2. Author: Charles Meaden
Obligatory Bit About Me
• Been involved in analytics for 25 years now
• SEO and Usability are my other core areas
• Based down in Mumbles, South Wales
– No, that’s not filtered
3. Author: Charles Meaden
A Lot Of The Things I Do Are A Process
• To get my work done I need to extract data from lots of different systems
– Google Analytics
– Crawlers such as Screaming Frog and Sitebulb
– Google Search Console
– Ecommerce systems
• Then transform (clean!) it
• Before invariably loading it another tool
4. Author: Charles Meaden
This Talk Is Platform / Tool Agnostic
• I can’t code for toffee
• But I can work out logical steps
• My preferred tool is Analytics Edge for Excel
5. Author: Charles Meaden
Statement of the Bleeding Obvious – Why Automate
Because so many of the
jobs that we do consist of
the
same steps every time
6. Author: Charles Meaden
Example – Extracting All URL’s That Contain a Query String
• We do Google Analytics audits
• One of the tasks is to
– Work out how many URL’s contain a query string
– Which values are shown in the query strings
– How many page views does it affect
• To get 3 months worth of data out of Google Analytics is at least 10 clicks
to get data I can work with
– Plus the time waiting for Google Analytics to go from screen to screen
– Then, I’ve got to merge the files together
• In Analytics Edge, I’ve got a ready made macro that does the above, plus
– Uses regex to split and identify each query string
– Calculate the number of different values for each query string
– Calculate how many page views if affects
• Result – I get to the data a lot quicker
7. Author: Charles Meaden
15 Minutes to Write this Macro
• Any tool has a
learning curve
• Once you’re
confident with
one,
automation
becomes a lot
easier
• This macro
took 15
minutes to
create from
start to finish
8. Author: Charles Meaden
Lesson 1: Focus on the end result
• Forget for a second what tool you’re going to use
• Have a clear idea of what it is end result going to be
• Also who is going to be using it
• Sometimes working backwards helps to uncover new methods
9. Author: Charles Meaden
Lesson 2: It’s Always Going To Take Longer
• Ideas that seemed great in theory, often taken longer
• Things are going to get in your way such as messy data or systems
changing the format
• This can be a good thing – some of my best solutions came from having to
adapt previous solutions
10. Author: Charles Meaden
Lesson 3: Stand On The Shoulder of Giants
• As Isacc Newton once said
• Just like we’ve all learnt from Simo Ahava how
to use Google Tag Manager
• Dig into the Python, R or whatever tools you
don’t use and see what people are doing there
• For a user intent tool, we’ve just replicated a
stemming library to remove plurals inside Excel
11. Author: Charles Meaden
Lesson 4: Accept that Data Will Always Be Dirty
• The first thing we always look for in any data set is the variations from the
norm
• Things such as
– Upper and lower cases
– Misspellings
– Wrong formats
• It’s an ongoing process especially if humans are involved
• Build out processes to trap those errors as early as possible in the process
12. Author: Charles Meaden
Lesson 5: Build A Library
• Build a library of all the parts that you use
• As well as the code, write down the process
you used
• My favourite tools for this are
– Evernote as I can add tags to everything
– LucidChart for building quick flow charts
13. Author: Charles Meaden
Lesson 6: Accept That Not Everything Can Be Automated
• A lot of my processes are semi
automated
• At some point in the process, I need to
quickly check for outliers
• I have routines that clean and then
identify anything that it new
• Especially true when dealing with
campaign tracking or new variables that
someone forgot to mention
14. Author: Charles Meaden
Lesson 8: Use Your Eyes
• Before sending the results of your data
off to someone else, take a look
• Does it make sense?
• What could you improve?
15. Author: Charles Meaden
Someone didn’t here…
• This is an anonymised
version of a chart that a
digital agency sent us
• No other documentation
to explain what it was
• A classic example of a
‘data puke’
• A quick eyeball of this
would have spotted that it
wouldn’t make sense
16. Author: Charles Meaden
Lesson 9: Educate Everyone About Benefits
• Don’t hide your talents
• Make it clear that while automation has a cost, the ROI can be amazing
• That it frees people up to actually do things with the data
• Once people see what can be achieved, more projects will come your way
17. Author: Charles Meaden
Lesson 10: How Can I Build On This
• Take a look at the process
• Encourage feedback from users
• Continually evolve the process
18. Author: Charles Meaden
Tools That We Use
• Analytics Edge for Excel is the tool that we use most frequently
– Best £150 we spend each year
– Technical support is superb
• Others have raved about SuperMetrics
• Google Data Studio and Sheets have some really good automation
features
19. Author: Charles Meaden
Thank You
• You can find me here
– Charles@digitalnation.co.uk
– https://twitter.com/charlesmeaden
– https://www.linkedin.com/in/charlesmeaden/