Category Archives: English

Business Optimization and Testing Basics

Optimization and testing became buzzwords in the digital world when big technological companies like Amazon, Google, and Netflix shared their experiences and results. Business optimization is not a new concept created by the Internet era, but rather, one that has been a vital part of business for years.

Business optimization is an ongoing process driven by data intended to discover hidden opportunities that can shift or improve your business. If you have some known issues with your current product or a desire to keep shifting your business strategy, optimization and testing can prove valuable tools. Long-term goals, including cost reductions, increases in revenue, or improvements in customer satisfaction and general experience, can be accomplished through optimization and testing.

Unfortunately, when I meet with young digital analysts, one particular issue proves very common: the narrow view of the optimization and testing process only within the context of digital assets they can see. Expensive tools are purchased, but emphasis is given to testing the shape and color of buttons. While content testing is one of the areas we want to work on it, it is only small part of your digital optimization.

Optimization is based on scientific testing techniques that are used to arrive at conclusions when results are not absolute. For example, when a scientist develops a new medicine to cure certain disease, she needs to test her medicine with real patients to evaluate not only the results of the treatment, but its side effects, too. Not every medicine will cure every patient of the same disease because of each individual’s differences. In this case, optimization and testing yields the desired results, as the optimum combination of people will be cured (increasing customer satisfaction), with the fewest side effects (improving customer experience), and the greatest possible revenue will be produced for the firm.

Two examples in the airline industry can help illuminate the utility of optimization as a technique. Airlines have proven to be pioneers in the optimization area, using complex mathematical techniques for years. As an old airliner, I have great sympathy for those who are trying to optimize this process, as it proves very complex and must consider the safety of passengers, the profitability of the airline, and various rigid regulations.

Airline Optimization Example 1: I have X number of planes with different configurations and have Y number of destinations all around the world. I also have an average of Z passengers flying on these routes which can change depending on season, day of week, or by any other operation reasons. How should I assign airplanes to these routes to make sure operations run smoothly while still making a profit?

Low cost carrier Southwest Airlines knew the difficulty of this optimization dilemma, deciding to address it by using only one type of aircraft for all routes, saving money on maintenance costs while simplifying the optimization formula.

When you assign aircraft to routes, you also have to consider the sleeping time of an aircraft. Airlines generate money when their aircraft are in the air, creating a need to minimize the aircraft’s time spent sitting idly in the airport.  “Sleeping” at the airport results in higher airport parking fees, more maintenance costs, and zero revenue.

As the example demonstrates, optimization proves a vital part of an airline’s business strategy, requiring the analysis of large amounts of data to accomplish automated optimizations.

Airline Optimization Example 2: Another airline example revolves around airline pricing models, which frequently result in two passengers paying different rates for the same trip, achieved by using revenue or yield management.  You might feel upset discovering this pricing discrepancy (especially if your neighbor in the seat over paid less than you), but airlines must use this pricing model to keep flying, as they have some of the lowest profit margins in the business world.

Airlines sell seats for an exact route at an exact date and time. When the flight takes off, the airline loses its chance to sell its empty seats, so the product, in a sense, is perishable. Optimization techniques answer the question of how much an airline should charge for this perishable product. An aircraft’s take-off cost will be very similar whether it has 200 or 10 people on board. It will still have same amount of check-in personnel working, a similar amount of fuel used (depending on the weight of the aircraft), and you will have the same amount of in-flight crew employed, as regulated by the FAA depending on aircraft type. In this case, it is in the best interest of the airline to sell as many seats as possible to help cover the fixed costs of the flight. Consider an example where 200 seats are available on a flight and the break-even point would be reached if each passenger is charged an average of $250. Our knowledge of supply and demand would alert us that some passengers would be willing to pay more than $250 for the flight, while others wouldn’t even consider such a high price. Airlines find optimum price points by relying on revenue management systems which continuously check demand. Each fare and rule levels are associated to certain booking classes which are closed and opened depending on time of day and overall demand. For example, if you buy your ticket very early, you are likely to get cheaper tickets, as they were released at the very beginning of the booking process.

These examples demonstrate the vast potential of optimization in business and may help us evaluate digital businesses in a more careful way.

When you are using Google, you are part of an optimization and testing process that helps Google find more relevant search results in the shortest possible amount of time and with the least possible effort. It is in Google’s best financial interest to help you avoid irrelevant results. This motivation helps explain why Google’s AdWords algorithm does not move the advertiser who bids the most to the top of the page, as Google must ensure that the advertisement is relevant and the company behind it has a solid track record.  Google will feed itself more data everyday to make this process better. That’s why older Google AdWords accounts with good track records can bid less than those who started bidding recently, which require Google to gather more information to ensure quality.

I hope this post gave you more ideas about what to test and optimize. Please remember that your content, mobile apps, and websites are vehicles to your end goal. Digitalization changed the business world dramatically, but the fundamental goals of a business remain: generating the greatest possible profit, continually-lower costs, and consistently improving customer experiences.  We have to keep this concept in mind when we perform our digital optimization.


Why Do You Need Switch from Google Analytics to Universal Analytics?

Universal Analytics has launched in 2012 but got out of beta in 2014. I think it is time for everybody to switch to Universal Analytics for many reasons. I’ve always preferred Omniture to Google Analytics for enterprises for many features that did not exist in Google but with Universal Analytics they added so many new features that were part of Omniture for years. It looks like Google is catching up Adobe in analytics and that is very good news for all analytics professionals. Hopefully we will see more innovation on both sides.

Please see my other posts that compares Omniture (Adobe Analytics) with Google Analytics Premium. These are old posts so in the near future some of them might be irrelevant.

Let’s start with our subject:

1.Start doing visitor based analytics rather than visit

Google Analytics is a visit  based tracking system which is now ancient way of reporting. I have started web analytics in 2003 with Web Trends but I started to learn more in  2007 with Omniture. Visitor based analytics was always core of my analysis and this was the biggest reason why I always preferred Omniture to Google Analytics. Why? Cause visit is not an exact metric with an absolute definition. I don’t know what to do with this metrics these days. Let’s see:

Page view: each time page is viewed, this metric is counted. We have an absolute definition and it is also very easy to count this metric exactly.

It is hard to count “Unique visitor” information due to cookies, different browsers but it has an absolute definition. # of people who visits your site at certain time frame.

But visit metrics is so vague, and with mobile native apps, I want to forget this metric. First of all most analytics solutions use 30 minutes as their visit definition. But what about your site’s visit definition? How do you define visit? There are sites which does not expire session for hours, and if you take 30 minutes as your visit definition, what happens?

2-Bring Offline Data into Google Analytics

The Measurement Protocol- another tool that Omniture offered for years (Data Sources) -is now available under Universal Analytics only. This tool lets you to send offline data to Analytics so that you can combine offline data with online. I am strong believer of combining online and offline data to be able to find great insights that might create significant difference in your organization.

3-Data Classification is Finally Possible

This is such a big feature I have been using in Omniture for years. Let’s say you are an e-commerce company and you track product ID or SKU. Products can be classified in many ways. For example a shoe can be classified as one of these: women shoe, size 37, red shoe, high heel, classic, Manolo Blahnik and etc..There is no way we can capture this tons of classifications during a transaction, yes we can theoretically but it would be such a waste of time and resources because that data is already defined somewhere in your internal database. Ideally you would like to feed this information by using common key such as SKU. This feature is so crucial for enterprises. Omniture SAINT and new product Classification Rule Builder has been matched as Dimension Widening. I have not used it yet so I am not sure if there is an automation like in Omniture but to be able to upload Excel data to Universal Analytics for classification is very good. We can’t complain especially if you are using free version.

3-Configure Campaign Expiration Period in Admin

Google Analytics is creating session and visitor data based on cookies. Universal Analytics is using a new method to set visit and visitor data. New method creates session data at the server level. What does this mean? This feature brings so many new features I love in Omniture. There are some admin configurations in Omniture that was done very easily such as defining campaign cookie expiration. In Google Analytics default campaign cookie is 6 months (no it is not 30 days, that is default in Google Adwords) unless you added JS code to the page to alter this. So with Universal Analytics you will not need to add JS code to the page and can modify this value in admin section without development. FINALLY!!

4-Define Referring Domains in Admin

Other important new configuration is to be able to define what is referral and what is not. If you have multiple domains or subdomains you might end up seeing one of these sites as a referral. This configuration lets you define what is referral for you. Another Omniture feature that is finally added.

Since Universal Analytics has so many new cool features that can be used for free, I plan to write more about Universal Analytics. Stay tuned..

sibel akcekaya


Packaging Matters: How to Initiate Curiosity for New Customers Offline?

Physical packaging matters! Still in this era of digital age..

I want to talk about the interesting experience I had in London. I have been here since last Friday before attending a conference on Tuesday. Between Friday and Monday I have been everywhere, went to museums, walked on different streets, got on a subway,shopped in different stores, got out of a bus and etc. I have seen many people with one same bag: PRIMARK..It was a brand I never heard, I have  known  shops such as Top Shop, Selfridge and  Harrods but I never heard about Primark. I thought this brand should have some value. Because I kept seeing it everywhere and bag was not cheap plastic bag, it was actually nice recyclable brown paper. 

Packaging Matters

I kept seeing these bags for days and I decided to look up this store online and visited offline. When I visited store I realized that it is more kind of H&M type of store, very cheap textile products, which  is actually  less stylish than its shopping back. But I ended up buying some basic items that would make me happy.

Sometimes we lose ourselves in digital world and forget simple marketing tactics that is very effective . Everybody is using similar kind of digital marketing techniques such as Google, social media and retargeting. Yeah these are all very important but let’s do one more innovative thing offline to differentiate ourselves from others.


Relearning Unique Visitor Metrics in the Multi-Channel and Multi-Device World

Do you remember how digital technology looked like early days of Internet, e-commerce and web analytics?  There were no smart devices or tablets; Internet Explorer was the dominant technology. Laptops and people with second home computers were very rare. Not everybody had 24 hour unlimited  access to the Internet and we had that nice slow dial up connection. When people wanted to reach us on digital channels, there was usually one device or one digital asset involved.

Then technology took off; computers got cheaper and better. Internet became better, cheaper and more accessible. Chrome, Firefox and Safari browsers started to become very popular and many people started to use them interchangeably. Smart devices and tablets started to be part of millions.  Customers now have many different devices and companies has many ways to meet them in digital world. Websites, mobile sites, micro sites, native apps, social media apps and etc.. Multi channel and multi device era had started.

Multi Device Era
Multi Device Era

As you know cookie is based on a browser , so with this new era there is really no way of counting absolute unique visitors. Some analyst might think that cookie is computer based, well it is not and  browsers don’t share cookies. If you use 3 browsers in one computer, you will be counted as 3 unique visitors. What we measure now is a unique visitor  per browser.

Unique Visitors
Unique Visitors

Let me tell you about my multi channel world. I have 2 personal laptops (1 MAC and 1 PC)  at home. I also use my client’s computers but I will not take that account to keep things simple.

On my PC I am using both Firefox and Chrome browsers so I am 2 users here. On my MAC I use Safari and Firefox, here I added 2 more users. I also access to similar websites on my Iphone, sometimes through browser and sometimes through native app. I have same behavior on IPAD. As you can see I am actually 8 unique visitors in digital world. Of course not everybody is  like me but we know on average people uses at least 2 devices: Computer and smart devices. On average one person can be  counted as 2 to 4 unique visitors.

As you can see it is pretty hard to define your absolute unique visitors number.Don’t get me wrong, I am not the type of analyst to obsess with perfect data; I  use visitor number for trends. But when I use this number, I know what it is and I make sure everybody also does.  For instance if you are an executive you should question how accurate this number is and if you are an analyst you should make sure that your executives don’t take this number so seriously.

Is there a no way to measure unique users who interact with my digital assets? Yes there is if you have lots of users logging in onto your site. Most e-commerce companies require customers to login to complete purchase. Other companies encourage social login to serve customers  personalized experience.  You want to measure engaged users on your site, that way you can make sense with data. Don’t forget anonymous customer trace is not a very good material for digital marketing anyway.




Google Analytics Premium versus Adobe Analytics (Omniture) Part 3 – Segmentation

This is my third article about Google Analytics Premium and Adobe Analytics (Omniture Site Catalyst) comparison. Please find my other articles below. In this article I want to talk about segmentation analysis. I am sure you heard it already, it is being called  by analytics experts all the time. crowd Most of the consumer businesses have different variety of customers: Women, men, young, old, rich, middle class, world traveler, TV lover, mobile phone user, CEO, analyst, Indian, muslim, urban residents and etc..These are example of some demographic segmentation that classical marketers always use when they develop new products or improve performance of their  existing products. Online marketing world created a need for new type of segmentation , I call it immediate segmentation which forces immediate personalizaton and targeting. Your product are: your mobile apps and your website. You have sub-products such as features, offers and content on your digital products. Some of the sub products are created at a particular instant and need to find the target segmentation at that instant. There is no time for a marketer to develop communication plan to meet that sub-product with its target market, it has to be automatic. For example 15% hotel discount for NYC only for 24 hours because you did not sell enough is a kind of immediate sub-product. Yes you might put banners on your site, you can email everybody but that’s really not  marketing, it is more of a mass communication. That’s where you actually need to design your digital channels to reflect this kind of marketing if you really want to target to your customers. The easiest way to do this is to use segmentation analysis. As I said above online marketing created immediate segments such as:

  • People who searched for NYC hotels for next week and did not buy
  • People who suddenly started to look at smart phone telephone news on your content site
  • People who came from Google, came until purchase page but did not complete sales

As an analytics expert our first job is to discover these immediate valuable segments. There is no certain segmentation here like women, christian, urban resident, these segmentation are hidden and you are the one who needs to discover them by looking at data. In order to do that Adobe Analytics (Omniture ) gives you great tools with Site Catalyst and Discover (new name:Ad Hoc Analysis) Discover is a tool to fall in love if you are a data junkie. Segmentation capabilities here are so detailed and fast. For example I can create a segments like:

  • People who came to my homepage, spent 30 seconds and moved to smart phone page
  • People who looked at business content in June-2013
  • People who came to my homepage from Google, continued to my account page, added “clothing” type of products to her cart, went to checkout but did not complete

Sequential and time based segmentation is really new and advanced. It makes me so powerful as an analyst. For more information go and read Brian Au’s great blog posts about the subject. When it comes to Google Analytics Premium, segmentation analysis capabilities are very limited.First of all you can’t segment your funnels which is so limiting.. I never look at my shopping funnels without segmentation because they are extremely different depending on the segment. You can’t apply one segment onto another, with an unlimited breakdowns.You can’t create sequential and time based segmentation which carries your view to another level. You might say all these things are very  advanced for you since you don’t even do basic analysis. This is an excuse I keep hearing all the time. I never employed that kind of thinking, and  all I can say: “Be hurry, train is going to leave soon and somebody else might get on it while you are loitering in the entrance of the trance station.” peace sibel akcekaya


Even They Fail Once In a While: How Amazon can’t find the product you’re looking for

Professionals in my field want to optimize user experience for our visitors. If you are not working in big Internet companies like Amazon, Netflix or Google you always look at them for an inspiration. They are usually very good examples. But nobody is perfect, indeed.

When you catch some of their failures, then you realize its OK to make mistakes.
You also realize that there is always some room for optimization, there is no limitation whatsoever..
I want to show one good example I caught recently.
I’ve got to admit, this is pretty advanced. I have been using Amazon for more than 9 years and I have witnessed their great improvement with its product search engine over the years. Most e-commerce companies can not even come close to Amazon’s product. But as you can see  from this example they can still do more optimization and hence make more money. Let’s see my example.
I started search for “Hemp Seeds” product  and I got my search results.
Then I clicked the first result and added this product to my cart. Then category next to the SEARCH button changed from “All” to “Grocery and Gourmet”. No I did not do that and I was not aware of that change during that time.
 Ten minutes later I remembered that I also had to buy new headphones as my old ones were almost dying. I went  back to my browser and decided to search for “headphones”. This was my result..VOILA!
I did not understand what was going on? No real headphones on Amazon? Or was there a new cool headphone product I was not aware of it..Shame on me..
Apparently Amazon brought me headphone items in the Grocery search category which was not selected by me.
As a customer I was confused , distracted and of course moved onto some other task. We all know famous STEVE KRUG’s motto: Don’t Make me Think. Most of our shopping is impulsive and if something distracts us, we are most likely stop shopping.
This example is a pure user experience failure for an e-commerce site  because it actually fails to find what customer is looking for. It is not an user error, since user was not the one who changed the category in the SEARCH.
I am sure Amazon has a good reason to change the category based on search, but as you can see it can fail very bad in some cases.
Happy Optimization
sibel akcekaya

Google Analytics Premium versus Adobe Omniture Site Catalyst Part 2 – Path Analytics

In my previous post I blogged about differences between Adobe Analytics (Omniture Site Catalyst) and Google Analytics Premium. Please go to below link if you want to get more info about these tools.

In this post I want to go into more detail about path analytics. I have this client with many pages and complicated transactions on their site but they happen to use Google Analytics . When I needed to do the path analysis on the site my only way was to set up funnel visualization which was not retrospective. It meant I had to wait sometime to collect certain amount of data. Then of course I can only have certain amount of them, I can’t keep setting up funnel visualization reports as much as I can.This frustrated me a lot. Can you imagine all the paths that can occur on a big site? How 25 goals can be enough for this job?

Path analytics will show you how your visitors navigating on your site or on your mobile apps.

You might want to see what kind of paths are happening on your site..

You might want to look at certain path that is interest to you.

When I say path I just don’t mean page flow, I mean everything..

What people searched? Links they clicked?

Let’s say 100 people searched for sport shoes, then 60 of them searched for running shoes and maybe 30 of them searched for hiking shoes in one visit.

Of course I also want to see where people go if they back out on something on my site.But I want to see their whole journey, not just one page before or after.

If you use Omniture and of course if you tagged your pages properly with s.pagename , then you can see it all. You can see FULL PATHS, you can filter these paths with the pages you want, you can create FALLOUT (similar to funnel visualization) and you will see historical data too.

With this feature I would know how people navigate on my site which is such a useful information for product managers, interaction designers and user experience professionals.

I can also segment this path data. For instance your path data will be always different if you compare one traffic source to another. Then you will understand what kind of people you acquire with your marketing channels.

All good stuff.

With this in mind, I promise to write something good about Google Analytics next

cheers and loves
sibel akcekaya


Google Analytics Premium versus Adobe Omniture Site Catalyst

I love Google.I can search anything for free and can access all the world’s information through Google.I love Gmail, it is one of the best e-mail platforms out there. I love Google Maps, I can find my way around anywhere in the world. I love Google Analytics. Internet businesses can start to track their websites clickstream data by just adding 1 line of JS code to their pages. I don’t know what would happen if there was no free Google Analytics. I want to say thousand THANKS to the Google Analytics creators in Google.

Yes Google Analytics  is awesome and free but do we know what it actually does? Can it be used instead of enterprise solutions such as Adobe Omniture Site Catalyst? My answer is to that would be it really depends..

Google Analytics free version is not definitely for enterprises or even medium companies since it is a sample of data and you really can’t depend on that data to make critical business decisions.

Let’s also make distinction between web analytics and online marketing analytics to make sure we don’t think they are the same thing. Web analytics is more about measuring your product performance, clickstream data and measuring your site optimization efforts. Online marketing analytics measures your online campaigns such as adwords, email etc along with SEO..They are not same but they are connected.

For me the biggest difference between Omniture and Google Analytics is this: Google Analytics’ main skill is online marketing analytics and Omniture’s main skill is web analytics and optimization. I mean in Google Analytics you can see both Adwords and SEO performance together which is very vital for your acquisition efforts.

But when it comes to real web analytics that will help you to understand your product performance, Omniture is the best tool out there. All the customizations you can create with evars, props and events can’t easily be done by Google Analytics.

For me this is one of the big advantages of Omniture. You host your own JS library and you can add some code to that library so that you don’t have to add custom tagging on the page. Unfortunately Google Analytics requires page level JS code for every customization.

I also happen to love Omniture’s new solution “Processing Rules” which helps me to create new variables without any tagging at all. I can simply create some variables by using existing ones. Unfortunately this feature doesn’t cover complex cases so I just hope Omniture can improve this product more along the way.

Omniture VISTA is also another way to create variables without tagging. This is done by Omniture engineers at the database right after your data is sent and right befoire data shows up in the interface. This is more advanced than processing rules, but of course this one has a cost.

This one is so important. I want to be able to track my customers on the site at an individual level to target them later. In Omniture I can do that but Google Analytics doesn’t still allow that. I think not to be able to track your members, or logged in users closely, is a big issue with GA.

This is also one of my important tools. I stopped looking at pathing at a global level as it really doesn’t tell me any story. I look at pathing report by segments. I can look at my funnel by many segments you can imagine:

  • Campaigns to understand the point in the funnel where I have issue or opportunity. Then I can alter my ad campaigns to address that specific issue or opportunity.
  • How my loyal customer move through funnel?
  • What about users from Germany? Should I design my product different for people from Germany
As you can see this list can go on and on..Segmentation is the most essential tool for us and funnel segmentation is the top priority.
Omniture admin console is pretty sophisticated and it is a must for enterprises, let me give couple examples:
  • I can share segments, dashboards or reports with anybody or with any group very easily
  • I can limit who can see what report at many levels
  • I can include/exclude IP addresses
  • I can customize menu and display most important items
  • I can create calculated metrics for my users
  • I can define when and how my conversion variables are expired, something that Google Analytics custom variables does not have
As you can see I have my own reasons why I prefer Omniture over Google Analytics. I am mainly focused on web analytics and optimization and unfortunately Google Analytics can’t provide all the tools I need for my optimization efforts. I talked to Google about analytics premium product and asked them If I could do all these things that I can do now. I was told that it is a beta program and I would be working with Google consultants try to realize all my objectives. There is an issue here. Why will I pay 100,000 USD for the product with unclear results. I just hope Google makes this product cheaper so that more customers can try it out. That way they can build a better product that can compete with other products in the market.
I am strong a believer for competition. Competition makes companies work more and create better products at a cheaper price :) That’s what society needs..
sibel akcekaya



Web Analytics, Testing and Marketing Learnings from Street Food guy

There is this street food guy on my way to subway. He sells delicious grilled corns. I keep buying these corns once in a while. He is awesome, he recognizes me and he always gives me a warm smile.

TIP1: Building relationships and customer lifetime value
He is always out there to build great relationships with its customers, he values customer lifetime value. Even if you did not buy from him, he always smiles at you from his heart to keep our relationship forever.

Then I stopped buying these corned meals as it became something usual for me . I kept smiling at him without any purchase and he kept smiling back at me. Then one day he crossed my road with a kind behavior and offered his new product, grilled chestnuts and he told me:
“Last time you passed by I wanted to offer you a complimentary chestnut but you walked very fast, please take one”

TIP2: Monitor your customers behaviors and understand changes in their behavior over time. If they are no longer loyal with you find out why? Maybe your products are obsolete. Even if you have nice new products, how you convince them to buy your product? Do testing and find the best method
He apparently watches his every customer and monitors their behavior -WEB ANALYTICS- and takes different actions each time to get their attention -TESTING. Striking this for me was his understanding of how  he needs to try different ways to get my attention.

TIP3: If you lost some of your customers, try new incentives and give out some coupons
He is very creative and he knew, he could get my attention if he offers me a free stuff to convince me for a real purchase.

At the end it worked. I bought a pack of chestnuts. I actually like chestnuts and of course as a business woman I have lots of respects to this street guy who employs natural business skills without any school education. Instead he gets his education real time from life with observation.

The most important thing that most intellectual and educated brains miss, simple human observation.

Well done my friend

sibel akcekaya