Archive for the ‘Analysis’ Category

The Mirage that is Direct Traffic

Wednesday, July 19th, 2017

All of us have seen it. Few talk about it. Most of us don’t want to admit it.

What is this ghost called direct traffic? How could customers/readers suddenly divine my name and land on my homepage. There are various reasons that someone can be captured in this catch-all bucket. All of them point to either incorrect implementation of the Google Analytics code or limitations of it.

So, should we just ignore it? In essence, all that is happening is that the referrer domain is null for some sessions landing on our site.

I posit that we can still find some uses for this nullity. What can they possibly be?

  1. Proportional allocation: Attribution of traffic in such terms as Google Analytics can show us, is not only simple but also simplistic and illusionary. An article about your business or a mention of it somewhere online does not mean that a user will click on it immediately. I see a lot of ads for products that might be interesting and then just go type the URL. Does that make me [direct]? Not at all. I am responding to your branding in media channels. I have known social campaign launches to create a significant increase in direct traffic. I would actually attribute an increase in direct traffic proportionately to all my branded marketing spends.
  2. Use as Control: Some marketers have argued that as much as 60% of direct is actually organic traffic. While that maybe true in experiments that they ran, it will not be the case for you if your implementation is correct. I would propose to use direct traffic as your control traffic. They are not attributed to your marketing efforts so the behavior of this traffic should contrast with your mailings (be a dear and use UTM parameters), social media, paid media etc. This difference in behavior (read conversions to goals) that you will have induced from your marketing will tell you how successful your marketing efforts really are.
  3. TV spends: It is notoriously hard to determine the success of TV spends. Several great analysts have written about it but no one really has a rock-solid method to determine the success of TV campaigns. However, if your Direct traffic goes up in the vicinity(time and place) of TV spends, you know they are working. We can slice and dice our traffic based on geography as well as DMA and have a clearer idea of the impact of TV. One thing to remember is that TV will impact other channels too so please compare and contrast direct with all of them. It has been shown that direct will move the most. Of course, this is quite subjective and I would use it carefully.

Anyone else using direct traffic in their analysis?

How to track UserID in Google Analytics

Thursday, July 13th, 2017

Why do we need to do that?

Ostensibly Google says that you do this for tracking true users of your site. Since multiple devices generate a visitor cookie each, they get counted as different visitors. You can stitch sessions cross-device by creating a userid view.

However, if we want to track CAC and LTV by referral sources and we have only Google Analytics and an internal database around, we can piggyback on this method to them.

It is tricky and will require some testing before it runs in your environment. However, it works and I have managed to figure it out in Google Tag Manager and will lay it out for you dear analyst, step by step.

Step 1: Create the User-ID view in Google Analytics.

Go to the Property and expand “.js tracking Info”. Then follow the prompts as seen in the picture below.







After you have set this up, let’s go set up a dimension to store userid data into.

Step 2: Create the UserId dimension in Google Analytics

In the same property column, below the PRODUCT LINKING tab, sits a place to create custom dimensions as shown in this picture below.




Click on create dimension and add the custom dimension as shown below:
















Please note that you have to name it userID. This is because I am going to name the data layer variable later the same. They have to be so to capture your user information.

Step 3: Have your awesome developer pass the user-id value to the data layer variable called userID. The values that it should spit out if you click “view-source” on your site should look like this:

dataLayer = [{‘userID’: ‘yyyyyy’}];
dataLayer.push( {‘event’:’uidAvailable’, ‘uid’:’yyyyyy’} );

They can push this change live anytime and will be quite thankful for the Starbucks drink you will treat them to.

Make very sure that this userID being made cannot be Personally identifiable information(PII). So no names, email addresses, etc.


Step 4: Ok! Let’s push on now to the Google Tag Manager and the last part of this implementation.

Open up your tag manager and on the left hand side where it says, variables, click and add a user defined variable. We will call it userID.

Now, on the left hand side open up the tags section. On all the tags that have anything to do with an event or page view, we will make slight edits. Let’s start with the Google Analytics tag (it should be set to universal Analytics otherwise this will not work).

We add two variable to be filled when the tag fires, one is the userId(please note that this is the exact name you need to set. The other is the custom dimension that we have created in Google Analytics so we can run reports based on our own user-ids, which I am sure, are aligned to email addresses.


Step 5: Now, let’s set our workspace on Google Tag Manager to Preview the changes that we have made. If this works, the Real-Time reports in Google Analytics will immediately start collecting data. Once it works, then go ahead and publish the new tags.

Now, you should be free to download the userID and referral source. You can download the LTV and userID from your own database and analyze in Excel or any other spreadsheet of your choice.

You would have noticed that I keep harping on the name and case of the variable “userID” and “userId”. I thought to name all variables the same and it didn’t work. However, hard-coding the variables works only!


If this doesn’t work, write to me and I will help you make it work. (ateeqahmad[at]gmail[dot]com)

A Framework for Measuring SaaS Product Success

Tuesday, June 13th, 2017

This is a generic framework I would use to measure the success of any web-based SaaS tool. It could be an app, a website, or even your own blog. This could apply to any tool but that is too broad a base and tons of books have been penned on the subject.

Grow Visitors/Users
It is imperative to grow visitors so that we can improve the visibility of the tool.

1. Where is the traffic coming from? (SEO, PPC, Social Media channels etc.)
2. What are the main landing pages on the site? ( are they optimally designed)
3. What are the main partners or referrers to the tool? ( any referral entities or affiliates)

Grow Registration/Subscriptions/Orders

1. Minimize steps to sign up or to shop.
2. Are registration pages/landing pages based on different personae?
3. What are the calls to action? Are they prominent enough?

Improve Onboarding

1. What happens when a user subscribes or signs up? How do we drive them quickly through the process?
2. Track sharing of information, queries through chats, email etc.
3. Is there an onboarding video walk-through for the users so they can quickly see what they are getting?
4. Improve Usability & UX based on user testing of people navigating after entering through the paywall.

Improve Speed and Performance
1. This is where you look at your technology and ask if you need to migrate to better code.
2. Improve all landing page speeds by using either CDN’s or removing external elements ( Youtube videos) that will slow it down.
3. Enable tracking in such a way that the technology and product teams are always aware of the load that each page inflicts on the servers.

Pricing and Retention

1. Every product has several pricing options online that have to be tested and optimized every year. Your costs go up every year but SaaS pricing people are very reluctant to do anything with product prices lest customers be driven away. If you communicate the reasons clearly, your customers will agree to small price changes without a problem.

2. What happens when a customer cancels? Is there a churn management system? How are customers enticed to continue on?

3. Is there a way a customer can upgrade to the next pricing tier? Are they emailed or otherwise contacted about it?

Gather Customer Feedback
1. I have already mentioned usability testing but talking to customers and how they want information packaged will be very important for any tool.
2. Customer satisfaction surveys and NPS scores, if used properly can help the product owners and hence ultimate user experience tremendously.

There are going to be hundreds of KPI’s that you can build on these measures but I urge you to have as few vanity metrics as possible. Otherwise, you will be drowning in unnecessary data that is of no use to anyone.

The Education of a Web Analyst

Saturday, May 20th, 2017
There are really two streams that we can see these days coming out of Master’s programs. One is the marketing/product Analytics practitioner and the other is data science practitioners. At the initial levels we need to learn  both streams because it will not really be clear to a student where their capability or interests truly lie.
Based on this premise, I would like to suggest a few types of learning that web analyst in particular need to imbibe in the start of their careers.
Business Writing is factual journalism.

I have experienced a lot of analysts coming in with lots of capability with numbers but none with writing business stories. I also see many analysts who don’t know the first thing about how to use stats tests in real life situations. Did the pricing change lead to more leads? Are the results statistically significant?

Several online course providers offer creative writing, business writing or journal writing courses and I think they should be part of any Analysts formative education.
Knowledge of the tools.
We need to provide hands-on understanding of the tools that are used in typical business analytics. I have seen a lot of spreadsheets(Excel wizardry can take you lots of places) being used.
There are some statistical packages like R(open sources and free, hence very useful), SPSS and SAS that the analyst should be familiar with. More importantly, they need to be able to live and breathe statistics because the core fundamentals are always useful. Some elements of data mining and clustering should also be learnt as needed or as a person’s curiosity demands it.
These days, data visualization done correctly, explains half the analysis by itself. So, some experience in Tableau, Microstrategy, Qlikview is very important in explaining the information clearly.
At the end, it is all code.
Whatever the environment of development in an organization, the analyst should have a fair idea about it since all the measurement hooks are mostly encoded. Even the ubiquitous Google Analytics is run based on a javascript pixel. A solid grounding in Python, PHP would be important so we need some intermediate technical courses to get a well-rounded technical perspective.
Understand the data river as it flows downstream
Although, large amounts of data exists, we don’t get training in information architecture.. How do you get unstructured data to structured data, to BI tools and to reports and metrics is something that needs to be taught too. The analyst can use some BI tools and data warehouse development and maintenance skills. As it become more common to build self-service reporting systems, we need to be able to teach some rudiments of this skill to our budding analysts.
Project Management
Most Analysts end up in large project teams as core members in-charge of numbers. So we need to teach them principles of project management and the current techniques like Agile and kanban. This will help them build their managerial skills too.
Design and Analytics
A good analyst will have a strong creative streak, not just have a good head for numbers. We need to be able to look the system that we are trying to improve (or make profitable) from a design perspective too.
In fact, a lot of web analytics is done just to measure and influence the design and flow of the information on the site for the customer. Conversely, not many designers are just going to rely on their intuition while building a site. They would ask the analyst about measuring the success of a particular design. So, a couple of introductory courses on creative design are essential for the education of an analyst.
I hope these make sense dear reader. Please let me know your thoughts.