Driving Growth with Personalisation & Optimisation – a Google Optimize Google Marketing Platform Sydney Meetup

In March we held our second Google Marketing Platform – Sydney meetup at Google HQ. We talked about Driving Growth with Personalisation & Optimisation – a Google Optimize Google Marketing Platform Sydney meetup.

Driving Growth with Personalisation & Optimisation

A Google Optimize Google Marketing Platform Sydney meetup

We spoke about how Google Optimize can help you take action on Google Analytics insights and deliver personalised site experiences to your most important users.
See the event slides below and find out how to get started with Google Optimize to improve your marketing in less than 1 week.
You’ll learn:

• Why every marketer should be using Google Optimize
• How Google Optimize is so powerful
• Google Optimize basics
• The statistical model it uses (Bayesian vs. frequentist)
• How to get up and running with Google Optimize to make your marketing improve in under a week

Our Meetup Objectives

1. Share Google product expertise
2. Share digital marketing expertise
3. Increase the adoption of best practices
4. Build a community of Google marketers

Want to join our next meetup?

Join the Google Marketing Platform – Sydney meetup here.

Google Marketing Platform News

Dynamic audiences in Google Analytics for Firebase
Audience builder enhancements include dynamic audience evaluation, audience exclusion and membership duration.
Get more info.

Personalisation is now available in Google Optimize
We’ll cover this one in more depth below …
Get more info.

New Google Analytics Course: Analytics for Power Users
Use analytics to improve website content, optimize your checkout flow, and focus your marketing strategy.
Get more info.

Driving Growth with Personalisation & Optimisation – Google Optimize

How to increase your digital marketing performance in just a few hours using Google Optimize.

Testing and Optimisation Before the Revolution

This is what testing used to look like before online tools. It took teams of people to do relatively simple tests.
Teams included:

• Marketing teams
• Research Executives
• Copywriters
• Designers
• Printers
• CATI researchers
• Research Analysts
• Statisticians  

Testing + Optimisation Revolution

Let me give you an example:

• In 2009 I managed an optimisation campaign for the BBC
• The campaign included TV, press, outdoor, Digital and Direct Mail
• Response method was post
• Our objective was to increase response rates from >80% to 100%

3 Month Process:

• Consulted with leading direct marketers
• Focus groups
• Eye tracking study – on control vs A/B/n

Team:

• Marketing Team (BBC and UK government)
• Data team (Gov. to run segmentation)
• CRM team
• Consumer behaviour consultant
• Marketing Executive
• Research Executive
• Research Analyst
• CATI researchers (x 50)
• Statistician
• Copywriter
• Designer
• Printer
• Direct Mail Production

In case you’re wondering – the campaign improved significantly. It did not hit 100%.

Optimisation Foundations – Cultivating Success

Optimisation Culture

• Before you start optimising its important to consider optimisation culture
• Overlook this step at your peril – it is vital to the success of optimisation efforts
• The culture requirements will vary depending on the size and structure of your business
  • Engage stakeholders: good testing usually goes deep, and is likely to affect areas across business functions. Plot who is involved and who you need to be brought into your product.
    • Pricing
      • Who can help you reduce the price from $1,000 to $900?
      • Who can help you split payments over a number of months using Zip pay?
      • Who can help you authorise a free trial?
    • Product
      • Who can help you build in a new killer feature?
      • Who can help you take away features that don’t add value?
    • Logistics
      • Who can help you deliver the product the same day rather than in a week?
      • Who can help you give a money back guarantee?
    • Effective testing is not simple. It is a challenging and collaborative process.

• • Framework – be clear about the testing framework you are going to use
  • Roadmap – define a clear roadmap for the business to follow
  • Performance expectations – be clear about what you expect from every test – add a $ value
  • Teach your stakeholders to understand results
• Learn from every test
• British Land Company – Digital Transformation to improve KPI tracking and performance
  • Managed two stakeholder groups with different functions to get cross-organisation buy into the project
  • Group 1: Snr Management – practical management and decision making
  • Group 2: Snr Management Team (inc. C-suite) – reporting and accountability

• It’s important to set expectations but don’t set them too low.
• The internet will tell you A/B testing doesn’t work – that’s not true.
• People aren’t brave or strategic with their testing.
• They test small. They test non-consequential items.
• If you do good tests, that are well considered, planned and executed you’ll be seeing more wins than losses.
• It takes stakeholder engagement to do the things that will make an impact to your business.

So what can you do…

Get More Winning Tests

4 simple testing rules:

1. Build a robust and disciplined test program – know what you’re going to test, who will be involved, what you expect out of your tests.
2. Understand the causal factors – understand your website, onsite factors, what is driving conversion.
   1. What pages are important to your website?
   2. What factors are driving response?
   3. Avoid RATS.
3. Make big changes.
   1. Changing a headline or the colour of the button will not transform your business.
   2. Big changes will deliver the big impacts you’re looking for:
   3. Change the flow of the information base.
   4. Change a funnel based on observed user behaviour.
4. Make more variations.
   1. Most people head into a test with A/B  (the odds in this case are 50/50).
   2. Get creative and create more solutions based on your hypothesis =.
   3. Do 3, 4, 5, 6 well-considered tests and you’re improving your odds of finding a winning test.

Optimisation Maturity

• Most people want to move from zero to complex testing without the experience.
• This is not a good idea and will likely result in test failure and loss of confidence.
• We advise you move through an optimisation maturity process over the course of a year:
  • Crawl: easy wins, get the confidence of stakeholders.
  • Walk: more complex experiments and personalisation, support of stakeholder.
  • Run: advance experiment and personalisation, collaboration of stakeholders.

Optimisation Frameworks

• There are 3 popular optimisation frameworks
• Each share many of the same theories
• But resonate with different companies

MECLabs

MECLabs is the most scientific approach.
It is a good blend between left brain and right brain thinkers.

• My personal favourite, as I like to quantify with a number or a value.
• Tends to resonate well with CFOs/CIO.
• C = Probability of conversion
• M = Motivation of users
• V = Value (clarity of value proposition)
• I = Incentive to take action
• F = Friction elements of process
• A = Anxiety about entering information

RELISH

Relish is a versatile framework developed by NeuroPower. Based on satisfying base level user needs, then moving towards advocacy.

• Relatedness: Clarify the purpose of the communication and the role of the person receiving it.
• Expression: Label the emotion that exists around the issue being communicated.
• Leading the Pack: Be clear about the objective of the communication.
• Interpersonal Connection: Empathise, connect and show that you understand.
• Seeing the Facts: Present key data, facts, information and milestones.
• Hopefulness: Address future implications and outline the next steps.

LIFT

1. Value Proposition: The value proposition is the most important of the six conversion factors as it has the largest potential impact on the conversion rate.
2. Relevance: Does the landing page relate to what the visitor thought they were going to see?
3. Clarity: Does the landing page clearly articulate the value proposition and call-to-action?
4. Urgency: Is there an indication that the action needs to be taken now?
5. Anxiety: What are potential misgivings the visitor could have about undertaking the conversion action?
6. Distraction: Are there items on the page that could divert the visitor away from the goal?

Google Optimize – Built for Business Growth

It is Not an A/B Testing Tool

Google Optimize is not an A/B testing tool

• Their mission is to help you grow and succeed using your data insights
• They do this using two features:
  • Experimentation – A/B, MVT, redirect testing
  • Personalisation (increasingly)
• We’ll take a closer look at each of these capabilities

The Market

What’s going on in the market:

• There are a number of other products
• Leaders
  • Optimizely
  • VWO
  • How does Google Optimize shape up?
• They are very different products
• Google Optimize is definitely more accessible
• Optimizely is seeing a decline in its use – but is still a leader

Google Optimize is Growing

• Google Optimize is growing
• It’s 360 product has more installs and is growing

Why is it growing?

• Simplicity is one reason (one of the most accessible tools)
• The good feature set is another
• Fast and actionable reporting another
• But native integrations with Google tools is the ultimate reason
• The integrations with Google Ads and Analytics are so powerful for 90% of users

As you can see:

• You can target audiences in Google analytics
• Then use Google Ads to bring the right audience
• Then you can personalise or test experiences to drive conversion
• This is as seamless as it looks
• No other tools have this level of integration out of the box

Google Analytics and Google Optimize

What this means to you as marketers is you get the full power of Analytics and Optimize.

• You can access deep insights through Google Analytics
• Turn those insights into actions with Google Optimize

Google Optimize – Google Marketing Platform Case Studies

Let’s talk about Experience and Personalisation with some real case studies.
Experiments:

• Google Optimize helps you test an idea  on your site to see if it improves performance
• It does this using A/B testing, MVT and URL redirects
• You can make all the changes to your page using Google Optimize
  • You can change text
  • Change images
  • You can add or remove sections
  • You can change button function

Case Study

Here’s an example of how experiments can improve business performance.
Mango used experiments to improve their mobile shopping experience and increase revenue.
Mango saw an increase in traffic on mobile. But no uplift in revenue from mobile.

• They created a simple test
• Add to cart button early in the flow for mobile users
• One change had a big impact

The outcome:

• 34% increase in products added to cart
• 3.9% increase in mobile revenue

Personalisation in Google Optimize Google Marketing Platform

Personalisation

• With Google Optimize you can create personalised landing pages targeted for specific audiences
• Like this one, which uses your geographic data to change the welcome message
• Really simple and effective

There are a wide range of attributes which you can use to personalise a digital experience
Simply, you can use

• Google Ads, on-site behaviour, geography or technology

You can go more complex

• Use javascript variables
• First party cookies e.g. if they have been to the site before, welcome them back
• Data layer variables e.g. action specific events on the submission of a form or download of a resource

Start with the basics; it will increase your confidence and get early wins.

Case Study

Sigma sports used personalisation to increase sales by delivering a personalised homepage experience.
Why?

• Research showed that 40% of users were returning users
• Only 2% of users were using the primary home page features
• The home page didn’t work for the returning users

Solution:

• Created 3 distinct audiences in Google Analytics – based on brand purchase or interaction
• These were used as targeting rules in Optimize 360
• For example; if you bought Specialized in the past, the home page carousel would be focused on Specialized products

The Result:

• Increased e-commerce conversions by 32%
• Increase revenue by 28%

Google Ads + Google Optimize Google Marketing Platform

Combining Google Ads + Optimize is where things get really interesting.
Let’s take a look at a customer journey:

Customer Journey

• Dave is in the market for a vacation
• Jenny is a monthly business traveller
• They both spend a lot of time looking at different travel web pages

As marketers, we spend a lot of time creating personalised adverts to get them interested in our service or products.

• But then once we’ve got them interested, we often serve them a very generic web page
• In this example, this page is better suited to Dave than Jenny
• Dave converts and Jenny doesn’t. 50% conversion!

This is where Google Optimize comes in. We can use Optimize in collaboration with Ads to personalise the experience for both Dave and Jenny.
So now…

• When Dave has responded to a vacation ad, he lands on a vacation page which meets his needs.
• When Jenny interacts with business traveller ads, she lands on a business travel page.
• In this case, both users are happy and both users purchase. 100% conversion!

Google Ads + Google Optimize Google Marketing Platform Case Study

Spotify is a great example of how to use search ads and optimize to get results.

• Using Google Analytics Spotify noticed their users in Germany were more interested in Audiobooks than playlists
• Using Google optimize:
  • Changed landing pages to show the wide range of audiobooks for those specific audiences
  • They targeted ad click traffic to show the audiobook experience
• This drove a 24% increase in premium subscriptions in Germany

Smart Reporting – Getting Results Faster in Google Optimize Google Marketing Platform

• Google Optimize comes with fast and actionable reporting built in.
• I’m going to give you a brief introduction to the statistical method and why it’s the right solution for most marketers.

All Statical Solutions are Not Equal

When it comes to measuring statistical inference, there are two distinct schools of thought.

• 1. Frequentist
  2. Bayesian

It’s important you know which method your marketing tool uses.

• Google Optimize (and others) uses Bayesian
• I’ll explain the reason why
• Although, you’ll see it doesn’t always make a difference (in this context)
• But one is more actionable than the other

Bayesian Statistics

What is Bayesian statistics?

• It’s a mathematical approach to calculating probability
• It users prior knowledge
• Conclusions are updated as more data is available

Frequentist Statistics

What is frequentist statistics?

• It’s also a mathematical approach to calculating probability
• Unlike Bayesian it is calculated by analysing the frequency of random events in a long run of repeated trials
• Conclusions are considered to be objective

Clear as mud! Let me give you an example …

Frequentist vs Bayesian

Example: You visit the Doctor with a pain in your stomach.

• Doctor A is a frequentist
• • • He has a model for diagnosing
    • He probes with his fingers and gets feedback
    • Your feedback would be an input into his model
    • Until he comes up with a diagnosis

The diagnosis is based on the Information available at the time

• Doctor B is a Bayesian
• • • She also has a model
    • She also has your patient history
    • Which states you’ve had a pain in your stomach in the past
    • She would go straight to the source and probe and get your feedback as an input
    • She would use the feedback in combination with history to determine her diagnosis

You could argue it is likely her model would be faster to determine the diagnosis and with a higher probability of accuracy.

Output is Important

The most important thing for non-statisticians like me is the actionability of the data:

• You can see an example here:
  • The frequentist method provides a P-score
    • It has a P- value of 0.01 or a P-value of 0.09
    • What action do you take?
  • Whereas the Bayesian method provides a probability as a %
    • It is 80% or 90% probably
    • What action do you take?

Example

Let’s apply both to a testing example to demonstrate the difference.
You have two websites:

1. Original
2. Test

You want to run a test to find out if your layout change on test A is better than the control.

The Test – Cycle 1

After 7 days (or 1 business cycle) it looks like A is winning.

• Bayesian
  • You get a probability of 89% chance of winning
  • This looks great
• Frequentist = P-value of 0.120
  • Your threshold is 0.05 – this doesn’t look good
• What do you do?
  • Keep testing
  • This is just one business cycle

The Test – Cycle 3

After 21 days (3 cycles) you take another look.
It’s not so obvious this time from the chart.

• However, if you look at the frequentist and Bayesian results there is a more definite answer
• Both indicate the test won
  • Bayesian with a probability of 98%
  • The frequentist with a P-value of 0.01 (under the threshold of 0.05)
• The takeaway
  • Both gave the same answer
  • However, the Bayesian got there first
  • Bayesian inference was stronger to start and got stronger over time
  • Bayesian is more actionable with a probability score

Actionable Reporting

Google uses Bayesian.
As a marketer, you should know …

• You can work with lower samples
• Get clearer results from your tests
• You will be able to communicate the result with greater ease

THANK YOU

Thanks again to Google for hosting our Google Optimize Google Marketing Platform Sydney event on Driving Growth with Personalisation & Optimisation. Many thanks to everyone that attended and of course, don’t miss out on any upcoming events like this one, by joining the Google Marketing Platform Sydney meetup.