Even though the resources are always limited it might be tempting to focus on collecting data the right way instead of collecting the right data.
Lukas shares a structured approach to gather, understand and prioritize requirements for digital analytics projects for larger enterprises using Adobe Analytics, Piwik/Matomo, Snowplow, Exponea/Bloomreach, Meiro to just name a few he has experience with.
SuperWeek 2023 - Building the case for Digital Analytics
1. BUILDING THE CASE FOR DIGITAL
ANALYTICS WITHIN AN ENTERPRISE
A SHIT-SHOW
@cataLuc
lcech
I needed to update my title
-- in reaction to @analyticsninja
3. Regulars at Superweek know that
there should always be a picture of
your dog in the presentation
-- shout to @mgershoff and @fredpike
4. Time
2008 2010 2012 2014 2016 2018 2020 2022
This is my experience in the field –
hopefully this qualifies me to talk here
5. Expected Takeaways
> What is special about enterprises in relation to analytics
> What we might do to tackle some of the issues
> How different solutions may (not) help us with that
> Something to discuss when you catch me without my guitar*)
*) when I have it, it is useless trying
to talk to me
6. Photo credits: Lukáš Bíba, Hospodářské noviny
This is almost the whole population of
Czechia (our market is small + we are
patriots) – me over there
7. Typical customers
:
:
Bigger customers on our market tend
to be owned from abroad – sometimes
this adds to complexity, but I still
enjoy working for them…
11. Too many Cooks spoil the broth
There is many different stakeholders
with variety of backgrounds and
mental models that we need to serve.
Usually also pretty heterogeneous
environment as well – various service
providers, frameworks, systems,...
13. This is "transparent slide"
We always try to say things like they
are – e.g. we need a lot of the
customer's time to capture all the
business requirements as clearly as
possible.
15. So called "Solution Design Reference"
seen with many Adobe Analytics
implementations never worked for us.
Please use more than a spreadsheet
with a list of variable names!
16. Usually the documentation only
captures "the data layer" – which
events and attributes go together and
how to pass them in code.
17. We always overlay this with carefully
documented requirements. Usually
there is also a lack of documentation
around the existing inventory
(websites, apps, channels,
integrations), so we fill that gap and
document these as well.
18. And last but not least there is also
the reporting and analyses that need
curation to be able to maintain a
long-term ability to fuel the
organizations with insights.
19. I am not saying that this is an easy
task to do – even Marie gave up :)
20. ➔ Bold conventions enforcement
➔ Community-driven standardization
➔ More features, less limitations
Complexity: How solutions might help?
👎
Sometimes limitations force us to
simplify and that might be helpful.
But sometimes we don't really cheer
for the enforcements when they are
not performed with caution.
22. Pitfall #1: Any untreated limitation will be exceeded once
We might hit the limits even when
these are 350 custom dimensions and
1000 custom metrics like in Adobe
Analytics if the way of handling these
is not carefully managed.
23. ➔ Bold conventions enforcement
➔ Community-driven standardization
➔ Relatively more features to handle complexity
➔ Standardization driven by the Experience Platform
➔ Robust validation via JSON schemas
Complexity: How solutions might help?
👎
There seems to be a rise of JSON
schemas for purposes of documenting
and validating the data – certainly
something worth to check!
24. After all this is not a rocket science.
Even business users should be able to
work with these, if we just hide the
lodgment of the "developerishness".
26. Give a man a fish and you feed him for a day.
Teach a man to fish and you feed him for a lifetime.
We think of analytics projects not as
a one-time delivery of a solution, but
rather as a form of education.
Photo by Hunter Brumels on Unsplash
27. Managing
Competence
Photo by Ivan Aleksic on Unsplash
Not really in the old-fashioned frontal
classroom style, but modern
on-demand structured ongoing stream
of recorded workshops and exercises.
29. Symbiotic relationships with competent individuals
So we try to find the competent ones,
make them friends and engage them,
so they can help us.
Photo by David Clode on Unsplash
30. Good old propaganda!
We also recommend to start sending a
regular newsletter within the
organization to share the latest
progress and opportunities for
education to further increase the
awareness.
31. ➔ Higher talent availability thanks to openness
➔ Community-driven education
➔ Heavily invested usability of Analysis Workspace
➔ But in general education is harder to get to
➔ Welcome serial for new users
➔ Do less and automate heavily
Competence: How solutions might help?
32. Pitfall #2: We might get too comfy with unclear requirements.
If we feel too confident that we can
answer any business question thanks
to the power of Adobe Analytics
feature set, we then tend to relax in
our effort to get the precise
requirements.
35. No battle plan survives the contact with the enemy
Expectation Reality
Goal Goal
36. No measurement plan survives the contact with the compliance
Expectation Reality
Goal Goal
37. Achieving
Compliance These non-functional requirements are
usually justified really well and that is
a huge opportunity to learn for us as
analysts as well.
Photo by Nick Fewings on Unsplash
38. Our agency always tries to be above
the average of the market in terms of
understanding all the nitty-gritty
nooks and crannies of IT process.
It usually makes also our deliverables
much more reliable and valuable.
39. ➔ Now attracting the attention for all
➔ Unclear situation for cautious enterprises
➔ Most robust privacy governance, better hosting options
➔ But still US company
➔ Ability to run fully on-premise (self-hosted)
➔ Code transparency, high development standards
➔ Ability to run fully on-premise (self-hosted)
Compliance: How solutions might help?
I am not a huge fan of self-hosted. It
usually introduces non-trivial overhead
of maintenance, but I expect to see
this more often thanks to privacy
regulations going crazy.