Avinash Kaushik worked at Intuit as the Director of Web Research & Analytics on their 60+ web properties and is currently an independent consultant working for companies like Google. He is the author of a recently released book called Web Analytics: An Hour A Day.
I recently had the opportunity to talk with Avinash about retailers use of web analytics. Join me in this Q and A session as he explains how retailers can get the most from their analytics.
What is the toughest challenge today that a retailer faces related to web analytics?
Overcoming the perception that web analytics by itself will solve all their data / decision making problems.
Most retailers, I think, “get it” that they need data and they have web analytics tools. The challenge is that they think just torturing that data enough will give them all the insights they need. The reality is that they probably need more than one tool and they need people/skills to be able to leverage those tools.
As an example at the minimum you need to have a web analytics tool, you need a continuous qualitative tool like surveys, and you need multivariate testing tools. You have those and you are cooking.
After you mature a bit you want to atleast get into competitive intelligence (why not benefit from knowing what your competitors are doing!).
In summary: New mindset (“trinity” is what I call it in my book) & People.
What are a few things that retailers should be doing with their analytics efforts that most don’t do today?
I have a 10/90 rule and it simply states that if you had $100 to invest in your analytics efforts then spend $10 on tools and professional services and spend $90 on people who will actually understand your business, analyze the data and find insights.
Lesson one: Consider doing a quick back of the napkin calculation to see how you are spending your $$, if you are 90/10, as is often the case, then you know what the problem is.
It is very common to find retailers where analytics, specifically the pursuit of collecting data, is a afterthought. Only after the pages are released does someone say “hey were is the data”, well it went out with tags, or after campaigns are released it is realized that the tracking parameters were not attached (no data to measure success).
Lesson two: Data capture and measuring success should be integral to your dna and a core part of every process to ensure data is being captured and analyzed. It cannot be a afterthought.
All of us, myself included, are full of ourselves. If you want to have a high conversion rate and make tons of money it is important to get over yourself (and the higher up the chain of command you are the more this is required). Instead shift to things of how you can involve customers in making decisions.
Lesson three: Consider doing surveys, remote usability etc. Absolutely have a A/B or multivariate testing strategy so that you can involve customers in determining how your website should look, what promotions and messaging works etc.
Other than the standard goals of conversion rate or cart abandonment rate, what are the most important goals to focus on in the early phases of analysis?
Getting the highest quality of traffic to your website. Period.
Garbage in, garbage out.
I always recommend that the first thing you want to do is measure quality of traffic (say for example by using that wonderful metric called bounce rate ) and immediately stop spending money on campaigns, affiliates, keywords where you are getting low quality traffic, say with a bounce rate of greater than 30%.
Also look at the top entry points on your website. You’ll be shocked at what you find and you’ll quickly get over the obsession of spending all your energy on creating a golden website home page.
In your book, you’ve got a great chapter on multichannel marketing campaigns. Are there proven tactics for pulling in metrics from non web analytics systems such as video analytics in the store, call center CRM call logs, and mobile web access into a single view of the customer experience?
I think there are really only two little secrets.
1) Primary Keys: 100% of the reason why we can’t do effective multi-channel analysis is the fact that we don’t have the right meta data to connect our disparate pieces of information. So if you are doing multichannel campaigns then please consider investing a bit of extra efforts into creating vanity url’s or unique 800 numbers of links with parameters or one of the many things I mention in my book. If you have the right primary keys (a database term) then nothing can stop you.
2) You can bring data into your web analytics tool, many allow for that, but if you are serious about this consider investing a true data warehouse environment were you will get tons more flexibility with what you can do with your data. You slap a off the shelf BI (business intelligence) tool on top of your data warehouse and you have created yourself a long term strategic data advantage.
Most web analytics research and analysis is around the concepts of improving customer experience and performance of the website. How can retailers utilize analytics in a more pervasive way to improve merchandising decisions, customer returns or other areas of the business?
For merchandising I think multivariate testing is quite literally god’s gift. Your smart Ms. Merchandiser can think of twenty excellent ways to sell yet most of the time 19 ideas are killed and one goes live, and there is a low likelihood that it will work. So throw all twenty into a testing tool, it will automatically create different versions of the pages etc and help you measure which one works best!
I am also a big fan of measuring off-line impact from online behavior. For example the folks from iPerceptions (www.iperceptions.com) were over to have lunch today and they were showing me how for their retail customers they were measuring brand impact and likelihood to visit a store purely as a function of a website visit. So not only can you measure conversion but you can measure if your website is creating the right brand halo and increasing the chance that someone will visit your store (or tell their friends about you).
I am a tiny bit stumped on the customer returns except to say the obvious that the web can help reduce your transaction cost and improve customer experience. For example when my Sony DSC-H9 camera was busted I was thrilled that I did not have to call sony. I could go to their website, fill out a RMA form and ship them the camera to be fixed. Without talking to a human. I am going to buy from sony again (and hopefully my DSC-H9 won’t go bust again!).
Web 2.0 trends such as social networking and RIA technologies like Flex, Ajax, and Silverlight pose new challenges to retailers working with web analytics. What techniques do you recommend for overcoming those issues?
There are only a couple of tools in the market at the moment that allow you to track these technologies in a elegant way. All vendors will tell you: “web 2.0, no biggie, we can do it”. Yet in reality most solutions are hacks.
But development is being done at a rapid pace and I am optimistic that new ways of collecting data efficiently, such a real “event logging” model, are going to be with is in the next so many months. Which means that it will become much easier, with all vendors doing real event logging, to code your rich interactive experiences to give you data and then for you to analyze it.
Exciting times ahead.
What predictions can you make for the 2009 holiday season for what retailers will be doing with web analytics?
2009 is so far away!
If I had to go out on a limb I would say that a lot more web experiences will be “fluid” and retailers will be able to better measure them. There might even be retailers measuring holistically impact of multiple content consumption channels (web, rss, mashups, social media etc) - this is very hard to do today. Finally I suspect that testing and experimentation will be a way of life, and measurement of experimentation will be integrated with web analytics.
So there. Now let’s wait until Dec ‘09 to see how right or wrong I am going to be!Tags: Analytics, research, surveys, tools