In an increasingly complex programmatic world, where we’re dealing with viewability, cookie deletion, cross-device attribution and third-party data enriched demand-side platforms (DSP), try raising the need for a data management platform (DMP) to a marketer. Either they’ll just shrug at the need to add yet another piece of ad tech, or they’ll tell you it’s the last thing on their shopping list in terms of priority.
And I don’t blame them. I read somewhere that for every 1 dollar a marketer wants to spend, they can expect to pay around 25-35 cents of it to deliver the message. That cost includes creative services, DSP charges, attribution, ad verification, and reporting. Add dynamic creative and a DMP into the mix and the cost goes even higher – and I haven’t even mentioned the pain of managing separate contracts for all these elements.
So while the industry is nowhere near consolidating around that problem, you can drive significant campaign efficiencies right now by using your own data for re-targeting, and audience extension (look alike) powered prospecting. This can deliver incremental ROI gains that pay up dividends for the entire stack.
As a marketer, you run campaigns to create buzz using attitudinal data. After creating that buzz, you communicate product features based off behavioral data, follow it up with re-targeting to convert – and then for the converts, execute cross-sell and up-sell campaigns. At each stage, you can use your own data to target, re-target, and extend audiences.
You are sitting on a gold mine of first-party data or second-party data (someone else’s first party data). You can also buy data to fill in the gaps.
A segment strategy requires you to understand the source of the data (online signals or offline actions), the method of collection (first party, second party or third party), the process of creating a segment (Boolean rules based or algorithmic look alike), and what the resulting segment attribute is (descriptive or predictive).
Some Segmentation Strategies influencing campaigns include:
Let’s say someone who was targeted based on attitudinal data from first-party or third-party data in a DSP. This person has seen a certain creative, so then the marketer can apply exclusions based on exposures to, or engagements with that specific creative. This is simple to do and paves the way for including more signals into that exclusion beyond just post-impression campaign data, to post-click data. This is where you will see the most impactful gains in your campaigns, and finally stop wasting money on advertising to the same person. This also means you can run control /test campaigns based off attitudinal data using a research company to validate your digital campaigns and their potential buzz/lift.
Customer Relationship Management (CRM) Based Targeting
You have the ‘what’ signals (what are users doing on your digital properties) and the ‘who’ signals (CRM data containing age, gender, Next Best Offer (NBO), churn potential, loyalty status, product owned) that you can now combine to create a segment to activate across programmatic display. You can even add a layer of geographical targeting, or in the mobile space, consider geo-fencing data, to make your targeting more specific to the user’s location.
When building a segment, think about applying recency and frequency based filters. For example, search a segment of ‘men looking at a creative for Product A, exposed to a banner 3 times in the last 7 days’. The advantage here is while DSPs offer frequency capping across a campaign, you can apply frequency capping across multiple DSPs, and even include channels beyond paid channels (paid display, on-site personalisation and email).
Once you have identified a combined segment comprising behavioural (clickstream, products bought) and descriptive (lifestyle, education level, NBO, Net Promoter Score (NPS)) data, you will see segments qualifying for NBO available for you to target at the right stage. An Analytics team that uses product ownership and market basket analysis can help find associations between product ownership with high confidence and lift. This segment can now be targeted by executing a campaign via the DMP on your owned channels (site /app/email) or a subtle targeting off site. This also means that you can now provide quants applying advanced statistical techniques on all offline data, and gain a faster way to go to market in targeting those segments. This also gives the team a way to learn quickly from the success of their targeting strategies and make changes to the recommendations coming in.
In simple words, the traditional market basket analysis now has a faster go-to-market on digital channels with a faster learning cycle. What’s more, the ability to add more data points beyond just product ownership when executing on models delivering NBO really boosts a marketer’s power.
This is pretty straightforward, especially for Travel and Hospitality, Retail or FMCG industries. You might even have experienced this yourself as a consumer – when searching for a sector, a destination or a hotel property, you’ve started to notice specific ads following you around the Internet.
But rather than doing a simple re-targeting based off the current search, you can apply a layer of signals about that person to change your re-targeting strategy to one with segment based re-targeting – where offers might change by loyalty status, for instance. Rather than a simple ‘because you searched for this sector, here is a message reminding you to book it’ scenario, it’s now adding a layer of past purchase history to your data.
Web Behavioural Targeting
This segment is built by combining third-party data that’s either tied to behavioural or predictive signals, or signals from your own clickstream. The resulting segment could either be descriptive or predictive, depending on campaign strategies that you want to use to target individuals – based on what they have done in the past or their intention, and hence predict their future behaviour.
Let’s take a segment called ’In Market – Automobile enthusiast’ for example. This is a segment comprising of individuals who are interested in buying a car. How do we know that? Because these individuals have looked at car reviews in the last 30 days. This shows recency and relevance: we can be fairly confident they are in the market to buy a new car.
Dynamic creative warrants a whole post to itself. It provides some simple offerings that allow users to upload a file containing parameters you want to change (Fare, Sector, Location, Price) and create multiple creatives at a low cost (the main advantage here) to others which allows for real-time decisioning to render a creative based off data coming in, and a segment qualifying for some pre-determined logic. Stay tune for more information on dynamic creative and DMPs in a follow up posts.
As the name suggests, the idea here is to target a segment by using the ad impression data as an inbound data source, aiming to change the targeting message after a certain amount of ad exposures. This helps with campaign efficiencies by avoiding campaign fatigue and taking the prospect through a sequence of messages which could be different by exposures and – more importantly – by device. Plus, it can be changed mid campaign.
A DMP powered campaign strategy can bring in efficiencies as programmatic ad spend increases, and can help marketers focus on driving loyalty and customer lifetime value. 2016 is the year the Marketing Stack and the Ad Tech Stack come together, and the DMP sits right in the middle of your marketing cloud, making that possible.