Future State - The Oracle Consulting Blog
-
Oracle Corporation
March 7, 2018
Handy DMP life-hacks you won't abandon - unlike your resolutions!
Author: Stephen Hanvey, DMP Expert Services Consultant at Oracle Consulting
As the old joke goes "I can't believe it's nearly a year since I didn't become a better person". If your 2018 resolutions were to try and keep 2017's (or 2016's for that matter) then you're in good company as 42% of people who make resolutions fail on them each year and a paltry 9% would say they were successful in reaching theirs.
There are good reasons why it is so difficult to keep resolutions: from unrealistic expectations, not being ready for change, having too many targets or being too easily dissuaded by initial failures - if I may jump to a conclusion you could say that it's all down to a lack of strategy!
I suppose all this makes for pretty negative reading and it would be enough to put anyone off bothering, yet contrary to the high failure rate so many believe that the turn of the year is the perfect catalyst for a self improvement make-over. In seeking sage advice from the plethora of life coaching blogs, I've been introduced to the basics of making achievable resolutions that will stick, which if I bind into one sentence for your benefit would be - Don't make resolutions - have a strategy like setting SMART daily goals.
So what's that got to do with DMPs you would be wise to ask - well to me, managing a DMP can sometimes feel a little bit like that post-binge seasonal life-hack during the down time between 26 and 31 December followed by the subsequent mid February resignation, let me illustrate:
- I need to lose weight - This year we need to reduce our operating costs, improve efficiency (whether that's ROAS or ROI) and reduce our CPA
- I need to learn something new - We need to improve our knowledge of customers and make better data-based decisions from our owned data assets
- I need get fit and go to the gym - We need to work smarter, be prepared and be more efficient
So here are a few data management life-hacks as counsel and opinion from my own DMP consulting experience. Maybe one of these will be a SMART goal you can accomplish in 2018?
Start with data but move to comprehension
When it comes to DMPs and data, I am always reminded of a quote by H. Jackson Brown Jr when he said "Nothing is more expensive than a missed opportunity". To invest in a DMP, but not plumb in ALL your valuable customer data and then worse still not make it your goal to move from data integration to data comprehension is a huge missed opportunity.
A DMP cannot create miracles by turning poor customer data into rich and valuable insights or make an inefficient marketing strategy efficient overnight, in fact without a data strategy the DMP will simply turn into a very expensive audience building tool that sits at the front of your programmatic advertising process.
Moving to comprehension requires such a data strategy, and one under-pinned by a DMP, the ingredients of which are:
- Identify, Coordinate and Consolidate - The DMP allows a business to create a single source of truth, as a consequence it can bring about the demolition of data silos that exist due to disparate data production, collection and storage systems. BUT this is true only if the DMP is allowed to become that source of truth by strategically planning ingestion and consolidation around one privacy safe ID. Ingestion of what... well of offline customer records, online tag collected web and mobile web activity, mobile app interaction data, web analytics insights, data science derived propensity modelling, tracked media interactions (impressions and clicks), 2nd party customer data sources (such as POS data from partner retailers if you're a CPG brand) and any other non-PII customer data source that a brand/publisher has at its disposal. In addition to the need for the plan, what needs to co-exist is an executive level understanding, buy in and advocacy of a centralised data strategy. In short - to not have a fully bought into plan for the identification, collection and consolidation of all existing and future customer data streams into the DMP is to be setting yourself up for an "expensive... missed opportunity"
- Make data segmentation and analytics objective driven - Consider the challenges that most businesses are trying to solve through a data strategy, an example can be seen in this IDG study here. The top 3 are:
- 60% – Finding correlations across multiple disparate data source
- 47% – Predicting customer behaviour
- 42% – Predicting product or service sales
Stating the obvious, after all this section is called "move to comprehension" - the next step after data consolidation is to make sense of it all. The best way to approach this is to have a clear business objective in mind that will fuel your analytical approach.
Let's take for example the most popular challenge from that IDG study - Finding correlations across multiple disparate data sources. What is this really all about? Why would an organisation want to find correlations across data sources?
The goal of assessing your customer data in this instance is to arrive at a measure of three things:
- Propensity in terms of the natural inclination to behave a certain way. If I can analyse customer data to arrive at a measure of 'propensity to convert' then I can dispense with wasteful marketing to low propensity customers.
- Affinity in terms of liking something based on shared characteristics. If I can analyse customer data to arrive a measure of "affinity for January sales and low cost items" then I can be highly targeted to these individuals to improve CTRs. Not to mention incredibly valuable audience segments created from building detailed lookalike models.
- Congruence in terms of the similarity between objects. If I can measure data congruency then I can discover which customers 'act like' my highest spending customers, and across which channels they engage, to improve accuracy and timeliness of any marketing. There could also be valuable learnings about a customer's cross-brand activity and cross-device journey that will inform an improved sales and marketing approach.
Develop an Audience Segmentation Strategy, make it full funnel
If media efficiency and ROI are your businesses detox missions in 2018 then having an audience segmentation strategy is paramount. Good segmentation requires a thorough understanding of the following:
- The sources of data (see above)
- The method and process of data collection. What is the onboarding time - is it real time or ramped over 24-48 hours? What data in the tag is natively captured vs added per campaign and can it all be easily used to build audiences?
- The DMP rules on data storage and cookie lifetime. What is the TTL (time to live) of the DMP cookie: 30 days or longer? Does a cookie remain targetable throughout that TTL and on what basis: cookies may be deleted if not seen after 30 days? What are the average decay rates of these cookies: i.e. a half life of 18 days? Are there rules around how long a taxonomy category can be targeted if it's based on a 3rd party cookie with a limited TTL?
- The process of creating audience segments in the DMP such as: boolean logic, mutually exclusive events and exclusions, the algorithmic calculation of propensity, affinity and congruency to understand act alike and look alike models. Also whether there is in-built A/B testing to enable clean separation of the cookie pool for testing.
- How and where the segment will be activated and how the DMP supports that integration/audience transfer. An audience segment for a social resell campaign will behave very differently to a search retargeting audience and will likely be transferred to the execution platform in a different way - e.g. one may require sending all cookie profiles that qualify and another may need customers to qualify for a campaign by their actions on the brand website before being cookie synced with the execution platform.
This however is not a strategy, having a firm grasp on the above means that anyone creating audience segments will be set up for success.
The strategy comes into play when you consider an audience segmentation strategy should be based on campaign goals, success metrics & target audience. A DMP user should first create preliminary definitions of their target audiences, build them in the DMP and then, before activating, run pre-campaign analytics. So what are we looking for in these pre-campaign reports? We should aim to understand demographics, preferred engagement channels, device engagement data and 3rd party data segments that enhance our knowledge of the customers intent. These insights allow you to tailor the creative message and ensure that the balance of scale and accuracy of the audience is suitable.
Once a campaign is active, mid campaign analytics should allow discovery of ideal customer attributes. What attributes correlate most highly with the converting profiles and which 3rd party data assets are contributing most positively to those conversions. Use this detail to optimise existing audience segments and introduce exclusions for converters into the original audience to reduce wasted ad-spend.
Now consider the full conversion/purchase funnel in your segmentation strategy. Create a second audience of only converters and analyse that audience to identify what the shared data attributes are (inc. propensity, affinity and congruency). Use these attributes to improve prospecting by building tiered targeted look-alike and act-alike audiences (targeting based on more and less accuracy adjusting bid strategy accordingly). Create retargeting and cross-sell campaigns for those converter segments with exclusions for prospecting segments to ensure consistency of message and reduce wasted ad-spend.
When I look to start building an audience segmentation strategy, I don't just look at reactive campaign needs, as this can become the trap of audience building in the DMP, literally building audiences to an agency media campaign brief. Instead consider that an audience can be predictive, descriptive, reactive and/or passive. We can build audiences with zero reach that will increase over time as customers qualify.
This means it is possible to concentrate on having granularity in your full funnel audience segmentations. After all it is through this granularity that we enable better performing audiences, with controlled cost and reach for ad buying that have the desired detoxifying impact on ROAS and ROI.
In the next article I will offer my opinion on something that is crucial to the DMPs value in the marketing and advertising industry - Organisational Readiness - It's not just about the right people, it's about the right team, the right decision making, the appropriate education and a relentless focus on test and learn reiteration.