Big data and avoiding analysis paralysis
Share this article
Wow, we’ve got a lot of data. In fact, we’ve got astronomical amounts of data. With each channel and device and app and site we add, the more of it we get; every day it grows bigger. We add data to augment our data (like CRM data) which in turn creates even more data. A single web page view generates 475 bits of data for Adobe SiteCatalyst to process. Don’t believe me? Download your clickstream file and you’ll see what I mean.
A single web page view generates 475 bits of data for Adobe SiteCatalyst to process.
But that’s not the point of this post.
The point is, with so much data, we can very quickly become entirely overwhelmed and end up letting in the arch enemy of insights, “analysis-paralysis”. And you know what happens then… you blindly go down alleys, you look at things you shouldn’t, you get distracted, you forget about what the important things are, and you generally get really unproductive.
5 top tips to avoid analysis paralysis
It’s not quantity, it’s quality. It’s what you do with it that counts.
Here are 5 key things to help you focus and keep your head above water when dealing with a sea of data:
Go back to the KPI’s. The goals. The objectives. Make sure your goals and objectives are clearly defined, preferably at the business and the digital level. Your digital goals and objectives should be aligned to your overall business objectives. Make sure that you can clearly measure the goals. If you can’t clearly measure them, you don’t have a proper KPI.
It’s highly likely that you’ll have different objectives at different stages of the brand journey and the customer lifecycle – define them, agree on them, and stick to them.
2) Data accuracy
If your data isn’t accurate, you’ll quickly run into problems. You’ll spend all your time either trying to validate your data 9 ways from Sunday, or trying to convince everyone it’s correct. If people challenge your data they will also challenge your insights. And there’s nothing worse than being in a room full of people when someone says ‘I don’t agree with X’, or ‘Why does my internal data differ significantly’ – you suddenly find yourself switching to justification mode, rather than storytelling mode. So, make sure your data is as accurate as possible.
Stay on top of the browser changes, stay on top of code changes, match your data to internal data – you can be within 1% with a bit of work. Re-validate your entire data set at least every quarter. It might seem a lot to do, but if you can justify your data you’ll be saving yourself days of trouble down the road.
3) One source
Get everyone using the same dataset. More than one source and what’s that noise you hear? Oh, that’s just the death-knell ringing in the background. This leads back to my previous point about making sure everyone is using the same dataset. SiteCatalyst is designed to be that single source of truth…so use it. If you’ve designed it properly and you’ve made sure your data is as accurate as possible, there’s no reason why everyone shouldn’t be singing off the same song sheet.
This does take a little work though. Agencies can be a little tricky here, especially when it comes to claiming credit – so make sure you have different attribution models in place and you’re not just using Last Touch (or First Touch). 63% of consumers come to your site through multiple channels before they convert – so understanding that conversion journey is important (and that’s where attribution comes in).
As the owner of digital analytics, you have to be an evangelist for the power of your data. At every opportunity you have to be talking rationale and possibilities. The reason why you and your team do what you do, and the possibilities that it can generate. You have to be continuously showing the business the opportunities based on your insights from your data.
4) Meet on a regular basis
One technique that works well is to make sure that everyone meets on a regular basis – weekly or monthly – to go through the data and talk about what is being seen. Yes, questions will come about, but providing your data is accurate, it’s a true source and you’re aligned to your KPI’s then these sessions should be invaluable to the business stakeholders. You should also be able to generate plans for the next month, quarter etc.
Testing should play a big part of this too. You should be thinking about scenarios to test, running those tests, reporting back on the results, and generating new tests. Make sure you have key stakeholders at these sessions – a cross functional team from digital, marketing, social, BI, call centre – as they will give them more value than just sending out a weekly report. This also ties back into you being an evangelist – if you’re not excited about your data, maybe you’re in the wrong job.
5) Know your audience
Finally, you have to know your audience. The way you “report” is based entirely on your audience. I recently came back from Summit in the US and was entirely inspired by the Data StoryTelling. Make sure that you tell stories from your data. People connect to stories but they challenge data. Think about the “aha” moment you want them to have. Don’t fill it full of data. Tell the story. Tell them how and why but don’t worry about the what (you have that in reports and analysis). Get them to the “aha” and then make recommendations for next steps. Your stakeholders are likely all business people; traditional and digital marketers, social managers, strategists, planners etc. They don’t want data (numbers) – they want thoughtful plans and recommendations based on insights – and insights are not data. Insights are based on data.
Deliver a difference
So yes, we have quite a lot of data, and it’s getting bigger. Our job is to sift through it, remain focused, and pull out those nuggets that help us plan and really deliver a difference to the business based on what we’re observing.