Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Marcus Tewksbury, global vice president for product strategy at Experian Marketing Services.
How fast is fast?
Racing to a friend’s wedding on a farmland highway west of Chicago, I found out that fast is 97 miles per hour, thanks to the highway patrol. We’ve figured out how to measure this type of speed, but defining what constitutes "fast" remains contextual. School zones, major interstates and the Indianapolis 500 each have an established standard that needs to be taken in context with the raw miles per hour.
Within the marketing industry, we also have the tools to measure raw speed. Email sends per hour? Check. Dynamic page loads in a minute? Got it. Retargeted impressions in a second? We measure it.
Determining what qualifies as fast requires context. Evaluating the speed of your data and delivery will vary depending on the channel and customer expectations. Before trying to make your data faster, you need to define fast and set benchmarks for each delivery.
Real-Time Digital Personalization: Marketing’s Indy 500
Digital personalization — whether it’s on a brand’s website or within a display ad — is where rubber hits the road in marketing programs. When a website visitor lands on a page, decisions as to how to make a good first impression need to be made in the time it takes to load the page. Any slowdown increases page load time and the likelihood of frustrating or losing visitor interest.
While there are a growing number of program types in the ad tech space of DMPs, RTB and ad exchanged inventory, the most profitable application of real-time personalization is for website product recommendations. When a visitor visits a website, do you want to show everyone the same set of products? Randomly sample across the general population? Or do you want to make an informed decision based upon that visitor’s past purchase history? It’s an easy answer, putting digital personalization at the top of many marketers’ to-do list.
Getting to the right speed to enable real-time digital personalization requires a couple of components. First, you need a decision engine that can do all its computations and return a decision virtually instantaneously at scale. Second, you need a source that can feed data at the same rate. These come in two general flavors:
- Anonymous visitors: Data operating at the demographic level, such as gender or location, can be delivered through a call to a data partner’s private DMP setup on the site’s behalf.
- Known visitors: data that taps into deeper insight, such as purchase history, recency and frequency of past transactions. Most modern channel marketing platforms can expose the data via API to real-time querying. Older systems must make export lists and pre-stage them somewhere accessible to the decision engine.
Speed check: Does every page view draw upon up-to-the-second transactional history to tailor product and content shown to the specific visitor?
Operational Messaging: Marketing’s Highway
Order confirmations and receipts aren’t quite as time sensitive as digital personalization, but consumer expectation is still high for speed of delivery. When an order gets submitted via a POS or ecommerce platform, the details of that order, along with available content specific to the purchaser, need to be merged and used to create a personalized message. In this way, personalized operational messages have a lot in common with site product recommendations. In turn the operational messages, usually via email, need to be pushed out and delivered.
To execute on this, the POS or ecommerce platform needs to deliver the transaction detail a record at a time to the cross-channel platform. It’s here that detail is merged with customer history to dynamically create the operational message.
Speed check: Do your tailored email or SMS images deploy within a minute of the order being completed?
Attribution: Marketing’s School Zone
Not much today is truly “school zone” territory. Anyone with more than five years of marketing experience may remember slower times, particularly direct mail or catalog mailings. Today, marketers want attribution to understand how marketing spending across all channels impacts transaction rates, delivered at the speed to make corrections in flight. Most currently use it for postmortem reporting and justification.
Attribution can measure the impact of digital display on in-store sales. Procter & Gamble and other major brands plan to double down on addressable digital channels in the coming year. This type of attribution works opposite digital personalization. Publishers produce a campaign history of exposed first-party visitors. This campaign history is then decoded through the same targeting sources and related back to purchase history.
Lining up these purchases with candidate marketing touches is the purview of specialized response attribution packages. With levers for things like in-home dates, channel preference and response lag, these tools enable marketers to get a fairly precise view of the contribution of specific content (great for deepening A/B testing) and campaigns.
Speed check: Is your attribution reporting generated on demand, including transactions completed through yesterday’s close of business?
Speed Training: How To Make Your Data Faster
There is no single, catch-all trick to improve the universal speed of your data. Different channels and tactics require different approaches. Start speed training your data with three steps:
1. Start measuring: You can only get better at something you measure. Come to grips with your current reality. Evaluate how long it takes to deliver targeted communication or measure performance. Add these as new dimensions to your marketing dashboard.
2. Use transactional history for dynamic personalization: Transactional history is a powerful source for relevancy. Get it off the ground in any channel. Start with basic category assortment for some broad recommendations then move into basket detail for replenishment or cross-sell opportunities.
3. Test cross-channel execution: In our dynamic, cross-channel world our customers bounce across channels. Fully understanding the effectiveness of a specific message must account for downstream response in virtually any channel. Start by using email or loyalty membership to connect in-store sales.