I’ve dedicated 2014 to the topic of analytics hygiene. As a result, I am noticing some folks who are interested in “digital hygiene” … which appears to be a fascinating topic around personal data, privacy and publicness. Totally different from what I am talking about. I am referring to a general desire to rid analytics programs of diseased reports, processes, and data… and some methods to help with that.
Governance is one thing, hygiene is another.
As I define it, analytics hygiene could easily be understood as something similar to governance; there is overlap between the two. It can be hard to pick them apart, even. The complexity there is probably finding a firm definition of what governance is as a starting point. Judah Phillips gets at it pretty well in his mammoth, Building a Digital Analytics Organization. I would quote the piece now and get into it, but that digression will cost me hundreds of words. Thank me later (maybe grab that title and keep it around as reference material, it’s pretty intense!).
Well, what is it then?
I’ve seen, too many times, poor quality get in the way when it could have been corrected years, months, weeks, or days before it became a problem. Maybe this has happened to you, or maybe you do not work in analytics. It’s not just data quality. Vanity metrics. Analysis paralysis. These aren’t buzz terms. The aim of analytics hygiene is to help to mitigate against these evils and others.
How do I get started?
You already have analytics hygiene. Is it good, or is it bad? That is the question. You do not need to read my musings to get on a good path. Countless brilliant people have written and spoken about web/digital analytics and actionability, monetization, and cost/benefit ratios… about using appropriate success metrics. One can (and should) reference Eric Peterson’s Big Book of KPIs (free download w/email address) to get a sense of what a decent metric consists of. Pick up Jason Burby’s Actionable Web Analytics to motivate yourself to focus on action.* Read and trust in Gary Angel. Holy crap. Stop reading my drivel and read this (come back please). YES. These tie right into the concept of analytics hygiene.
The principles of analytics hygiene
One can not have a methodology with a super cool name like “analytics hygiene” without having some defined principles… this is my first attempt at putting them together. I think I will come back and write a separate post on each principle, but here’s what I’ve got for now:
A year or two ago, I might have used “resiliency”, “fault tolerance”, or “incremental learning”. These are not good fits for what I mean. Via Antifragile: Things That Gain From Disorder, Nassim Nicholas Taleb enlightened me to a better term: antifragility. Simply put, when something that is antifragile “breaks”, it re-emerges as a stronger being. Think Hydra (reference stolen from the aforementioned).
I’ve yet to be part of a software/data project where antifragility wasn’t of critical importance. Something went wrong? Fix it and place monitors/detection to aid with potential future occurrences. Be proactively antifragile where it makes sense, fixing potential issues as well as actual issues. If you can’t fix it now, document it… otherwise known as stick it into a … framework!
Frameworks are important… and, whether recognized as frameworks or not, they are everywhere. While they certainly would seem to offer the most value in a large organization, you’re going to need frameworks even if you are all alone. Even if you don’t use a preconceived digital or paper template, having a general project framework in mind will help tremendously with planning and prioritization. Framework is a sad word that tends to do the opposite of excite people. Don’t hold that against a good framework. Frameworks are just project outlines that aim to bring consistency, organizational memory, and rigor to whatever they’re applied to.
You’ll need frameworks for how requests for data are handled.
You’ll need frameworks for how data is delivered.
You’ll need frameworks for how analysis is done.
You’ll need frameworks for how data is monitored.
Frameworks provide consistency and rigor.
Frameworks provide organizationally accepted and documented ways of doing things.
Frameworks provide ways to compare how things have been done over time.
Tim Wilson, of Web Analytics Demystified fame, has provided a great framework for optimization that applies universally well to organizations of all shapes and sizes. I hope to come back later to write detailed posts on frameworks as well as the other hygiene principles. When I do, I may spend more time on this one. Frameworks can help to create good organizational habits, but they can also help to perpetuate bad ones. We want to deliver incremental learning or value of another kind in all that we do. Creating frameworks that will not be used is a waste, too. Keep that in mind, and consider your audience when creating or selecting frameworks.
Know the past, present and future. Organizationally. You know… fit it into an omnivision framework. Know the answers to questions like: Why are you doing things the way they are done today? What are the pain points? Where do you see the analytics team and software involved going over the next 1, 3, 5, 10 years?
Part of supporting the mission is generally to do so in a cost-effective manner. You can’t be cost-effective if you don’t know what things cost. The analytics footprint should be placed into the omnivision framework.
Omnivision can not be achieved without outsiders. Otherwise, your vision is biased by the organizational habits and the potentially conflicting goals of individuals.
Outsiders need to be involved. Easy for me to say, as an analytics consultant… but seriously… your organizational habits are just that. You need to step outside of the organization periodically to be sure you are heading in a direction that makes sense. You don’t need to head in the same direction as others, or others like you, are heading. You should be moving in a direction that makes sense given where you are now, and where you will be in those next 1, 3, 5, 10 years. You can bring outsiders in, and you can send people out. Do both, if you can. If you can only do one, send your people out into the world… maybe to eMetrics.
You can get outsiders involved in your program without hiring a consultant. You can get out to a local analytics MeetUp, a DAA get-together, Web Analytics Wednesday, etc. You can hire someone into the company, they’ll be an outsider at least for a few weeks…
Thanks for taking the time to read my first thoughts on what good analytics hygiene consists of. I can imagine that I’ll revisit these after feedback and more thought!
*Don’t worry about web analytics materials more than 5 years old. For as much as has occurred in 5 years, much of the teachings were created to be forward compatible. We have become too dependent on the publishing date of digital analytics content. The bibles are already written, even if they will be re-written over time. Indeed, many of them were written before our industry existed. If you’re focused only on materials published in the last 12-24 months, you are missing out.