Monday, October 21, 2019

Artificial Intelligence (AI) for Marketing 101

These days there is a lot of buzz in the marketing community about use of Artificial Intelligence (AI) and Machine Learning (ML) in marketing.

In this post I will cover some basics of AI that you need to know before you can explore how AI and ML can help you in your marketing efforts.

What is Artificial Intelligence? 

There are several definitions of Artificial Intelligence or AI. The simplest one to understand is from Oracle.com: "Artificial intelligence (AI) refers to systems or machines that mimic human intelligence to perform tasks and can iteratively improve themselves based on the information they collect."

So in a nutshell AI refers to machine, which can learn and become intelligent like humans.

AI is an umbrella term that includes algorithms, concepts, tools, technologies etc. that perform these complex human like tasks.

One of such and widely used concept in AI is Machine Learning.  Keep in mind that all machine learning is AI but not all AI is Machine Learning as AI include much more than just Machine Learning (ML).

What is Machine Learning (ML)?

Machine learning is the practice of using statistics to parse large amount of data (structured and unstructured), find patterns in it, learn from it, and then make a determination or prediction about something in the world. So rather than hand-coding software routines with a specific set of instructions to accomplish a particular task, the machine is “trained” using large amounts of data and algorithms that give it the ability to learn how to perform the task. (definition modified and adopted from: Nvidia).

Machine learning builds a mathematical model based on sample data, known as "training data", in order to make predictions or decisions without being explicitly programmed to perform the task. (source: https://en.wikipedia.org/wiki/Machine_learning)

There are three major types of learning used to train these models - Supervised learning, Unsupervised Learning and Reinforcement Learning.

Supervised Learning

In supervised learning, the algorithm builds a mathematical model from a set of data that contains both the inputs and the desired outputs.
(source: https://en.wikipedia.org/wiki/Machine_learning)

Example: Predict churn propensity of a customer.
You can provide training data that contains customers purchase data, past behavior data (input) and then each customer is labeled if they churned or not. Based on this data, model learns what purchase and behavior data will cause all the customers to be labeled as "Churn Risk" or "Not Churn Risk".


Unsupervised Learning

In unsupervised learning, the algorithm builds a mathematical model from a set of data which contains only inputs and no desired output labels. Unsupervised learning algorithms are used to find structure in the data, like grouping or clustering of data points. Unsupervised learning can discover patterns in the data, and can group the inputs into categories. (source: https://en.wikipedia.org/wiki/Machine_learning)

Example: Uncover customer segments.
Unsupervised learning can help find various customer segments in your customer data using customer attributes, sales, onsite behavior etc.. This can then be used to drive better customer engagement and better marketing performance.

Reinforcement Learning

In Reinforcement learning, the agent (also called Machine, model or AI) is given a problem to solve and faces a game-like situation. It is given rewards for positive behavior and punished for negative behavior as it tries to solve the problem.. These rewards are provided by the developer of AI. The machine uses trial and error to come up with a solution to the problem. The developer does not provide the model any hints or suggestions for how to solve the game. It’s up to the model to figure out how to solve the problem and maximize the reward. The end goal is to make the model learn desired behavior that maximizes the total reward.

Example: Provide recommended products to customers.
Reinforcement leaning can be used to develop a online product recommendation engine.

Other Terms that you should be aware of

Structured Data

Data that can be organized in rows and columns such as Customer Demographics, Sales data, onsite behavior data etc.

Unstructured Data


Free form data such as word documents, call scripts, pdf, images etc. Anything that is not structured is classified as Unstructured data.

Marketing Uses of AI

There are several ways AI can be used in Marketing.  Here are some examples, this is not a complete list. I will add more articles in future to cover several use cases.
  • Customer Segmentation
  • Ad budget allocation across channels or by channel
  • Content creation
  • Chatbots - Which understands humans questions and then responds with appropriate response.
  • Churn Prediction/Customer Retention
  • Product recommendation engine
Hopefully this article provides some clarity to the confusion around AI in Marketing.

Your turn now. Are you using AI for marketing? If yes, how? If not then why not? What are the challenges. Let's talk.


----------------------------------------------------------------------------------------------------------

Signup for my online courses. Here is a coupons for 50% off on the first month. Use coupon code 50OFFFIRSTMONTH at checkout.



Thursday, October 10, 2019

Essential Excel Features & Functions for Marketing Analytics

Excel is one of the most used tool for data analysis by marketers, marketing analysts, web and digital analysts.
We (Optizent.com) have developed a course that will teach you Excel Functions and Features that you need for Marketing Analytics. You will also get several practical examples of how to use them for your analysis. Not only do we teach excel functions and features useful for Marketing, Digital and Web Analytics but also you will get cool ideas about what data you can analyze.

What you’ll learn
  • Understand various types of errors Excel throws and how to fix them
  • How to suppress the error or replace them with more meaningful data
  • Learn different ways of combining data from multiple columns
  • Understand different types of cell references.
  • Bulk delete blank rows to clean your excel sheet data
  • Find duplicate rows of data, highlight them and/or bulk delete them
  • Learn how to use VLOOKUP - Uses search engine (SEO) keyword data as an example.
  • Learn Pivot Tables - Example:  Digital Marketing Campaign Analysis
  • Find Optimal Campaign Budget Allocation with Excel Add-In Solver - This will change the way you allocate your marketing budgets.
Course requirements or prerequisites?
  • You should be familiar with the basics of Excel.
Who this course is for:
  • Anybody who wants to learn about some key features in Excel how to apply them to common data issues.
  • Marketing Analysts
  • Digital & Web Analysts
  • SEO analysts

Get this course on Udemy

or signup for Optizent Institute and get this course and several others with One Month Free Trial (Cancel anytime)

Tuesday, October 08, 2019

Search Engine Optimization with Google Analytics

There are several tools in the market for Search Engine Optimization (SEO). Some are free and some are paid. Google Analytics is one such tool, which is free.

I have developed an online course, that will teach how to use Google Analytics to get better at Search Engine Optimization (SEO).

Google Analytics provide lots of valuable information that you can use to your advantage and have higher ranking sites on Google, Bing etc.
This course also gives you an overview of Google Webmaster Tools also called Search Console and show you how you can get that data into Google Analytics.

What will you learn:
  1. Understand which Google Analytics reports to use to get insights to improve SEO
  2. Get insight into (not provided) keywords that show up in Google Analytics keyword report.
  3. Get a quick overview of Google Web Master Tools (Google Search Console) and connect it to Google Analytics
  4. Understand and use Landing Pages, Devices, Countries and Keyword reports from Google Search Console
  5. Learn how to compare various traffic drivers showing up in Google Analytics reports.
  6. Segment Organic traffic in your reports.
  7. Get a free SEO Custom Dashboard template that you can use to develop your own dashboard and use it right away and wow your boss.

or get access to all of Optizent courses (50% off First month) at https://training.optizent.com/p/full-access-to-all-courses/?product_id=1025897&coupon_code=FREETRIAL






Thursday, March 14, 2019

Context is Critical: Creating a Culture of Web Analytics

Continuing my series on Creating a Culture of Analytics I would like to touch on a very critical aspect of creating a culture of Web Analytics and that is Context.

What is Context

According to Princeton.edu context is
  • Discourse that surrounds a language unit and helps to determine its interpretation
  • the set of facts or circumstances that surround a situation or event

Context takes the ambiguity out of the equation. As an Analyst it is very important that you provide full context when reporting your web analytics data. Context gets everybody on the same page. Do not leave anything for interpretation by the end users of your reports, give them the insights in a simple and easy to understand format.

Let’s look at an example to understand critical context is.

60 Degrees

If I say it is going to be 60 degrees tomorrow. What will be you reaction?
If you are in Minnesota – You will yell “Summer”
If you are in Seattle, you will think – ““Spring”
If you are in Florida, you will say “ Damn… Cold”
If you are in India, you will say “WTF….” (Indians measures temperature in Celsius and 60 degrees Celsius is 140 F)

Some other question that might pop in people’s mind are:
  • What is the temperature today?
  • Is it normal to have 60 degrees this time of the year

Without context 60 degrees does not mean much. Right.
Similarly when you report your numbers and tell report on visits, page views, time on site etc. it does not mean much unless you provide the full context.

Web Analytics & Context

Just saying that Visits are down by 10% from last week is not enough. You have to put that 10% decline in full context. Tell your end users what happened and why they should or should not worry.

So add something like : Visits are down 10% from last week and also 10% lower compared to the same time last year. Prior to this week we saw a 10% year over year growth but last week was abnormally down. Isn’t that getting better now?

You should go even further: Last year we got some free advertising from local newspaper sites that drove 20% additional traffic same time last year. Since we did not have the advertising deal this year, it impacted our visits this year. We noted the potential impact of newspaper site advertising in our last year’s annual recap (here is the link to last year report – people forget so remind them). If we take out the impact of spike from newspaper sites then we have a consistent pattern of 10% year over year increase. As noted in last few reports, that increase is due to our social media efforts this year. Now the picture is much clearer. Of course you should look into the full impact e.g. conversion, bounces, sales etc. (Note: How you present this story will depend on what format you chose to present your report)

Now everybody is on the same page and knows exactly what those numbers mean. Without that context, everybody would have had their own interpretations of the data. Misinterpretations lead to wrong action and/or mistrust in the data and the analytics team.

Final Words

Do not provide any reports without providing full context. Keep in mind that most of the canned and automatic reports do more harm than good because they do not provide context.

Other posts in the series


Note: This post was originally posted on March 4th, 2011 but it is still very relevant.

Follow me on twitter @anilbatra

Facebook Page http://www.facebook.com/pages/Anil-Batra-Page/130050670343547

Optizent.com - Digital Marketing & Analytics Consulting and Training



Tuesday, September 18, 2018

Advanced Digital Analytics workshop at DA Hub on Oct 17th



I will be conducting a full day advanced digital analytics class at DA Hub in Austin on Oct 17th.

Do you have experience with influencing business change? Are you ready to move into reflective analysis that drives real return on investment for your company? Want to advance your career prospects? This course will help you deliver all of that.

This interactive full day advanced digital analytics course will give you practical tools to elevate your analytics skills. It will also help you transition into a thought leader within your organization. 

Building upon the Fundamentals of Digital Analytics, this course encourages participants to critically examine methods and metrics in ways appropriate for their specific business model.

Learn how to effectively communicate analytical data, best practices for site and campaign design, and examine campaign performance standards from the CFO perspective using control groups and simple predictive models.

Upon completion of this course, participants will be able to:

  • Define gaps between measurement practices and business goals for the company 
  • Create a framework for standardizing the evaluation of current performance measurements for website, search, display, email, affiliate and social channels.
  • Understand how to effectively measure campaigns
  • Understand how to find right opportunities for improvement and optimize them
  • Provide feedback on the design, execution, and outcomes of testing efforts 
  • Manage a results-oriented analytical culture adept at driving business change 


The course follows the Digital Analytics Association's syllabus and is endorsed by it. Participants will be awarded six Professional Development Units (PDU) towards their DAA Certified Web Analyst certification renewal credit.



This works shop will be held in Austin, TX on Oct 17th along with DA Hub. Register at https://www.digitalanalyticshub.com/us_2018/anil_batra?reg_type_id=19719

About DA Hub

DA Hub is bringing together top analytics professionals for in-depth idea exchange. Leading digital analytics and optimization practitioners come from across the US to discuss and share the latest developments, challenges and opportunities in the industry. It is a unique opportunity to be part of the conversation. With over 50 huddles to pick from, run by the industry’s foremost practitioners and only 160 places available, you should book your place for this year’s event to meet, learn, share and network.  Get details at https://www.digitalanalyticshub.com/us_2018/home?reg_type_id=19719


Saturday, August 18, 2018

23 Email Marketing Metrics That You Should Know

A while ago I wrote a post on Measuring Online Display Advertising. Continuing the theme, in this post I am describing 23 metrics for measuring email marketing.

Get this article as an eBook at Global Analytics Academy.
  1. Sent: The numbers of emails (generally unique email address) that were sent. This number excludes any suppressions that occurs due to business rules, privacy compliance etc. Your email service provider (ESP) will have these number as a standard metrics in their reports. This is a raw metrics that is used to calculate other email performance metrics and is generally of low value by itself so should not be used as a Key Performance Indicator (KPI). A trending of this metric overtime can provide you a view into the health of your marketing list.
  2. Delivered is a count of emails that made their way into the recipient inbox. If an email was rejected by the recipients email provide then it is not counted as delivered. If an email shows up in junk folder, it is still counted as delivered. Your email service provider will have these number as a standard metrics in their reports. This is a raw metrics that is used to calculate other metrics. A trending of this metric overtime can provide you a view into the health of your marketing list. Delivery rate (described below) is a better indicator of Delivery issues than the raw number. This metrics by itself is not a KPI but forms a basis for other KPIs.
  3. Delivery Rate – It is calculated as Emails Delivered divided by Emails sent, expressed as a percentage is Email Delivery Rate. Delivery Rate measure the quality of your email list, goal is to have 100% delivery rate, but I can guarantee that it is not going to happen. Any deviation from 100% should be investigated to see what is causing the issues. If there are some hard bounces (see below) then those should be removed promptly. Too many hard bounces can lead to spam triggers and further delivery issues.
  4. Bounce – An email is considered a bounce when it cannot be delivered to the intended email address. There are two types of bounces – Hard Bounce and Soft Bounce. Hard bounce generally means that the email address is wrong or no longer exists. Soft Bounce generally means that the addresses exists but either the inbox is full or is having temporary issues, the message is too large to deliver etc. You should immediately remove Hard Bounces from your email list since they are dead and you will never be able to deliver an email to them. Raw number of bounces should not be used as a KPI.
  5. Bounce Rate – Bounce rate is measured as Bounces divided by emails sent, expressed as a percentage. It is exact opposite of Delivery Rate. (Note: this should not be confused with the Landing Page Bounce Rate)
  6. Total Opens: Total opens measure the number of times your email has been viewed by the recipients. It gets counted when a recipient opens the email. Emails use a small invisible pixel (image) that gets loaded every time an email is viewed, the loading of this invisible pixel is counted as an open. Few things to keep in mind
    1. Any recipient who have disabled the images will not be counted in the open metrics since the invisible pixel won’t be loaded.
    2. Any recipient who has preview pane open will be counted as open as the emails gets loaded in preview pane enough though the person might not actually open it.
    3. Multiple views by same respondent will increase the open count, one for each view (open).
  7. Unique Opens - This measures the number of unique recipients who opened the email. Unlike Total opens, multiple views (opens) by a same recipient will be counted as one unique open.
  8. Total Open Rate – This measure the effectiveness of your subject line and your brand (shown in from column of email) in driving people to open the emails. Email open is the first action by user in their journey to engage with your email. This metrics is calculated as Total Opens divided by Delivered, expressed as percentage.
  9. Total Clicks or Clicks - Total number of clicks on any link in the email is counted in this metric. Keep in mind that a click does not mean that a person landed on the intended destination of the link hence you will likely see a discrepancy in this metric, as shown by your ESP, and the number shown in your Web Analytics tool. There are multiple factors that could lead to a click but not a visit to the destination. If one recipients clicks on multiple links then each click is counted in this metric.
  10. Unique Clicks – Unique Click counts the number of unique recipients who clicked on one or more clicks. Unlike Total Clicks, Unique clicks counts each person only once, no matter how many links that person clicks.
  11. Click to Open Rate – It measures how effective your newsletter content is in driving people to take actions. It is calculated as Unique Clicks divided by Unique Opens, expressed as a percentage.
  12. Total Click Through Rate – It is calculated as Clicks divided by Delivered, expressed as a percentage. If a person clicks on 2 links then the number of clicks will be 2. Considering that one person (email recipient) can click multiple links in the email, this number can potentially go over 100%.
  13. Unique Click Through Rate – It is calculated as Unique Clicks divided by Delivered, expressed as a percentage. Even if one person clicks on multiple links, only one click is counted in this calculation. Keep in mind that if someone talks about Click through Rate then they are referring to this metrics. This is also used for industry benchmarking by various vendors.
  14. Email Conversions – Email Conversion is defined as the count of action that you want the visitors to take when they arrive as a direct result of a click on the email. Some examples of conversions are – purchase, download a whitepaper, sign up for an event etc.
  15. Conversion Rate – It is calculated as Number of Email Conversions divided by Delivered, expressed as a percentage. Some vendors use the sent metrics as denominators and in some organizations I have seen the Unique clicks (visits in the Web Analytics tools) used as the denominators. (also see, 21 Metrics to Measure Online Display Advertising)
  16. Unsubscribes – Number of emails recipients who chose to unsubscribe from your future mailings. This number is available in your ESPs report.
  17. Unsubscribe Rate – Unsubscribes Rate is calculated as Unsubscribes divided by Delivered and is expressed as a percentage. It measure the effectiveness of your email marketing strategy and the quality/relevance of your email marketing. If this number continues to rise, you have a problem that should be immediately fixed. The fixes range from adjusting the email frequency to increasing the relevance of the message.
  18. Email Complaint or Spam Complaint – Number of email subscribers who have marked your emails as Spam. This number is readily available in most of the ESP. SPAM complains can totally kill your email marketing so this number should be watched closely and steps should be taken to ensure that you have users permission to market and are sending the relevant messages at the right frequency. This number should be available from your ESP
  19. Email Complaint Rate/Spam Complaint Rate Number of emails complaints divided by total emails delivered, express as a percentage.
  20. List Growth Rate – Measures, how fast your email list is growing, it is the net results of new subscribers minus the unsubscribes and email/spam complaints. You have to make sure that your list continues to grow rather go in negative direction. Growth (new subscribers – unsubscribes- email complaints) divided by total list size is your growth rate. Your email marketing program depends on List Growth so watch this number closely and take actions to actively grow your email list.
  21. Forward Rate/Share RateThis measures the emails forwarded (shared) by your recipients to their friends/contacts. It is calculated as number of forwards divided by number of emails delivered and is expressed as a percentage. It provides a view into the effectiveness of your email in not only engaging your recipients but also driving new subscribers, people who become of your brand as a results of receiving emails from their friends. This number is available in some ESPs.
  22. RevenueThis measures the Revenue generated as a direct results of email. Several version of Revenue as a KPI are
    1. Revenue Per Sent – Revenue attributed directly to the email divided by number of emails sent. This is also sometimes expressed in terms of Revenue Per 1000 (RPM).
    2. Revenue Per Click - Revenue attributed directly to the email divided by number of unique clicks.
    3. Revenue Per Open ­ - Revenue attributed directly to the email divided by number of unique opens.Revenue numbers won’t be available in your ESP but can be tracked in Web Analytics tools for online sales.
  23. CPM – CPM stand for Cost Per Mile (1000 in Latin). This is generally used when you rent/buy emails list from third parties. It is the cost of renting 1000 email address and is calculated as (Cost/Emails)*1000. This rate should be provided to you by the vendor from whom you are renting the list. If not provided then you can use the above calculations.
Get this article as an eBook at Global Analytics Academy.

Here are few more email marketing posts that you will like:
  1. One costly email mistake that you can easily fix Growing email list is a hard job. All you Growth hacking goes down the drain when you make a simple mistakes that costs you subscribers that you just gained. This posts you one such mistake and how to fix it.
  2. Email Personalization Not Working? Read This This posts explains why the email personalization might not work. The bottom line is that you have update your personalization criteria over time and test it.
  3. 3 Techniques for Expanding your Email Reach Email marketers are facing a tough time with growing emails remaining unopened and unsubscribes. Acquiring new subscribers using old techniques is expensive. In this post I have listed 3 techniques that you can use to spread the word of your emails/newsletters beyond the email list that you are sending the emails to.
  4. Are You Depleting Your Email List? Email marketers, in order to maximize short term conversions, often bombard irrelevant emails in subscribers inbox However this short term mentality results in erosion of long term viability of their email marketing, due to increase in unsubscribes causing depletion of email lists.
  5. 15 Things to Test in your Email Campaign This post talks about 15 things you can test today.
  6. Targeting Cart Abandonment by Email Targeting Cart Abandonment is a great way to drive conversions however, use incentives/offers cautiously.
  7. Conversion Tip: Making the Most of the Email Confirmation Thank you Page Use your Confirmation page effectively, this posts shows an example of a good page and a not so good page.
  8. Number One Email Marketing Mistake Number one mistake marketers make with email marketing is to send “Irrelevant” messages to their customers. Find out why this strategy has a far-reaching impact on your email marketing program.
  9. 7 Ways to Create Relevancy in Emails 7 tried and tested ways of creating relevancy in emails are described in this post.
  10. Relevancy Matters in Email Marketing This post shows an example of an email that missed the opportunity to convert.
Get this article as an eBook at Global Analytics Academy.


Questions/Comments?

Subscribe to my mailing list

* indicates required

Thursday, August 02, 2018

Google Tag Manager Workshop in Seattle, WA

I will be conducting a full day Google Tag Manager Workshop in Bellevue (Seattle),  Come and learn how Google Tag Manage works and how you can start it to use it for your business.
After this class, you will know how to confidently use Google Tag Manager and deploy Google Analytics and Facebook pixel.


Date: Sept 17th
Time: 10 AM – 4:00 PM
Location: 3600 136th Pl SE # 300, Bellevue, WA 98006
Pricing
$699 – Before Aug 15th (Early Bird)  - Save $100 - Signup now
$799 – After Aug 15th


This is a great opportunity for you to remove your fear of Google Tag Manager, and get trained so that you can use it with confidence.
What will this training cover?
This course will cover every thing you need to know to start using Google Tag Manager with confidence. Covers the latest version of Google Tag Manager (2018). I am very confident that you will love this course.

Here is what some of the students of my online class are saying:

Troy – AWESOME COURSE! I bought like 4 courses (including stuff for google tag manager) on udemy to teach me this and NOTHING came close to what Anil delivered in this course! I don't usually rate courses and place comments but I will make an exception in this case. Phenomenal class, covers everything, & well worth the money!

Ashish Batra – Initially I wasn’t sure if I should subscribe to this course or not as I usually buy courses with 100+ reviews. I am glad I purchased it. Anil has done a fantastic job in this course. If you are a technical marketer, you must do it. Previously, I have done some other GTM courses and watched youtube videos. But this course is definitely among more practical courses and added value to my existing knowledge. p.s. Coincidentally, I share last name with instructor, but we aren’t related 

Kate Proyka –  The course is well structured, clear and covers all elements of the tool. There are several examples which can be easily implemented and make sense.
Bryan Bloom – I already love love this course. It is at the correct speed and amount of explanation. I was so scared of GTM and now I am learning it and loving it!!!
In this course you will learn
  1. Fundamentals of Tag Manger (Applies to any tag manager)
  2. Signing up for Google Tag Manager
  3. Details of Google Tag Manager Interface
  4. How to setup Google Tag Manager for Google Analytics and track page views
  5. How to setup external link tracking as Events in Google Analytics via Google Tag Manager
  6. How to setup Button click tracking in Google Analytics
  7. Track JavaScript errors using GTM
  8. Deploy GTM in Wodpress
  9. Use Data Layer in Google Tag Manager
  10. Facebook Conversion and Re-targeting Pixel
  11. Facebook event tracking
Note: You will need basic understanding of HTML and JavaScript to use some advanced tracking using GTM.