Enter a search term such as “mobile analytics” or browse our content using the filters above.
Check your spelling or try broadening your search.
Sorry about this, there is a problem with our search at the moment.
Please try again later.
I had the pleasure (or dubious honour?) of moderating on Econsultancy’s attribution analysis and modelling table at this year’s Digital Cream event at the Emirates stadium in London.
It was an intriguing insight in to what ecommerce teams are doing and what is holding them back. The common theme was a slight unease about how best to use attribution modelling to help the business grow.
There was variety in the type of company represented, from membership organisations to high street retailers, as well as the job roles of the people attending, from ecommerce managers to business analysts.
Here’s a summary of the six key issues that were discussed and the challenges that businesses are facing.
Please note that under Chatham House rules, it's not possible for me to disclose any company or delegate specific information as that would compromise the confidentiality that is crucial to Digital Cream.
If you're looking for case study material, this blog doesn't provide it.
1. Not knowing where to start
Inertia is holding people back. There is a perceived complexity with attribution due to the endless potential of what you can model, which leads to projects stalling as quicker wins are prioritised.
The best advice is to focus on a key question that, if answered, would help the business make better decisions and improve KPIs. Get the analytics tools set-up to capture and model the data, and then learn from the process. Once you have seen results, you can expand the modelling and increase complexity gradually.
For example, a common challenge with paid search is knowing the optimal blend of keyword bidding. You could focus your attribution modelling on increasing knowledge in this area and helping you determine bidding strategies and rules, including day-parting to manage cost and click through.
2. Disparate systems and lack of trust in data
This is an age-old analytics headache. For many reasons, companies can grow up with multiple analytics tools e.g. different managers prefer different tools and there is no formal migration process to maintain a central data store.
Inevitably, because each tool has its own unique way of capturing and displaying data, never shall the data sets align perfectly. Discrepancies in numbers are causing internal division and a lack of faith in data.
The concern from senior management is that the data is unreliable; therefore it’s not sensible to use to make investment decisions.
Good advice is to not let the risk of failure hold you back. Attribution is an imperfect science, so you have to start by accepting a margin of error.
However, you can learn quickly by modelling data and seeing the impact it has on your KPIs. This can influence business decisions.
Pick your data set and validate the data to make sure it’s being captured accurately. Then use this data set for attribution, don’t worry that the numbers may be slightly different somewhere else.
Focus on trend analysis, not absolutes.
3. Lack of in-house specialist skills
There is often an imbalance between commercial decision makers and the analysts who can implement and manage attribution programs. This leads to a desire to create a project but nobody to deliver it, which is why some outsource to external specialists to avoid having to recruit internally.
A good example given was the inability of the business to translate commercial requirements in to a technical specification for tagging to ensure that the required data is being captured.
This requires an interesting blend of commercial and technical skills and the ability to validate code to check that the tags are working correctly.
4. Investment conflict
Interestingly, in some businesses the responsibility for attribution sits within the ecommerce team yet the budget for investment in tools sits within IT. This can create a conflict of interests and battle for prioritisation.
This is an extreme example but does highlight potential internal barriers to adoption.
5. Lack of resource
A common issue is not having the required staff to make use of an attribution model. Attribution modelling can churn out a lot of data, so if there is nobody there to analyse, interpret and act, the data is wasted.
It’s interesting that many businesses see the business case based on whether or not the data can be used by their existing team, not on the benefits that attribution can bring vs. the costs of enabling.
Perhaps there needs to be a mind-set change to see attribution as a core investment, rather than a cost, and to create the business case based on a more long-term ROI.
6. Complexity of multichannel/multi-device tie-up
Where do you start with that topic? This got many heads scratching. Some believe you can link devices; others are sceptical.
Again, this comes down to small steps. If you try and nail the whole picture upfront, you risk drowning in a sea of data complexity. However, there are some quick wins that can be tested.
For example, some multichannel retailers offering in-store WiFi (e.g. supermarkets) are encouraging customers to sign-up for free WiFi. They register their account to get a unique access code.
This enables the business to link the device they use in-store with their main account. It’s not a perfect science (e.g. shared devices like iPads) but it’s one piece in the jigsaw.
In most sessions when we came on to any phrase including ‘multi’, a look of sheer terror flashed across the eyes…
Summary: the role of attribution is not yet clear
It was a fascinating insight into the barriers that are holding businesses back from fully embracing attribution. Whilst some teams have made the first steps and are proactively using attribution analysis as part of their optimisation program, many are tentatively dipping their toes in the water hoping for someone to give them the answer.
Only two of the 25 delegates had a functioning attribution model in place, the rest were focused more on analysis.
If you’re looking for advice and guidance on setting up attribution modelling for your business, I can highly recommend contacting Lewis Lenssen at DC Storm. He provided a wealth of advice and information to the delegates and has experience of working across different markets to help businesses get to grips with attribution. I learned a lot just listening to him.
In my next blog I’ll take a look at some of the business questions that attribution analysis can help answer to provide more context.