August 2006


31 Aug 2006 12:42 am

montereyEricB’s recent post on his excellent blog Inside Analytics was on the topic of Analytics and Results. It covered two things (Eric correct me if I misstate anything):

1) Concern in the blogosphere and elsewhere around potential for misuse by Google of data they have from AdWords and Google Analytics and

2) Need for defining a new industry standard for measuring “conversions” or “actions”.

I started to write a comment on his post around the first point but it got so long that I thought it might be better off as a blog post.

Context and "Back Story":

There seems to be some concern in the industry, newspapers and blogs around the potential for “mischief” by Google if you do Pay Per Click (PPC) campaigns as well as use Google Analytics to report on that data. The core premise being that since Google will have access to both your ad spend and return on that ad spend (hence value) they can manipulate the market to make more money (because they know how much you  are making from that 50 cent ad click). Or variations of this theme.

Exploring an Alternative Point of View:

The concern is coming from different quarters, from Google’s current competitors, from Google’s perceived future competitors, from general folks who seem to be genuinely worried. 

If Google wants to be profitable and they have access to your spend and, say for ecommerce,“order size”, it is a simple small step to go manipulate market / customers to make even more revenue from Search Engine Marketing (SEM) campaigns. Is it genuine? Are there any flaws in the thinking behind the opinions?

There are some significantly smarter people talking up these concerns surrounding Google AdWords and Google Analytics, but I am a bit mystified that some obvious points have been overlooked. At the expense of sounding naive here are three thoughts for you to critique (in no apparent order):

1) The mountain is very high (and is it really worth it?) :

Most of you who read this blog, and I, actually work with Analytics tools and I think we can safely say that this is a non-trivial challenge to overcome even when we use dedicated analytics tools and just for one company. Think integration of WebTrends and Omniture and ClickTracks etc with our PPC data and getting even half decent insights fast enough. It is really complicated, no matter how you look at it.

If we can't even do it well for one company, ours, for Google to do it for the rest of the world and all the different business models and acquisition strategies is a non-trivial challenge.  Even though there are examples of extremely efficient arbitrage, specifically in the foreign currency markets, I am skeptical that it is “easy” for Google to find "insights" and do so in a timely manner to over charge us and do so while balancing that against real ROI Google would get for itself.

Now I'll admit that if Google's new mantra is "Only Do Evil", they most definitely have the brightest collection of people to actually pull this off. But with all the eyes and hard core scrutiny on Google how would this story look on the front page of New York Times, and what would be the impact on the company’s future?

2) Single acquisition channel vs a world of many :

All the concerns raised assume that Google is going to be the only acquisition strategy option for all companies out there. This is far from certain. PPC is important but for most Fortune 1000 companies it is hardly all their spending.  In as much PPC competes against all other acquisition strategies that we have at our disposal (direct marketing, email, affiliates, SEO, bill boards, crazy tv ads etc etc).

If I run a business and suddenly I see that Google PPC is not as cost effective for me as an acquisition channel,  we will pour money into other channels that get us better ROI. 

The Internet is beautiful and in the past switching acquisition strategies was complex, expensive and hard to pull quickly, now it takes me 10 minutes to switch strategies (don’t ya love the web!).

3) “Efficient Markets” and helping oneself vs customers :

I believe in efficient markets and the power of markets to constantly find optimal ways for companies to reach customers.

Google is not a monopoly, it is not the only option out there, there are larger companies out there trying to compete with Google (though some are in a process of getting their act together :) and Google will have to compete with all of them to be a part of my acquisition strategy.

If they change their mantra to Only Do Evil and they mobilize 5,000 PhD's to build artificial intelligence algorithms to mine my goal page data in Google Analytics to charge me more via AdWords , I'll simply shift to Yahoo! or MSN or a company that I don't know about today but is sure to come along.

Google is not omnipresent and it can't rest on its laurels or bank its existence on a strategy to "rip off" its customers (us). To stay in business for a long time (and at $380 per share for stock) they are going to have to compete for my marketing dollars with all other options I have. For that to happen their strategy will not be "how to make the most amount of money From Avinash" but rather “how to make the most amount of money For Avinash at the lowest cost for him". That will separate them from competitors, that is the difference, for any company, between a short-term strategy and a long term strategy.

A metaphor:

I recently gave a speech at Google at the nice Googleplex in Mountain View in Building 40. The reason I am being specific about it is because I spent almost four and half years working in those exact buildings when they were owned by SGI (Silicon Graphics Inc). 

When I joined SGI nine odd years ago there was a Business Week front page cover titled “coolest technology company on earth” (I am paraphrasing). Today SGI is almost bankrupt (and it truly breaks my heart because I loved that company).

As I walked in B40 for my Google speech it struck me that there could be a story in Business Week today with the exact same title and it would rightly apply to Google and they are in the exact same building.


If Google wants to avoid the fate of SGI (and it took SGI less than three years to go from the “coolest company” magazine cover to start layoffs), it has got to constantly keep innovating, stay relevant, fight in a dog eat dog world and most importantly maintain 1) a persistent focus on having the best damn customer experience for all that Google offers, search and beyond, and 2) be the best, and most competitive ROI, customer acquisition channel for companies.

The fastest way to a SGI like fate is to do the kinds of things the GA + AdWords concerns suggest, and I think Google is smarter than that.

Small but important point: 

Does this mean you should use Google Analytics? That you should link GA with AdWords? Or that you should not? All these are questions that you'll answer for yourself and do what makes best sense for your individual company.

Full Disclosure :

  • I think Google is a very innovative company and I have a favorable opinion of the company
  • I know three people who work at Google, I think they are nice (this could cloud my judgment)
  • I have given a speech at Google and they gave me a free drink (Odwalla Mango Tango)
  • I have absolutely positively no insights into Google’s strategy and I have no inside information into Google Analytics or AdWords

So what do you think? Is this a very naive view of the world? Do you think Google can “crush” all competition and become the “king of the world” forever? Is there something obvious that I should have considered that I am missing? Please share your feedback via comments (and I have a thick skin so please be as brutal as you have to be).

(Eric I am sure you’ll agree this was a really long comment on your post! Thanks for posting something that stretched the brain. : )).

[Like this post? For more posts like this please click here.]

PS: Hello from the beautiful and historic Hotel Del Coronado in Coronado, California. Stock photo below.

Del_coronado

28 Aug 2006 12:28 am

curvesHow many metrics can you call adorable?  Site abandonment rate is an adorable metric, to me : ), for these reasons:

  • Money, money, money baby. IMHO there isn't a metric out there that can tell you a lot so quickly and any improvement you make to it will directly and immediately impact the bottom-line.
  • It measures the customer interaction in a very small number of web pages, probably one for cart and two to three for checkout. How many metrics out there can compete with that? Few pages, awesome optimization potential.
  • Win-Win: By the time visitors fall into the influence of this metric they want to give you cash and you want that cash. How could that not be simply the best thing in the world?

Inspite of its absolutely adorable qualities it is surprising that this metric does not get as much prominence as, say, conversion rate (stop obsessing postbest practices post). People simply don’t seem to walk around with Abandonment Rate on top of their dashboards and consultants don’t center their pitches around helping improve this metric. My hypothesis is that

  1. we don’t understand the true power of this metric 
  2. we don’t think we can improve the four odd pages that this metric covers 
  3. there is some confusion around around a standard metric definition and that does not help 
  4. usually this number is really “high” (YMMV) and we are embarrassed to put it in front of senior management
  5. there is no hype around this metric (hopefully this blog post will fix that! : ))

(While it is absolutely possible to apply this metric to a lead capture process, or other websites processes, for reasons of absolute clarity and maximum impact this post will focus purely on the metric in the ecommerce world.)

Metric Definition:

Site Abandonment Rate (in percent terms) = [1 – (the total orders placed on the website divided by total add to cart clicks)].

In english it is the number of people who intended to buy, by clicking on the add to cart button, to those who actually made it out at the other end, by clicking submit order. If on your website people purchase multiple items in the same session, trends of this metric will accommodate that behavior just fine.

The process that this metric measures on most websites is: click -> add to cart -> click -> start checkout -> create account (or login) -> click -> provide credit card -> click -> review order -> click -> submit order. Hence at its highest level the metric helps you understand how much money is left on the table by your customers.

I recommend this definition of the metric because few people measure the process described above end to end and if we measure it in silos (explained below) we tend to solve in those silos (cart and checkout). That is a problem because the most impactful solutions for improving this metric lie in end to end thinking.

There are no standards for Site Abandonment Rate that I am aware of (if you are please post in comments) but the range most brandished around is that the metric is usually between 50 to 70% abandonment. In my personal experience at different companies I have seen it between 25 to 55%. That should give you some range to think about.

Depending on the cost of items you sell on your website each percent point of abandonment could represent tens of thousands to millions of dollars per month in revenue. Hence my recommendation to have a almost irrational adoration of this metric.

Next Level Segmentation Recommendations:

Once you measure Site Abandonment Rate and know where you fall then it is time to segment. Two logical first level segmentations:

Cart Abandonment Rate (in percent terms) = [1 – (the total number of people who start checkout divided by the total number of add to cart clicks)].

Spheal1In english this is the number of people who were motivated enough by your spiel to make a initial commitment to buy from you. They could do this for any number of reasons (saving the item for easy access, check shipping etc) but it is still a deeper interest than “browsing”.

In the context of your Site Abandonment Rate the Cart Abandonment Rate is very helpful because it will immediately isolate where most of the departures are occurring. Usually this should explain most of your Site Abandonment Rate because of the “relationship” with the visitor is fragile until this point (it is stronger than the "browsers" but still a bit fragile).

Checkout Abandonment Rate (in percent terms) = [1 – (the total number of people who complete checkout divided by the total number of people who start checkout)].

In english this is the number of people who decided to bail at the very last step of a very complex and complicated journey that is the purchasing process.

I have gone out on a limb in the past and have said that this number should be zero percent simply because there is no excuse for your website experience failing to deliver at this most critical of stages. Of course the number will never be zero but should be really close to it, after all can’t we all optimize the three odd pages that make a normal checkout experience to make it a flawless experience?

In context of your Site Abandonment Rate the Checkout Abandonment Rate will be a smaller number than Cart Abandonment, simply because if people make it this far they tend to get through (Important: YMMV).

So now you know your Site Abandonment. You have taken time to understand the distribution of departures between the cart and checkout processes. And sadly it turns our that your site is performing very sub-optimally!! Just kidding. : )

What could you do next?

Some suggestions on moving from simply reporting metrics to finding actionable insights, and taking action: 

  • Segment the Abandonment Rates to really understand where your pain points are. As is always the case you are looking for anomalies in the data and differences in the Abandonment Rate for different segments and causes for those differences. This can be great source of actionable insights. Segmentation examples would include:
    • By “campaigns” (this would include PPC, Pay Per Click, / SEM, Search Engine Marketing, and direct marketing and other campaigns you run)
    • By Referring URLS (who is sending you traffic that might abandon at higher or lower rates)
    • By Products (this is not obvious but you’ll find large differences in rates for different product, or product groups you sell)

  • Love Multivariate Testing: Both the cart and checkout are absolutely perfect for multivariate testing. They are usually “page level” experiences and hence lend themselves to multivariate testing very well. Use your favorite vendor tool to first “modularize” the page then create variations of content for each module and then let the “recipe creation” process create different versions of the page, go relax for a day (or more depending on how much traffic you get) and identify what page works best for your customers.

    Remember there are hundreds of thousands of dollars at stake here so please give this testing process some love and a lot of attention.

  • Challenge Fundamental Beliefs: If your Abandonment Rates are really terrible, step back to challenge some fundamental beliefs at your company. You can test these to prove if they move the right levers for your customers. These would be different for each company but some general guidance on what this could be:
    • If you have “interruptives” in the process (buy this also please or upgrade for $5 etc), try to remove them and see what happens.
    • If you don’t offer anonymous checkout, offer anonymous checkout. If you don’t offer an option to sign up for an account, offer that option.
    • If you have a painful long checkout data input page, see what happens when you don’t insist on asking people to tell you the color of their eyes or some other expendable information.

Two minor best practice tips:

Think of these as things that have either been sore points for me personally or surprises that something so simple is so often overlooked, you can choose the reason you like best : ). 

  • Investigate if there are simple things people are looking, information that is only in the cart or checkout process that could “falsely” get people into that process. Two examples:
    • Shipping Costs: If people just want to know what the shipping cost make it easier for people to find shipping than having to go to checkout or add to cart. Not only will this reduce abandonment, and give you your real abandonment rate, but it also sucks for your customers that you make this painful.
    • Delivery Schedule: How long will it take me to get the product? Do you have it in stock? Let’s not wait until the end to share this nice information.

  • Check that your website is carrying through promotions correctly and reiterates in the cart and checkout process (in bold gigantic letters) the discounts that your customer was promised in the offer or on your affiliate site or your product pages or in your campaigns. It is amazing how many websites don’t do this simple thing well.

I hope you are convinced of the potential of this really simple metric to add to your personal pocketbook, and that of your company. It is really rare to get a metric that focuses on such a small part of the website experience yet holds so much power to help our customers as well as us.

Agree? Disagree? Did I miss something above? Have a tip to share that worked for your company? Please share your feedback via comments.

[Like this post? For more posts like this please click here.]

Next Page »