. Action is not.
Good luck!
Your turn now. Care to share your ideas on how you improve your bouncy / low conversion pages? How do you close the purpose vs intent gap? What was the last thing you did that had a dramatic impact?
Please share.
PS:
Couple other related posts you might find interesting:
Via 
Good damn! This is one of the best articles i have readet in long time, thanks for sharing this!
Hi Avinash,
I really enjoy Fivesecondtest. It's so simple. I have created tests with my A/B testing landing pages. I already have some interesting results to exploit. The most difficult and the longest part in my job is to convince teams in my society. Especially to listen to the voice of users. Your blog helps me decreasing the time spent in convincing. More time for fun …
Best,
Morgan
Good post as usual Avinash, but for once I disagree with one of your comments.
The post click segmentation Dell does helps it further narrow down products that are appropriate to user. Whilst I agree the page could have been better targeted towards "colour laserjet printers", you can't infer what the purpose of that printer will be from the search query alone.
It's arguably better to further segment visitors and then present more targeted, appropriate printer options i.e. printers suitable for home use versus printers suitable for business use. Each of those markets have very different needs and respond to very different selling points – upfront segmentation has the potential to pay dividends in terms of conversion.
Whilst I agree that Dell haven’t done a great job of implementation and I concur with your point that they make you worry over price etc. I don’t agree that you should always point broad search queries to narrow product pages without further qualification.
Heavy post. You have given me a lot to do but I know it will be productive.
Thank you!
Love the Crutchfield bit – Marketers need to realize that "conversion" is a process that *starts* with the web site. First you convert the sale, then you have to convert the customer!
This is something Crutchfield (and many other companies with mail order roots) are quite good at because they have decades of test results, and Marketers should pay attention for clues. Analysts can help by understanding which campaings or products lead to higher 2nd purchase rate.
There is a reason for the "bonus stickers and custom helpful manuals that are in the box" – reducing 2nd purchase bounce rate, and there is tremendous financial leverage (ROI) in reducing 1x buyers from 80% to 70% – just like with page bounce rate.
Excellant post Avinash , the thing is : many webmasters are not defining the web page purpose for each page inside the website , and another point is that webmaster should trate visitors as a customers not cookies as you explained in your book .
Now I think everyone should read this post in parts.
example : before moving into this part
"Uno: Insights from Sources (URL’s, Keywords, Campaigns)"
Everyone should finish part one which is defining the web-page purpose and customer Intent ..
and should be able to implement this on one of the pages he is dealing with *. (*If you are a web analyst and working on a website nowadays)
btw: I will write another final comment after I finish all parts :)
Thanks Aviniah for making the web analytics better and better
You convinced me to do some A/B testing, now you do not need to convince me anymore :)
I got a 50% increase of my conversion rate within 3 days…
http://www.ceondo.com/ecte/2009/08/ab-testing-boost-conversion
Really, split testing is something that should be implemented as a default in all the CMS.
Thanks a lot!
This is one of the most excellent posts I have read in a long time. Ad-to-Landing Page connection is the #1 thing I (try) to harp about for bounce rate / conversion rate / transaction rate optimization.
I really like this blog post because of the use of things like fivesecondtest and Site Overlay / Heat Map, as well as real website examples that in the context of this post, can really benefit those site owners! I'm also still a big believer in Site Overlay, even though some of my colleagues / co-workers are not drinking the kool-aid anymore.
I would just like to capitalize on the point that Avinash makes about installing a program and hoping that actions fall out of the sky into your lap. Not only do you have to put your hard hat on and roll up your sleeves to find insights, you also have to sell it to your customers / prospects / boss(es). Make presentations, highlight important data, use arrows, circles, scales, colors, bold, italics, and underlines. Take it seriously. Imagine that you are in a court-room and you are defense council, presenting your closing argument to a 12-person Jury who will decide whether your client goes free or goes to jail for 25 years. You HAVE to convince all 12 Jury members, or your client goes to jail. For your clients / boss(es), you HAVE to convince them that controlled A/B or Multivariate Experiments + Data Mining are (business) life or jail for your client's or company's web site.
Thank You!
After reading this post I felt a sudden urge to do something (as happends a lot when reading your posts).
I wanted to take a look at clicktracks. Seems it is part of something else now. While I was under the impression from your post that clicktracks was free and decent.
Bummer. Or did I misread you ?
but that won't hold me back, I will try some of your suggestions.
Same as above comment – "heavy post"! Read it twice now and tomorrow going to sit down with our design team and come up with the next iteration of our site, thanks for the tips!
Great post as usual. We're in the middle of redesigning our own website and your blog has been a valuable resource when working on our content strategy or how we should be analyzing our years of analytics data.
I had a question about the first diagram depicting Customer Intent and Webpage purpose. Does "Your Intent" have its own circle that should be overlapping or would you consider that to be a part of "Webpage purpose"? If your intent is its own circle, should we be concerned if only "Webpage purpose" + "Customer Intent" overlap? Or, should we just relish in the success and try to create another page that matched what we truly intended the webpage to be?
Avinash,
Agreed–another great post. Read the blog, ran to CrazyEgg and started an account. I'm using SiteSpect to insert the javascript for the test so I didn't have to enter an IT ticket to update the site. From reading the blog to collecting heatmap data took 10 minutes. Gotta love the web!
More importantly, am already learning from the heatmap data (it's nice having the refer and keyword dimensions available). Great complimentary information to all the rich data we're getting from the Omniture Suite. Wow, this sounds like I'm a shill for three different organizations here (wait, four counting Occam's Razor)–not intended.
Cheers
Absolutely outstanding analysis, Avinash. WTG!
It seems that the same % bounce rate could be deemed high for some sites, and acceptable for others. Is there research on this anywhere?
Fantastic article and the best I have read (in entirety!) in a long time. Thank you Avinash, I really look forward to your posts. As Nike sez…Just Do It!
Avinash, great post and thanks for the mention. Next steps for our landing page are:
1) Test a version that incorporates the account signup form, rather than send you off to another page.
2) Try out some different call-to-action phrases – "Create Account", "Sign Up", "Join Now", etc.
3) Work out how we're going to fit this in with the rest of our workload! ;-)
Avinash,
I really liked all your insights but even more liked the fact that I became familiar to the tools that just now mentioned on Summit for Success in Burlington!
Thank you
How do you do it over and over again! Great article. A lot of the tools you are loving we also using and as you say they work verry well. Recently we got 20% more quotes just by changing the call to action button. The conversion (sales/quotes) did not change so absolute we got more sales!
By one of our customers we used Crazy Egg we saw that the visitors love the search option an tagcloud. But we also saw that more than 50% did not pass through the homepage. But this was the only place were they offer the functionality……Now they offer it on all the pages…and yes customers are bouncing less hard.
You have to love testing and the visitors who make our job every day a little box of chocolats…you never know what you gone get!
Thanks for sharing and giving me inspiration!
Awesome post – great tips and you managed to avoid saying MVT or multivariate even once! :)
At our company, we use GA, CrazyEgg, Test&Target, SurveyMonkey, fivesecondtest.com, ClickTale and usertesting.com. (No, we're really not vendor-crazy, I wish somebody could package some more of this stuff up nicely!) The combination of quantitative and qualitative data is unbelievably powerful. The key is to come up with great hypotheses that you're willing to be proven wrong about. And listen, listen, listen to both the explicit and implicit customer feedback.
Steven: I welcome your disagreement, thank you.
My feedback to dell came from realizing that no matter what button I choose (home or small biz) on the resulting website 98% of the inventory (be it printers, computers or anything else) was pretty much the same. Hence my recommendation that perhaps they can satisfy the customer first and then ask the customer to choose the division of dell they want to do business with.
I don't work at Dell, so perhaps they have tried other things and only their current approach works.
Jim: Awesome, I love this: "Reducing second purchase bounce rate."
How rare for us to think about second purchase, and yet it is so critical.
Djemmers: You are right, clicktracks is not available by itself any more. Its existing users, like me, can continue using it though.
Brian: Yes in the picture "your intent" (website owner's) is synonymous with "webpage purpose". Ideally what you want the webpage to do is deliver what your customer wants from your website.
I added Crazyegg as an alternative, and it does what it does well. :)
Geno: There are no established benchmarks, and one important reason is that they will be different for each sector and site purpose. In tools like Google Analytics you get benchmarks for your industry vertical for websites whose size is similar to yours.
In general though it is really hard to get a bounce rate under 20%, anything over 35% is cause for concern, 50% (above) is worrying for me.
There are rare exceptions, for example for blogs I don't look at overall bounce rate, just the bounce rate for the New Visitors segment.
Everyone: Thanks for all the comments and kind words!!
Avinash.
very quality tips and information relating to high bounce rate….I do experience this situation most of the time..but now I know the solution..thanks..
I am a member of too many blogs and forums. As a consequence it takes me two hours a day to read them all!!
That said, very insightful pieces like this makes it all worthwhile.
Thanks, Avinash. I didn't know that Google Analytics provides benchmark data like this for individual industry verticals. Need to dig that part out, and study it. Thanks again. Keep up the good work!
Hi Avinash,
The most annoying thing I noticed on many websites is the flash lay overs, where customer has to complete the form or give feed back in order to view the page. There is no option to close the popup to view the website!! How horrible it is!!
Some of these include subscription to their offers or simply you have to register to view their site. Is n't it wise to let them browse the site first and get feed back or offer to register
at the end of their visit or after couple of page views? At least from the Google analytics it's easy to figure out the pages with high bounce rates and stop displaying the flash layover for couple of days to see the difference?
As someone mentioned about #Dell I would like to share my personal experience,
@Dell, what I have noticed is sometimes the offers are not available in a particular Geo-location but still they run the display Ads on Yahoo Home page in that location. This was happened couple of months ago when I wanted to buy a new PC.The landing page was simply something like "OOPs!! The offer is not available in your region!!" in the content section. At least they should have showed me the alternatives!
Well I bought a Acer PC at a much lower price & I love it!!
Avinash, can I first of all thank you for introducing Fivesecondtest in your post in section Seis: Insights from First Impressions.
As you say there will be other apps like this available, but I'm happy to spend time using this and helping make more informed user interface decisions, rather than looking for alternatives!
In fact, having posted a comment about this fab application on my twitter account earlier, and then having a 'usability consultant' question the whole idea of gaining first impression insights, I decided to post an article on Econsultancy titled 'Take five seconds to test your call to action and USP'.
Your comment of 'Fivesecondtest will not be your be all and end all for improving your web pages, but for me it has been a great source of very rapid feedback that then provokes discussions with the Usability, Design, UI, IA, Analytics teams' is absolutely spot on and I agree wholeheartedly.
On a seperate note but related to a current bounce rate challenge we have with a client site, they are a translation services companies who over 40 lanuguage specific landing pages – great for SEO and traffic generation, but when their core target customers are UK based businesses, getting traffic from all corners of the earth into these landing pages certainly isn't delivering helping to keep the bounce rate low and actual conversion rate up.
We are just about to start testing alternative versions of these landing pages and pretty much your 6 point blueprint on this post isn't a bad reflection of the areas we will be focusing on!
Best thing I read on the internet all day.
We used Crazy Egg to test a home page recently for our client and found that 90% of the users were clicking on just three items. The company had us re-design the home page to make these three items prominent on the page, and re-organize everything else so that they were below the fold. It worked very well for showing the mindset of the user. This client has us run A/B tests on all of the pages where they are trying to attract conversions, and we have put Crazy Egg on all of the A/B tests going forward. It even allowed them to segment their branded-term ppc traffic to a landing page that is now different (also redesigned recently) from their keyword-specific but non-branded-term landing pages. It was a great success.
Geno: To get benchmarking data in Google Analytics you would click on Visitors and then on Benchmarking. Here's a screenshot:
The first red arrow shows the menu item.
The second red arrow to open the industry, vertical categories and sub categories and choose the right one for him.
In the graph itself the blue line is your performance and the grey line is the benchmark.
-Avinash.
Thanks for a great article! Nice to see some comparison of alternatives to Google Analytics.
Totally agree regarding Dell, as a professional programmer we often see the same when designing solutions. Our company often see things from their perspective instead of the customers perspective. Budgets and who pays for what is not always the best way to design a website.
Kenneth
Great stuff! Thank you, Avinash.
Simply superb! You highlight what everyone commonly ignores and that's what I like the most about the content your write. Thanks for sharing your knowledge with all of us. In the past I have followed some of your suggestion and seen immediate results. I am going to logon to Analytics now and start looking into the bounce rate and modify my content accordingly to show my visitors what they came to my page for.
Thanks a million!
Now this is something every marketing guy should read. As should every web developer. And so should the usability guys.
Heck, everyone who has anything to do with the web should read this. Brilliant stuff.
My richest source of non-analytics data has been internal site search data. It is the direct feed of 'what your users could not find'. I have always loved it!
Ace post buddy ;)
Great graphs, a picture is worth a million words. I'm about to bounce
This article was awesome, very insightful. This will definitely help in my web development future. Just one note of criticism, there are a few spelling and grammatical errors on the page, may want to fix those.
Where is the Content Detail Menu as shown in your first Analytics screen capture and why can't I find it??
Connie: Just go to the report of any individual page and you'll find it (the clue is in the second screen shot, right under the one you are referencing).
Here's a detailed path:
Overview > Top Content > Click on any page and on this report you'll see in bold how many times the page was seen and right underneath it you'll see Analyze > Click on drop down.
Oh and to the right you'll Segment and you can click on the drop down for segmentation options.
Andrew: Thank you, I will go through the post again and look for errors.
I only have time to write late in the night and the just put it out there in the time I have. Not the best strategy for optimal prose. I need to work on that.
Avinash.
This is great! An article with all the words and ideas I needed for my upcoming meeting.
One of the challenge I had was how to explain the gap between site visitors' intent and web pages.
The site I'm working on has many product detail pages that is used in various marketing channel, and various referring traffic sources bringing in customers with two to three different kinds of customer intent. First, visitors coming to landing page and routing to product support section. Second, visitors will exit or go to non-shopping cart pages. Finally, few % of users who go to shopping cart to purchase.
The challenge is to show the HiPPO's the reality and take actions. What I am thinking…
We can either take advantage of paid ads to drive targeted traffic, and "potentially" optimize the landing page. Other option is to revamp the page from SEO perspective, by suggesting to split the page into two; leaving current version as is, and creating another page making it relevant for support or eCommerce.
Basically, one other challenge is, you have a site catered to support and eCommerce, with two different major customer intent for support and to buy.
Do you have any other ideas, based on your experience that you can share?
Kris: The primary emphasis of your search efforts should be on SEO, making sure that the site is well constructed, the urls are nicely created, the site is crawl able. One other important thing will be to make sure keywords and content on the page is optimal.
For PPC I think the great thing is that you control the landing page, the search engine does not.
My strategy is to use the normal (SEO) product pages if the calls to action will match the keyword intent. For example if my keyword is "Buy Lenovo Thinkpad X301" and my product page is a ecommerce page then I'll just send the PPC traffic there. Match of keyword intent and webpage purpose.
But if it turns out that the keyword I am buying is "compare lite laptops" then I might not send the traffic to my normal page, rather I would construct a custom landing page to match the keyword intent and landing page purpose.
Hope this helps a bit.
-Avinash.
Great stuff Avinash — it took me a little while to digest & assimilate :-). Your post goes to prove that using a 'rifle approach' to web analytics (where in you put your energy and focus on improving one or a couple of metrics, in this case bounce rate) can more often than not, provide you with a much better return than using a 'shotgun approach' (where you are trying to work on a broad array of metrics — and might end up optimizing nothing).
I have been reading your blog for quite sometime, the articles are really well written and well explained. I have become a huge fan of Avinash Kaushik. This article is a gem.
Great post, although your use of the "Fear This" font in the venn diagrams does seem a bit scary. :-)
I think that an increasing focus on the keyword mix of pages that do have sufficient kw data needs to happen, like you're talking about. We spend all of this time trying to figure out what people want out of our web site, and here we have a whole slice of the pie doing just that, to a certain extent. Maybe we want to pay attention?
Evan: It's not "Fear This"!! I would not do that to my readers. : )
its a font called : Kristen ITC
I use it sparingly, but I do thing it has a unique fun feel.
Its here:
http://new.myfonts.com/fonts/agfa/kristen-itc/
I concur with you, that we do need to pay attention a lot more to what the keyword data is telling us when we focus on specific "bad" pages.
-Avinash.
One thing that's a real "gotcha" for bounce rate is how can you tell the difference between users who bounced because they didn't like it and users who came to read the latest post through an RSS feed or other means, then left, because they've read your previous posts? The latter would be normal and expected behavior.
@connie
Well, you definitely won't have any quantitative data to determine this (unless you're using a pop up survey to ask them or have a camera installed in their home that can read thoughts :-) ), but I would say that you'd have to make a common sense judgment based on the type of content the user is landing on and where they came from. In your case, if your blog post doesn't have any clear "Also read this" links or a call to action on the page, I would expect bounce rate to be astronomically high.
Try putting things like these calls to action or "see also's" on the page to improve stickiness, and make these things easy to see. Bounce rate may remain high because the user got what they wanted, but it will probably be much lower than if you didn't have these assets in place.
@connie
It also bears mentioning that no metric is a problem solver. Metrics like bounce rate are just watchdogs (VERY good ones) that bark when they see a potential problem. Sure, when your neighbor comes over to borrow a cup of sugar, the watchdog might mistake them for a burglar. But it doesn't make the dog any less valuable.
We're going to have to remember that after we start the engine and put the car into drive, we still have to stay awake and pilot the car. These tools are just that: tools. No metric truly has a "gotcha", just like a price to earnings multiple doesn't have a "gotcha". If you didn't dig deeper to realize that the great P/E telling you to buy 10,000 shares of Enron was a false positive, it's not the metric's fault, it's the buyer's.
Great question!
Great idea about the calls to action and similar articles and such. Perhaps it's obvious, but it wasn't to me =) Thanks!
Great post – thanks for this. It's always tough explaining the relationship between marketing and design, but I think you showed this brilliantly.
I'm flat-out embarrassed that it's taken me this long to come read your new post and leave you a comment, Avinash. Life has been BUSY! I've been working non-stop to help our company embrace the power of analytics. We finally just got the green light to put 4Q on our site this week. Within a day we had 120 respondents that barely gave us a C letter grade in satisfaction.
Nevermind HiPPO hyperbole or "I have a cousin whose friend knows a guy that used the site and loves it." We have data now. 120 people resoundingly told us "meh."
Hopefully our HiPPOs will allow us move forward on a large project to correct some of these issues.
Thanks again for this wonderful post and all the advice. I've already pre-ordered your next book, Web Analytics 2.0, from Amazon.com. I'm sure many of your readers will want to do the same. Cheers!
damn good post man. I learned something new about google web analytics today.
Thank you, thank you!
The light bulb just switched on……I am now running keyword phrase tests on all of my domains, and going for the perfect match.
This blog is now my most favorite!
Tim
Well written post. I agree with everything you stated. Bounce rate is a great indicator for a number of metrics. If you struggle with growing traffic, maximize the traffic you currently get by focusing on ways to increase page views per visit, i.e., link to older posts, move recent posts to the forefront.
When I first started my blog, the bounce rate was around 90%, thanks in large part to Digg/StumbleUpon/Mixx traffic sources. However, since focusing on more industry specific audiences and constantly playing around with the layout, the bounce rate is consistently around 12%.
Hey A, first time visitor here but have heard a lot about the blog. Thought it valuable to note one point that may be implied in your post, but worthwhile spelling out:
In my opinion often matching, and perhaps more specifically filtering, customer intent starts before the customer lands on your page. If an online marketer uses SEM or display to drive traffic to their site, the creative being used can set the expectation of what a user intends to see on the associated landing page. For example, for a software company looking to sell it's product for a fee, using FREE in its ad copy will likely drive a lot of traffic to its website, however not result in many conversions as the customer expected a Free product – not to mention the potential damage to that company's brand.
As an online marketer, consideration of customer intent starts with how you plan attract customer intention to drive traffic. You make a great illustration here with your "survey"example in your post.
Keep up the great work and I look forward to reading more.
Wow great post. Lots of great information. Makes me think I have a lot to learn. Thanks for the great tips.
I stumbled upon your site by typing in google looking for best internet marketing book. It took me to one site that had a link to yours. I think it brought me to the right place :) I am happy I found you. Now I have a lot of reading to do here.
Thanks again
Jim
Hello Avinash,
This post about understanding customer's intent versus content was one of the best posts I've read lately – thanks!
I'm trying to understand "bounce rate" when having the same entrance source (search engine, ie. Google), same keyword, same landing page, but different bounce rates when it comes to paid vs. non-paid traffic. Some differences I see come as high as a 20% or even greater. Some keywords have minimal difference on same scenario. But, I'm most concerned with the bigger gaps.
Why would there be such a disconnect in customer's intention with content if the distance apart may be just a few pixels away (sponsored link vs. organic result)? Any thoughts on this would be appreciated.
Saludos,
Nacho
Great and detailed blog post! Made me write a blog post about clarifying the purpose of the website into our project blog. In our project, we analyse website traffic and profile users of travel websites. Our experience on ClickTracks is just brilliant! We also utilise Google Analytics and together these two give us a very good picture about user behavior on site.
Your blog definitely gets into my RSS reader!
Very relevant post, as I am trying to make sense out of the bounce rate on our sites. Have you come across an acceptable benchmark for bounce rate – realize that it will probably depend on site type (informational, transactional, etc.). And, it will probably be unrealistic to expect near 0%!
A rare, useful article on SEO that doesn't repeat the same basics over and over again. Thank-you.
tips-for-improving-high-bounce-low-conversion-web-pages confirms my feeling and thinking expressed in previous comments here and elsewhere: The free value you give away in blog posts is awesome and amazing. You are worth your weight in gold. If ever in Atlanta, would be happy to increase your worth by taking you out for b'fast, lunch or dinner :-)
Cheers,
Beat
Another excellent post avinash. Always can count on some good posts here on this blog.
It's also been a great resource for me to forward to those I train.
thanks again.
awesome post!!! I just discovered your blog yesterday, and it is really the best web analytics blog I have ever seen on the web. Keep it up!
Great post Avinash! We really like using CrazyEgg, http://newsletter.blizzardinternet.com/crazyegg-heatmap/2009/09/15/ it's a great way to visualize user behavior, which IS a goldmine. Along with analytics it can be a dangerous tool.
Just discovered your blog today..So much to read – very insightful. I love your simplistic and easy to comprehend examples
Very well written article & illustrations. Too much emphasis has been put on #'s and stats that don't compare to the best practice of any business: Give customers what they want!
As a company, we use to do only SEO / SEM, but found it frustrating when no one looked at the actual offering!
Thanks for educating the market!
Darrell Keezer
– Owner, CandyboxMarketing.com
Hi Great post thanks for the info. I can never know enough about good conversion pages.
I feel that the best way to target your traffic more effectively would be to improve your SEO strategies relative to what you are actually marketing, which was mentioned briefly mid article. If your landing page is about duck mating calls then make sure that your keywords contain words relative to duck mating calling also your all of your other meta tags and title tags. Just follow the usual guidelines of on-page SEO. Also make sure that your anchor text of your inbound links contain relative keywords as well. Dont try to target something that your site isnt about and then wonder why you have a low conversion rate and high bounce rate, it just goes hand in hand
Hi – Great post. Have you ever found any metrics for how plug-ins and extra downloads impact bounce rate? I'm just trying to benchmark a game we developed against the norms (we have a high bounce rate largely, we suspect, because of a browser pug-in required). The metrics are surprisingly hard to find. Thanks!!
Shel: I am not sure if there is a way to detect how many visits started a session on your site with a missing plugin. Though I suspect that your web server logs have that data and it is a simple matter of grep'ing the right string.
You could also explore features like Custom Variables in Google Analytics (http://goo.gl/NjLV) to see if there is a clever way for you to collect the data that someone was missing a plugin. Work with a GAAC if you need help, a list is here: http://www.bit.ly/gaac
If the download link is yours, it is pretty easy to encode that and capture the clicks on it with any web analytics tool.
-Avinash.
Always good to see an old thread spring back to life!
Shel: one thing to also remember – you're not going to be able to use a virtual page view or an event to signal whether a plugin is installed or not (although you could modify the PV name) because that will create another server call and nullify bounce (you will get either a page view + event or a double page view, which are both considered non-bounces by analytics). A lot of people forget this and then they end up with super-low bounce rates (or zero!). A custom variable might be the best solution for sure.
Good luck! Have you posted this question at the GA forums for help, too?
http://www.google.com/support/forum/p/Google+Analytics/?hl=en
This is an awesome post about SEO and on page modification.
I think your post is one of the best I have ever read on the subject of increasing conversion etc.
Another thing we did that had a 16% increase in our conversion rate was to add a Video Trust seal from shopperguard.com to our checkout.
This is such an informative article! This is one of the best resource that I've come across, the images are such a great addition!
Thank you for this, this would really help me out in minimizing my blog bounce rate!
Great advice about Google Analytics :)
Great post, all six suggestions are good ideas for improving a website. I particularly like the The Five Second Test. With so many options on the web, many visitors aren
Really helpful article Avinash – thank you.
Do you have any thoughts on creating custom landing pages to closely match particular adwords ads?
ie you write an ad that picks on a slightly quirky keyword group and you have a (possibly temporary) landing page which doesn't perfectly fit with the existing structure of your site – just to match the customer intent of people reaching you via that one ad?
Mark: It is always a best practice to try and create relevant landing pages for your keywords, especially for paid search and most especially for your head (and even big "torso") terms.
The only time this becomes hard is for the long tail. For those long tail keywords an optimal path is to create logical clusters based on relationships and intent and then create a "mostly customized" landing page for each cluster.
Either way it is a best practice to do all you can to improve relevance.
Avinash.
Mark:
Also remember, paid search is traffic rental. Even if the landing pages don't fit into your "regular" site hierarchy, find a way to fit them in to your site map and/or make relevant content "hubs" for related search phrases to build relevance in the natural search results. Try to not create these pages as islands, as much as possible.
As Avinash said, you want to build as much relevance as possible, so remember to "network" your relevance together so you can do well and be responsible to SEO, so you're not just renting, but paying the mortgage too! Build content relevant to specific search phrases, then build content up the hierarchy for what these phrases/pages have in common, and up and up until, voila! You have a new site section, rich with content!
This is very informative great examples, and well documented.
Thank you for sharing.
Mike
Hi,
Great article with some really great tips. It's so difficult to see clearly when you work for yourself and are struggling to make ends meet while you establish yourself. You end up sometimes just trying to tweak everything and chucking everything into a page to make it appeal to everyone and so appealing to very few.
This is a straightforward, practical guide that I'm sure will help me no end.
Thanks again
Andy
I never realized how I could go through every page I had and see exactly what the visitor could see. I definitely relate to how I must make it easily determinable what the visitor is there for. On my reviews pages I went back and made much bigger product pictures and more informational data with headings and my bounce rate instantly is dropping in only days!
Great insight into bounce rate issues. I had not thought about readers coming in from RSS feed and just reading that one post.
"Experiment or Die" – good motto! I find myself unwilling to make many changes…although I don't really have much to lose, and tons to gain.
Excellent advices to increase the time each visitor will remain on the website.
Sometimes it's good to have links to key content the user is looking for right at the top (An analysis of the keywords for the page will show the user intent).
A home page may have quite a few keyword matches and showing the top 5 relevant pages may be a good idea.
Personally, I've see significant conversions and lower bounces after doing this.
Great tips, all!
But how are "A/B" landing pages tests set up? Do you have any tips on how that's done?
Dee: If you are using Google Website Optimizer this is perhaps the best guide you can fine, it is easy to read and act on.
~ The Techie Guide to Google Website Optimizer
If you just want an implementation guide then this should work:
~ Website Optimizer Installation Guide
A specific guide to A/B tests is here:
~ Website Optimiser Installation Guide – A/B
There is also a helpful Advanced A/B Testing guide.
If you are using a different tool than GWO then please reach out to the vendor, I am positive they will have lots of helpful resources.
Avinash.
Collecting user analytics and customer feedback is very important for management of low conversion web pages. Most sites do not have proper SEO strategies and unnecessarily focus on irrelevant keywords, and then they wonder what's wrong with it.
Thanks for the article, btw. Very informative and nicely structured.
We have a very low bounce rate about 16% but also a very low conversion rate, like 2000 clicks a day and 1 conversion. I searched the web for low bounce, low conversion but there's nothing out there
We are like some kind of Ferrari car. Lots of lookers but very few buyers.
I am looking for ways to increase conversions. Our prices are the best in the business.
Peter
Peter: If the bounce rate is so low (16% is not too low, but low) then I get this nagging feeling that the analytics implementation might not be right. It is not uncommon for people to have javascript tags twice on some pages, or to have the tag and also fire events when the person first lands or other such things that cause the bounce rate to appear lower than it is.
If possible I would encourage you to check that out first. If nothing else you'll audit your tags and that is always a good thing to do every once in a while.
In terms of a technique to understand lower conversions… for me the path is no different than what's outlined in this post. The same things will help you understand why people don't convert. Though I would start with #3 in this post. Then go to #6.
Also if you send order confirmation emails to those who buy, include a link to a very short one or two question survey to understand why they are buying. That will be very insightful.
Avinash.
Hello Avinash,
First of all, thanks for your great advices and guidance.
I have one question regarding conversion rates/ goals. I'll start with a short intro and then with the my stupid question (:
I run a subscription service (streaming of local sitcoms) I'm easily tracking my goal conversions (paid subscriptions) through my tagged campaigns (got the tip via your book) Basically my campaign advertize specific video content/sitcoms, so I know which content converts my tagged campaign visitors to paid subscribers
But I still have some traffic for which I don't know the reason/ what content converted the visitors to paid subscribers..
So, I just wonder if my method for finding out what video content made the visitors convert into (from the untagged campaigns) subscribers is appropriate
This is my method, I'm looking at the url of the exit page (last thing new visitor watched) for vistiors who came via google / organic or direct traffic and if there is indication of conversion goal completed (e.g paid subscription) right next to the "exit page" url (let's say movies/back_to_the_future), I decide that the visitor converted because of that content (in this example, because of "back_to_the_future")
Any info, articles and tiny tips would be greatly appreciated :))
Have a nice day,
Martin
I'm really confused. Your article was brilliant – but I have an Arsenal blog (gunnerstown .com) averaging 300k visitors per month, and a terrible bounce rate. I'm not sure how to apply your tips, since I'm not trying to sell something…
We deliver quality content – most of our pages are accessed directly, from news aggregators or social media links. Not much search engine generated… There is equal balance between returning and new visitors…
But advertisers point to the BR and we lose a ton of deals.
I'm at a loss… Any tips for bloggers? (As opposed to commercial enterprises?)
Thanks for a great article. Respect. :)
Batmandela: Here's the question you have to answer: "How is it that Gunnerstown making money/survive as a business or non-profit?"
The answer holds the key to if you should care about bounce rate or not.
If people just come to the site and read one page and see only one page of ads, you don't make as much money. So it is better to have a lower bounce rate because people will read more content, they will see more ads, you will make more money.
Or maybe you make money when people come to your site and sign up for the email program. Great. If I come to the site, and sign up for a email program by clicking on the box on your right nav, that is not a bounce.
So on and so forth.
If for your business simply me coming to the site and not doing *anything* and just leaving causes you to make money, you should not worry about bounce rate.
For a blog that is one other thing I recommend. Start by looking at reducing bounce rate for New Visitors because they've not built any relationship with you. Returning Visitors can read the latest content and leave and that is ok. (Though even for the Returning you want them to submit a comment, or share your post on twitter/facebook/google+. If they do either of those, GA will not consider that visit to be a bounce, if you have implemented GA right.)
Avinash.
Earlier my bounce rate was always 75%+ , after applying your methods my bounce rate decreased to 60%. I hope it will decrease more.
Thanks for the great post it helps a lot to me.
Great insight into bounce rate issues. I was missing so many.
Thanks sir.
Hi Avinash,
Thanks for sharing the tips on improving web pages, your articles was awesome and I found it very helpful.
Looking forward for more great tips.
Melinda.
hello
I have read the post, interesting – good things to learn.
towards the start of the post , you mentioned about checking the keywords that led the customer come to web page – i am mentioning about organic search. But now that is not there. Could you please guide.
Harish: Indeed due to Secure Search Google, Bing, Yahoo! and other search engines are not providing all the organic search keywords.
I'd written a post with how we can do keyword analysis now. Please check it out here:
~ Search: Not Provided: What Remains, Keyword Data Options, the Future
Avinash.
Avinash, this article was written more than 10 years ago, but the techniques described here still work.
Thank you for your article!