I have nearly recovered now from Adobe Summit and the fun of being up close & personal with Nina Conti. Going through my notes to write up summaries of the conferences, thought I would start with the five breakout sessions that I attended.
Analytics Rock Stars
This session was a series of tips with the speakers headlined by Brent Dykes of Web Analytics Action Hero fame. He was joined by three practitioners working client side. There were some nice tips presented during this session, sadly only one was new to me.
The tip is relevant to booking engines for travel related organised. It is to group booking dates into useful buckets with the example given of holiday periods e.g. Christmas period, school holidays, summer break, Easter, etc. For SiteCatalyst, this can be accomplished using SAINT although possible directly through the tagging with the help of a good developer.
The value then comes when evaluating performance of campaigns (internal and external) against the date period they prompt bookings (or searches for bookings) for. So your summer campaign would ideally prompt bookings for the summer but this might not be the case.
I would add additional dimensions here of the booking window (time between now and the start date), length of booking, location and value of booking. Comparing conversion rates from search to booking across all of these dimensions could provide some very interesting actionable insights.
Other tips included:
- Ideas for making SiteCatalyst dashboards more readable & user friendly (my advice is still to do it in Excel)
- Use Marketing Channels as your one source of truth for traffic source performance – totally agree
- Create calculated metrics using a combination of metrics and segments (limited by creativity) allowing you to compare performance across multiple areas in one place
- Use a logical structure and initiative names for the Calculated Metrics – this is relevant to everything, especially Page Names but also segments and campaigns
- Whatever the solution/tool, combine all date in one place
- For product merchandising, create a scatterplot of all products across two dimensions with different tactics for products in different quadrants
This was another tips & tricks session (with a theme of time saving) featuring Ben Gaines, everyone’s favourite Adobe Product Manager (plus another client side practitioner). Key tips here were:
- Set up the Key Metrics report for managers for their top 5 KPIs (including segment to apply to them) – so they don’t constantly ask you for the same information
- Use Processing Rules to define variables without the need for dev involvement
- Extend this using Context Data Variables so you don’t need to predefine if a piece of information is an eVar, sProp, etc
- Use segmentation to drill into your data
A basic summary for all of this is quite simple:
Step 1: Do a proper & complete set up of your web analytics tool based on the information you need to know
Step 2: Learn how to use the tool and its features
You will save time and finally start getting value from web analytics.
Customer Analytics Part 1
This session started very promisingly with Simon Ricketts from Channel 4 describing their approach to segmenting their visitors. Interestingly, it does include just exporting all of the data out of SiteCatalyst, instead creating a unstructured data warehouse by combining it with other data sources in a Hadoop environment – allowing for detailed ad hoc analysis.
Visitor behaviour is grouped based on their Recency, Frequency and Value (value defined based on Dwell Time). This leads to about 7 different visitor segments being created based on different types of website behaviour. Segments can then be created in SiteCatalyst to reflect these segments.
The missing element in this session was a detailed example of the process for setting up these segments and applying them to reports. It shouldn’t be too difficult to do (within Discover) but would have been nice to have been shown this. I was also unclear what proportion of visitors login across the Channel 4 properties so if this analysis applies to 50% of their traffic or only 5%.
This is a session I was genuinely looking forward to. I have been arguing for a while that Attribution is not possible (whatever the cost of the tool/consultancy) and that we need to find alternative approaches to evaluating and optimising marketing campaigns. Econometrics sounds like a possible option for this.
And the first 10 minutes was great. Dr Sid Shah definitely knows what he is talking about and introduced the ideas of Econometrics in a nice simple to understand manner. The basic principle is to calculate a baseline for a campaign (using other data in some way), compare against actual performance with the difference attributed to the impact of the campaign. It then gets complicated with multiple campaigns & different quality campaigns. But I was sold on this being a valid approach to accurately calculating the impact of campaigns.
Unfortunately we then switched to a presentation on Attribution which highlighted the many issues with it but went through how Sky attempts to use it anyway. No issues with the presenter but she didn’t appear to believe in Attribution either.
I really really wish this had been a full session on Econometrics.
Customer Analytics Part 2
The final breakout session I attended was another on customer analytics featuring my old boss Matthew Tod. I know he likes to be controversial and make people question what they are doing so anticipated an interesting session. His co-presenter was Adam Jenkins, an Evangelist for Adobe Analytics.
Adam started out by stating 3 facts (apologies if I don’t get this word perfect):
- Customer Analytics is the future of Digital Analytics
- Conversion rate doesn’t matter (an old favourite of Matthews)
- Digital data is only a part of customer data
Unfortunately all 3 “facts” are merely opinions. I disagree with all three which made this a very difficult presentation to sit through. I have had plans for a while now to write a blog post about how I believe there are three types of analytics for online companies:
- Marketing Analytics (pre website behaviour)
- Web Analytics (website behaviour)
- Customer/Member Analytics (post website behaviour)
For companies where the majority of visitors log in, e.g. community websites or those with most useful content behind a login, I agree that Customer/Member analytics is the most critical. For the vast majority of companies, it is only one element. Any analyst who focuses all of their attention on say 5% of website visitors (most retail websites) is leaving a lot of money on the table.
Matthew then stuck to his guns on how Conversion Rate is totally irrelevant. I continue to disagree with him, while the website conversion rate declining doesn’t necessarily mean bad performance, start applying segments and it is the key metric for most organisations.
So I was a bit frustrated with the last couple of sessions (and it only got better when I was dragged up on stage for everyone’s general amusement a little while later). There was mixed feedback from other sessions that people attended, some of the technical labs ones were said to be great as the presenter really got into the detail but others were considered quite basic. I believe the Marketing Innovations track was well received as it was more about concepts.
My request, and that of every other analyst I talked to with more than a couple of years experience, is that a new track is required for (real) Advanced Analytics. We don’t care what company the presenter works for, we want their knowledge & ideas – mostly we just want to learn something new. Please…