Secrets of Analytics Success – Collecting and Measuring the Right Data – Google Marketing Platform Sydney Meetup Recap

Last week we held our first of many Google Marketing Platform Sydney meetups. We talked about Secrets of Analytics Success – Collecting and Measuring the Right Data.

Sign up for our upcoming events

https://www.meetup.com/Google-Marketing-Platform-Sydney/

Secrets of Analytics Success – Collecting and Measuring the Right Data – Google Marketing Platform Sydney Meetup Recap

The difference between discovering good insights and great insights is the data you have available for analysis, yet many marketers neglect to take a strategic approach to data collection. At this event, our expert speaker provided an introduction to creating an effective measurement plan that will deliver online marketing success.
Joining live from France our speaker, Benoit Weber, is a Google Tag Manager and Analytics specialist. Benoit has worked with leading online brands including Gumtree, Expedia and Open Colleges.
Watch the presentation below:

View the slides:

Google Marketing Platform Sydney Meetup Recap

View our slides and notes below.

Thank you, everyone, for joining today.
2 weeks ago I attended SuperWeek in Hungary. For 5 days, digital marketing professionals, analysts and leaders of the measurement industry were gathered in a mountain resort, discussing data collection, machine learning, data ethics and security = in other words: MY DREAM.
I really invite everyone to have a look at the different speakers as most of them are influencers and leaders that I am personally following.

Doctor Who is my favourite series. I discovered this while I was backpacking through Asia back in 2014.
The Doctor is a Time Lord (an alien) exploring the universe in a time-travelling space ship (Tardis). The spaceship has the shape of a police box. He travels the universe with a number of companions to help people/civilisations in need.
Every episode starts with the Doctor and his team landing on a new planet, a new place in the universe that is facing a crisis. With the help of his companions, he tries to understand the situation, questions everything, he never makes any assumptions about what he knows or doesn’t, nor assumptions on the reason why people are acting in bad/good way, analyses the environment and uses his tools (screwdriver, or tardis) only when required.
He asks thousands of questions, acts a little crazy and weird but always ends up saving the world. Every once in a while, he needs to regenerate. Like everyone else, he can get hurt. If he is too badly harmed to heal normally, the Time Lord “transforms” into a new body. Since 2005, the Doctor has regenerated 5 times already.
Parallel with my world: the Doctor became my model. I’d love to think that one day I will be as cool as the Doctor.
Data Analysts:

• Are a bit weird: Obsess about data, accuracy, collection, interpretation
• We need to solve problems without making any assumptions on how to solve it or what tool to use
• Don’t be too confident in what you know and what will work
• Find out what matters to your colleague and to the different teams, understand how they work, what they report on
• Regenerate = In every work environment (in-house or agency), you will get hurt (personally or professionally) but you will need to get back on your feet, propose an alternative, find the best solutions to answer their needs
• The main objective will be to solve our issue or answer questions on our business = Basically save the universe

Data Governance: Definition, state of the art, how the others are doing it
Measurement Plan: What’s important and the process to build it
What to Measure: What’s really important to measure and how to prioritise and split the work into batches


Definition: “Data Governance is the exercise of authority and control (planning, monitoring, and enforcement) over the management of data assets”: Global Data Management Community.
Here are the top 8 concepts that are included in data governance:
Today we are going to focus on Google Analytics data but this should be applied on every data point you collect within your business.
The 3 most important points (according to me) are privacy, security and quality.
Privacy is the keystone of any sort of data collection = be ethical and don’t collect without consent.
You also have to make sure your data is secure and there is no breach in your data.
Third would be quality, as it will directly impact your decisions –  bad data just looks like good data and there is nothing worse than making the wrong decision because of erroneous data.
Do you have any sort of governance? Do you have anything in place that ensures that every point listed here is covered? Who is responsible for it?
There is no one solution but every business should have their own governance and for digital analytics, it starts with a measurement plan.

This is a study done by the Digital Analytics Association back in 2017. This was done on more than 800 professionals: team managers and directors, and staff in analyst and statistician roles, at organisations across a wide variety of industries.
The scary reality is that a bit less than 25% have governance.
The main reasons presented were:

1. Organisation politics: 18%
2. Lack of resources: 18%
3. Limited understanding of value: 17%
4. Not seen as important by senior management: 11%
5. Believe that “it’s someone else’s problem”: 7%
6. Considered too complex: 6%

This has, unfortunately, had a direct impact on the current state of analytics (Google Analytics).
This is a study that was done by Brian Clifton (Successful analytics (2015), advanced web metrics with GA (2008). and presented during Superweek.  

1. Collection of personally identifiable information: PII is prohibited and forbidden by GA. Your Google Analytics account can be terminated without any notice as it goes against Google policy.
2. Only 1 out of 4 sites properly tracked conversions. You need proper goals to be able to evaluate how well you are doing. Are users doing what you are expecting them to do on your site? Without conversion tracking, you cannot really action anything and therefore improve your business.
3. Less than 1 in 5 has the correct campaign tagging. This makes it hard to monitor your campaign performance and review your return on investment.

Data quality is in an investment.
You need to know your data is solid OR… eradicate your doubts. The problem is not so much about errors in your tracking but not knowing that there is one.
Question and challenge your current analytics implementation. Uncover gaps and errors in your current measurement.

To be able to solve a lot of these problems, the first thing to do will be to define what matters to you and therefore what to measure: this is what we call the measurement plan.

We need to find out how to make sense of our data but always keep in mind the why.
The ultimate goal will be to have direct impacts on your business.
Data is useless unless you can action it to make your business better. For this, you will need to give sense to your data.
To be able to do so, you need to juggle between having good biz expertise, tool expertise and maturity/culture.

1. Biz expertise: Figure out a good strategy to use, what we should do – measurement plan, solutions, strategies
2. Tool expertise: We need to be confident with the tool but it is meaningless without strategies
3. Maturity: Communication, participation, transparency, purpose, agility, shared goals

The merging of the three will be your own Tardis, your way towards impacts.

This is the optimisation framework. It helps to put data collection back into a bigger frame.
This is presented in the book  ‘Google Analytics Breakthrough’ Feras Alhlou, Shiraz Asif, Eric Fettman

• Strategy
  • Develop roadmap and KPIs
  • Measurement Plan, Discovery and Audit
• Implementation
  • Measure what matters
  • Tag Management, Analytics, testing and qualitative tools
• Reporting
  • GA and Supplemental Tools
  • Dashboards, custom & standard reporting, Data Integration and Data Visualisation
• Analysis
  • Derive actionable insights
  • What is and what is not working?
  • Goals, e-commerce, segmentation, traffic sources
• Optimisation
  • Testing and personalisation based on qualitative and quantitative inputs
• Business Impact
  • Increase revenue, reduce cost, improve loyalty

Data Collection is the keystone.
You can’t interpret and derive actionable insights unless you have and can rely on your data collection.
Collect
Implement GA so it gathers the necessary data to get insights beyond the most basic tracking such as video plays, user interactions, e-commerce …
Aggregate
Limit unnecessary fragmentation and capture data with classifications that reflect the way you view your organisation, your content and customers > settings, filters, content groupings and custom dimensions.
Segment
Consolidate and reclassify data as needed. Break down your data into logical segments so you can get better and faster insights.
Integrate
Integrate your marketing data (Adwords, AdSense, DoubleClick) and CRM with your GA data, so you can accurately calculate marketing ROI. Better correlate web/app interactions with offline customer engagement and lifetime value.
Visualise
Visualisation is essential to effective communication. Displaying data in graph format is often more efficient and specific than default tables – for your colleagues, managers and yourself.
Interpret
Understand the causality behind the data. It’s the most difficult, but also the most important challenge that analysts face. With both quantitative data, qualitative data (surveys, user testing) understand what is not working. Create a hypothesis on why/how to optimise your site, to improve usability and conversions. Test and validate observations to optimise profitability.

Avinash Kaushik said:

There is one difference between winners and losers when it comes to web analytics.
Winners, well before they think data or tool, have a well structured Digital Marketing & Measurement Model.
Losers don’t.
We often make the mistake of focusing on tools before strategy.  We need to start by defining our measurement plan.

1. Identify the business objectives
2. Identify goals for each business objective
3. Define meaningful metrics and KPIs for each goal
4. Define targets for each KPI
5. Identify important dimensions
6. Identify the segments of people/behaviour/outcomes that we’ll analyse to understand why we succeeded or failed

A measurement plan is everything but static.  You will have to make sure that it is evolving and adapting to your needs.
This chart shows us the 5 steps.
Step 1: Define a measurement plan that contains your objectives, goals, KPIs, targets and segments. If you don’t have one yet, build the first version and then improve it incrementally.
Step 2: Document it. Everything you need and have, keep it up to date, help you plan and prioritise. Split your implementation into a batch.
Step 3: Create the implementation plan that will translate your needs into technical specifications (data layers, containers, pixels) that you can give to IT or your agency.
Step 4: Implement according to the implementation plan.
Step 5: Maintain and refine your data collection every day.
At a minimum, it will take you 2 weeks to draft the first run with an easy measurement plan that does not require too much implementation if you have the knowledge.
If your implementation is more complex (meaning form tracking or e-commerce), you will simply split your measurement plan into different batches and based on your IT pace (sprints) you will synchronise and plan new implementation to roll out at your pace – around 3-4 weeks.
If you don’t have one yet, build a first version and then improve on it incrementally.

To start defining your measurement plan, don’t factor too much on what you know. The most important would be for you to find what is missing, what are the current gaps in your measurement?
Finding out what is missing will enable you to widen your data collection if you consider it worth it.
Find out your unknown unknown.

• What we know that we know: Assumptions
• What we know that we don’t know: Gaps
• What we don’t know that we know: Tacit knowledge
• What we don’t know that we don’t know: Discoveries

This is not an easy exercise, and this is almost impossible to figure out without questioning everything you know. 

Reports that say that something hasn’t happened are always interesting to me, because as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we don’t know we don’t know. And if one looks throughout the history of our country and other free countries, it is the latter category that tend to be the difficult ones. – Donald Rumsfeld, United States Secretery of Defense, 2002

This is an easy framework: The 5 Ws and the H.
Broaden your knowledge and don’t make any assumptions on what’s important and what’s not. Not only question what is needed but also what already exists.

• Who: Who is the data source? Is it reliable? Are they multiples sources? What’s their relationship? Who will use the data?
• When? When did we start collecting the data? Are there downtimes? Is it reliable throughout the timeframe? Were there major changes to the site over this period?
• How? How is it being collected? What’s the methodology? Can it be trusted? Are there obstacles?
• Where? Where does the data need to reside? GA Only? CRM? Other? Do we need DataStudio? Custom Reports?
• WHAT? What is the data about? What story do I need to tell? Are we missing data to tell the whole story? Do we need data for pixels?
• Why? Why are we doing this? Enhancing or fixing? What is the main purpose of the project?

KEEP IT SIMPLE STUPID
But sometimes we forget that we are not alone on the ship and we forget to ask what matters to others.

Everyone in your team or your work environment (including your agency) shares the same goal: Make your business better. Sometimes we forget about this.
Ensuring that everyone is involved and participating in the elaboration of the measurement plan, not only ensures that we include everything but also that the team is aware of what is needed and what they have to do.
It is often that teams are working on specific verticals: PPC, SEO, Email. And it is often that we forget or are not aware of the activities or campaigns run by others.
Example: I was in contact with only 1 person and the team that was in charge of email was never part of the discussion that I had while defining and when I presented my measurement plan. I provided conventions for emails to ensure that emails were properly tagged. Unless you do this, this will be invisible to GA. I gave the specifications but somehow they got lost in translation and were never reported to the team responsible. Nobody realised that emails were not being reported in GA due to errors in tagging and that everything was ending as Direct. 100,000s of emails were sent every month, but there was no visibility on how many users actually accessed the site and converted.
Unless we include everyone in the discussion and share the plan and strategy with everyone, we may simply miss insights and some recommendations may not make it to the concerned team/person.
Once we have questioned everything and reached out to everyone in the team, we will end up with a much clearer picture of what is important to the business.

The first step will be to define your business objectives.

• What are we hoping to accomplish for our business?
• What’s important to our business?
• What’s important to our customers?
• What’s the core and unique value proposition that we offer?
• What defines success for our organisation?
• How are we trying to achieve success?
• What are the three most important priorities for your site?

DUMB: there are 2 meanings – keep it simple stupid but also doable understandable manageable and beneficial.
Here are few examples of generic goals, and I would recommend you to not overdo or overthink your objectives. At IMWT, we generally end up having 3 main objectives.

Define DUMB objectives.
The objectives must be DUMB: Doable. Understandable. Manageable. Beneficial.
Make money, save money, make them happy.
For each objective, you will need to define goals (different strings that you can pull).

• Specific – target a specific area for improvement.
• Measurable – quantify or at least suggest an indicator of progress.
• Assignable – specify who will do it.
• Realistic – state what results can realistically be achieved, given available resources.
• Time-related – specify when the result(s) can be achieved.

For each objective, there are always different perspectives and strategies to achieve a unique goal. Here, are 2 goals per objective.
Try to keep it simple.

Time span: 1 quarter, 1 semester or 1 year.

To be able to drill down, break down and review performances for different segments of people/behaviour/outcomes, you will require dimensions.

So far, we discussed with everyone, we questioned everything we knew, we defined our business objectives, translated them into goals, listed important metrics, listed our main KPIs with achievable targets and specified important dimensions that will enable us to segment our users based on their behaviour, status, location… 

Here is an example of what we could do with an e-commerce site that has a blog.

• Adjusted bounce
• Calculated metrics to evaluate the real active time on page
• Enhanced e-commerce with checkout funnel and list of performance reports
• Heatmaps and session recording through Hotjar

But the problem now is that we will often end up with this exhaustive list of what we would like to measure.
But what really matters? Could we prioritise?
Example with a client with a media spend of 2 Million and 200 metrics. When we started working we realised that only 20 signals and 4 metrics were actually important.
Only keep metrics that make sense and that are essential for management to run their day to day monitoring.

How much can I get rid of? How much do I really need now?
Try to eliminate anything not mandatory and relevant. Try to define your minimal viable measurement plan.
Keep it simple, stupid.

Some people want to go fast and implement everything but experience shows us that it is better to split into smaller projects.
Step by Step.
Smaller changes with clear directions are easier to implement.
Find out what is your minimal viable measurement plan, start with this, build your business case to convince people to get onboard with your data collection project, and build on top of it.
Start with the minimum and enhanced measurements will come later on a regular basis. If you work in batches you won’t be impacted by the dependency of tasks.
Example: One of my clients has a publisher site. We defined a full measurement plan including enhanced-e-commerce for the publisher. We considered a blog post as a product, the reading of the article as checkout steps, the transaction as the full consumption of the article and content grouping. The client got really enthusiastic and validated the project. We wrote all the technical specifications for the technical team to implement (meaning the datalayers). One thing we did not plan was the amount of work and complexity that this would require for the technical team. As they did not have clear templates for articles, they ended up having to manually implement the datalayers. A project that was really cool on paper ended up as a nightmare for validations and QA as we are dealing with thousands of articles, each having different datalayers implemented.
The project that was supposed to take 2 months (according to their estimation) ended up lasting for almost 10 months.
Make sure you split your project into feasible batches with no dependency, taking into consideration external restrictions (technical or political). Start with your minimal viable measurement plan and build enhanced measurements step by step.

So how do we prioritise?

STEP BY STEP
The MVA: Most Valuable Answer

• What strategic question do we need to be able to answer this year?
• What data do we need to be able to answer it?

1. Potential: Heuristic and Voice of the Client/Customer/Boss
2. Importance: Traffic Volume, Cost, Impact on
3. Ease: technical ease, political issues, number of parties involved

Don’t make assumptions on what you need and know, involve everyone, communication is key, encourage multi-disciplinary teams, share goals, ask for more involvement from everyone.
Transparency, involvement, shared knowledge and skills and constant learning.

Think strategy before you think tool.
Take one step back and find out what you want to achieve before you plan how to do it with your tool.     

Exterminate, annihilate, DESTROY!!!
The Daleks are part of Doctor Who. Their only objective is to destroy and exterminate everything.
Sometimes you will have to go to the dark side and kill any source of failure: avoid organisation failure, client failure, self-failure, technology failure, failure to understand requirements, failure to follow logic, failure to deploy.
Question and challenge everything.

And lastly, I just want to reemphasize the importance of your team and on collaboration.

Thank you for joining us and don’t forget to sign up for our upcoming Google Marketing Platform Sydney Meetups.
https://www.meetup.com/Google-Marketing-Platform-Sydney/