If you're like most companies your first party data is your most valuable owned marketing resource, and likely your most underutilised, says iCrossing's chief marketing officer Alistair Dent.
But at an individual level there are limits to the usefulness of that data. Trying to analyse the factors on each individual path to purchase is tricky. Serving a personalised ad specific to just one user is impossible outside of your owned channels (website, email, SMS etc). Most media owners won't even let you target a group smaller than a few hundred people at a time.
But segmentation of your users exposes much of the value in that data. Start with retargeting.
Returning visitors to your website will likely show a different predicted conversion rate, so put them into a list so you can bid differently. Analyse the incrementality so that you can bid less for returning visitors if necessary. Put users who make a purchase in your list so that you can identify those who have shown they're willing to engage with your brand.
Most marketers follow the above strategies, but many stop there. Follow the next step: segment users based on categories that are relevant to your own marketing strategy. If a user bought an expensive bed, they're more likely than average to be interested in expensive bedding. If they reached your site via a search for "cheap flights to Rome" then you can forever tag that user as being price sensitive.
The third party data market can cover a lot of useful segments, but with some key downsides:
• There is little transparency about how the data is collected, so you can't judge accuracy.
• Very few understand how people move across devices, so you lose key behavioural indicators.
• Cookie expiration dates mean that old data can become useless quickly.
• You can't use third party data in targeting on search.
By looking at your own customers you know that you're seeing accurate data that is hugely relevant to your brand. You can use this to retarget, or more powerfully you can use this to target look-a-likes.
Lookalike modelling is available in many platforms, and can extend the wonderful customer segmentation you've done on your website visitors to find other users around the web whose behaviour is similar. If you sell posh green wellington boots and your best customers spend a lot of time reading both "Country Living" and "Good Housekeeping", then you can target only users that regularly read both sites. At an extended scale this lets you expand your targeting into prospecting, but based on the segments that matter to your business.
Lookalike modelling (and buying media on partners that index closely to your audiences) based on more sophisticated segments of your website visitors is reasonably rare, but the next stage is more rare and more valuable: defining those segments based on your CRM data.
Marketers often think of CRM as an email channel. A list of people who have engaged directly with your brand in the past. Some savvy brands start to analyse that data and tag the "best next action" for each user based on their history. Has an energy contract but not a service contract? Great, that's a tag that a service contract is the next upsell. That insight tends to stay inside the database, or maybe targeted email lists.
By exposing that best next action to the website when an existing customer logs in, they can be placed into an appropriate retargeting list. A DMP, a web analytics package or even a tag manager can all fire the right tag at the right time (or populate the right variable into an existing tag) based on a variable on the page put there by the CRM database. Imagine being able to do lookalike modelling on the existing customers who fit into a "frequent purchaser" segment, or a "high value purchase" segment, or even a "pays promptly" segment…
We're getting into some quite unusual but very valuable turf now, but we can go further. So far all of our data has been implicit: we've inferred it from a user's purchases, searches or website browsing. But it's well within our scope to simply ask them. This might be heavy-touch (e.g. a long incentivised survey) or light-touch (like allowing them to rearrange their homepage to remove content they don't like). We might offer them a way to improve their experience as a customer, or we might slide in a single-question choice once every few visits. Either way that data can be used to enrich the segment they're in.
This explicit data is more valuable than anything we can learn implicitly, because it comes straight from the user's own preferences about how we should talk to them. Expect data enrichment to be a core pillar of a wide range of digital agencies' services in 2017 as it breaks out from email and into media targeting.
This article was written by Alistair Dent, chief media officer at iCrossing.