In late 2012, Econsultancy published the latest edition of its Marketing Attribution Management Buyer’s Guide, at a time when attribution was a particularly hot topic for marketers.
Vendors were furiously marketing their attribution platforms, and there were blog posts galore on the subject. Since then, talking about attribution, particularly in the same breath as the dreaded term ‘big data’, appears to have gone somewhat off the boil.
Or so I thought, before attending a recent Econsultancy roundtable on the subject of marketing attribution, where discussion and debate was as lively as I have seen at a roundtable.
Attending were a mix of marketers at varying stages in the journey towards implementing an attribution solution, alongside a number of representatives from vendors, analytics providers and agencies.
All spoke with enthusiasm of their journeys and resulting opinions on the use of attribution tools to analyse the customer journey to purchase.
So, attribution; what is it for? Some potential answers:
- Optimising the media mix.
- Justification of spend.
- Remuneration of partners e.g. affiliates.
- Optimising the impact of content.
- Offline/online optimisation.
- Winning back lapsed customers.
- Forecasting/planning budget.
Whatever your reason, success with attribution for a marketer is likely to come down to three key factors:
- Internal skill.
- Internal politics.
- Data quality.
Emerging from the two-hour discussion were best practice tips and advice for those juggling with the above factors to make a success of attribution. Five of the best are below:
1. Don’t try to do it all by yourself
There is a shedload of great technology out there that can help to make sense of your data. Even the biggest brands can lack the expertise to combine datasets and carry out complex algorithmic attribution in-house.
Even if the majority of companies cannot afford to bring the expertise in-house (as one attendee’s brand had done, by simply buying out an attribution company) there are now some strong providers of attribution and analytics services on the market that can go a long way towards making sense of your data.
2. Make sure that you trust the data before making decisions
One salient point made was that though the quantity of data is increasing, the quality of data is getting worse due to the number of devices and channels that marketers are trying to track across.
One delegate was struggling with the start of their attribution journey because the results coming out were not matching up to what they were seeing on the back end.
There were murmurs of agreement when it was suggested that there was a lack of belief that the tools were accurately capturing the multichannel data.
Advocated was the process of using tools like Adobe Insights to bring together data sets and ensuring that they are clean and accurate, before embarking upon attribution.
3. Buy-in at every level is integral
Senior-level buy-in for any process or budget change is always essential, as they are the decision-makers. However, insights from attribution tools have an impact across the business, to even the smallest team that is involved in the consumer path to purchase.
Before attribution, it may have appeared like a certain team was doing a great job at achieving ROI and having a big impact on sales. They don’t want this questioned by an attribution model that suddenly gives more weight to a totally different channel.
Impacts are far reaching; budgets can be cut, rewards such as commission and bonuses reduced, and work loads change. As a result, teams can be reticent to get on board with attribution, making the collection of data potentially difficult, and the support of teams more difficult to gain.
One delegate explained that unless they could get a model that everyone agreed on, they couldn’t get buy-in to implement changes, and therefore it was a useless model. Testing, and proving the results by comparing to a forecast by the attribution tool, were considered important for achieving the necessary buy-in.
4. Spend time choosing the right tool and model
Depending on your business model, attribution can be based on conversion points other than sales, for example downloads or sign-ups. It was suggested that businesses need to determine their objectives from attribution, and ensure the model being implemented can produce insightful results.
One delegate at the roundtable had found it difficult to get buy-in for a model that only attributed on the last part of the journey to a sale, so had resorted to using attribution as a research tool to affect strategy rather than budgets.
Broken pathways to purchase as consumers jump across devices creates issues for companies that can’t track across device, which, let’s face it, is the vast majority. Working out if the attribution model’s results are a true reflection of these unknown complete paths is a hard task.
One attendee suggested finding a tool that has a forecasting function, and comparing the forecast to the end result of various models. If the model produces results within 5-10% of the forecast, a good level of confidence can be had in the model.
5. Test, test, test
Finally, testing is key to getting the buy-in necessary to make changes. One delegate told of how his company had tested lowering paid search budgets based on an attribution model that undervalued PPC; a huge drop in sales resulted.
Many are trying to move away from last-click attribution models which traditionally result in high PPC budgets, by erroneously lowering those budgets, resulting in a drop in sales and a subsequent drop in confidence about attribution.
Testing needs to occur before confidence can be put into any attribution model, which is where many companies can stumble with the implementation process. As mentioned, testing against a forecasting tool can instil confidence that the correct balance is being achieved, and another example was control group testing, including the charity test.
This can be used to prove display revenue, involving apportioning a percentage of budget towards advertising for a charity of choice in the same inventory as would normally be used, and tracking it as standard display ads.
Scepticism about attribution still abounds. The question was asked of how successful attribution is compared to simply testing different radical marketing mixes. This is particularly prevalent considering the lack of confidence many have in the quality of their data, and when the issue of how to match up multi-device and online/offline data is still very present.
The answer is unclear. However, attribution technologies are improving all the time, and as with any relatively new technology, will take a while for education and confidence in the process to be strong among marketers.