OVERVIEW

Remarketing can be a powerful, immensely effective display targeting option; a real tool to support the performance of other marketing channels, as well as an easy way to independently drive new sales/leads through your site at a profitable cost.

The most basic premise of remarketing is justification in itself to run remarketing campaigns – following users around the internet with the products they’ve recently viewed, taking them back to the same product page – but if you dive a little deeper and discover the true potential of remarketing – some example applications and targeting options (how far it can be taken) – its value becomes truly apparent.

This guide is a reduced run-through of just that; an overview of the key principles that make amazing remarketing campaigns so amazing, reasoning behind the approaches and some examples for each.

SEGMENTATION

Segmentation, segmentation, segmentation (within reason). The smallest amount of people you can target via remarketing on the GDN (Google Display Network) is 100 in a single audience; with this in mind you want to segment your audiences as much as possible, by as many possible variables which denote user quality – potential to convert – as you can. This enables you to define your bid for each group; increasing ad frequency for say, users in your highest quality group, resulting in more of them coming back to your site and reduced frequency and spend on your lowest quality group - keeping your spend in-line with your return. Here are six key dimensions you’d want to consider segmenting by – if possible (100 users, remember?) – but it can be taken further, and has been, for better performance:

TIME SINCE LAST VISIT

Not the most important one, but still a very important one, is the time (in days) since the users last visited the website. Grouping your targeted audiences by this is a no-brainer; having segments for users for say the last seven days, last eight to fourteen, last fifteen to thirty, and so forth and so on is straight forward and easy to implement, and rewarding in performance. If the traffic volume on-site is there and there (100 users in each, including your other segments – will explain further on) I’d recommend having your remarketing audiences segmented by individual days, having segments for each day. But this isn’t realistic for 99% of the advertisers out there, so here are a few example segments that would work for most websites:

  • Last 7 Days
  • Last 8 to 14 Days
  • Last 15 to 30 Days
  • Last 31 to 60 Days

The above example only covers a more immediate focused remarketing campaign – getting users back to site to convert – you may find other, more appropriate uses for the time since last visit segment if your remarketing objective differs; sell domestic holidays and want to increase awareness amongst people who booked this month last year? Twelve month time-since-last-visit. This could be taken further with segments for percentiles of spend – group up the users so you can have much more frequency of the users who spent the most last year, the top 10% of spenders, and lower frequency with people who only booked short breaks.

PRODUCTS/PAGES VIEWED

This is the most important one, segmenting your audiences by products viewed. There’s some trickery you can do to dynamically serve correct adverts with the right content for the pages/products the users has viewed – which would null this step/point – but for the sake of explanation we’ll presume you have two products on-site; one for a blue hat and one for a red hat, you want to show adverts with blue hats in to the users who viewed the blue hat on-site who didn’t purchase it, and likewise for the red hat – users who viewed both would be in both audiences (but you could set up extra layers of targeting so users who spent longer on one of the hat pages would only be placed into that audience)

So in this instance you would have segments for:

  • Red Hat Viewers
  • Blue Hat Viewers

PAGE DEPTH

Now you’ve got your base, top level segments in place (time since last visit and the product pages visited) you’ll want to start looking at how to group up by user quality, and what dimensions denote user quality. Page depth (or PD) sits more in the user quality group than any other, and it’s simply how many pages a user visited when they were on-site (not any pages specifically – by itself – but how many).

But as a rule of thumb, users who had a page depth of one (only visited one page before leaving) will have a much worse performance than users who had a page depth of say five, or six.

So your segments may be:

  • PD = 1 to 2
  • PD = 2 to 3
  • PD = 4 to 6
  • PD = 7 to 9
  • PD = 10+

Things to really consider when using PD as a remarketing segment; every website is different, has a different structure and users will carve out very regular, most common, conversion journeys through them, resulting in very different page depth X conversion rate signatures for each one. So before you create your PD segments, take a look in Google Analytics.

KEY PAGES VIEWED

Along with the others, you’ll want to weigh up the value of throwing what specific pages a user has visited in with the mix – for eCommerce stores this might be the delivery information page, for service based business this might be the reviews page, against users who didn’t. Users who visited these pages will have a much higher intent to purchase/enquire than ones who didn’t.

  • Key Page 1: Delivery Information (/stuff/delivery-info)
  • Key Page 2: Company About Us (stuff/about-us)

TRAFFIC/SOURCE CHANNELS

Another segment which more often than not denotes user traffic quality is the user’s traffic source. Segmenting by these can help you focus your remarketing spend/campaigns on users which will perform better. A few examples:

  • Google / Organic
  • Google / CPC
  • Bing / Organic
  • Yahoo / Organic
  • awebsiteco.uk / Referral

Worth noting; anything you can targeting – include – with your segments, can also be excluded from your segments, to prevent those users from falling into your audiences irrespective if they’ve fulfilled the criteria or not. So you may find there are some referral sites that drive low quality users to your site, or users who only visited one page and left after a few seconds, damage the performance of your remarketing campaigns too much. If so, you can just exclude these users.

DEVICE

I won’t go into too much detail (as it’ll require explaining the whole mobile PPC strategy/eco system thing as mobile remarketing plays a small part in a much, much bigger picture – I’ll save that for another time) but you might want to consider your device targeting, if at all segmented, when setting up your campaign. But presuming we’re not targeting just desktop devices, we’ll go for:

  • Desktop Users
  • Mobile Users

RESULTING STRUCTURE

Now you’ve got your segments, you also have your campaign structure – you just need to make sure you’ve planned far enough ahead that you know you’ll have 100 users in each targeted group (I did say we’d come back to it).

So you need to take all of your segments and undertake some multiplication for your campaigns. Lets presume we’re only segmenting by traffic source, time since last visit, product viewed and page depth. Our resulting audiences (how we’re grouping the users we’re actually going to be targeting) would look like:

  • Google/CPC | 7 Days | Blue Hat | PD = 1 to 3
  • Google/CPC | 7 Days | Blue Hat | PD = 4 to 6
  • Google/CPC | 7 Days | Blue Hat | PD = 7 to 10
  • Google/CPC | 7 Days | Blue Hat | PD = 10+

These four are just an excerpt of the groups you’d be targeting; if you had 5 traffic sources, 4 time-since-last-visit segments, 2 products and 4 PD groups, you’d end up having 160 groups! Perfect, if it sounds like there’s too many, you’re probably doing it right.

BIDDING STRUCTURE

So, why have your Remarketing campaigns segmented so granularly? One word; bidding. Bidding on the display network affects both your cost per click and frequency of adverts shown to the users you’re targeting. If you’ve structured/segmented your audiences well, they’re have varying levels of performance.

So you’ll want to be bidding more on more on users who have higher page depths, even more on those users who visited by a better performing traffic source, and then even more on the users who visited more recently. Initially, anyway; review and update as you go. You’ll have something that looks similar to:

remarketing

You’ll see how your initial bids are tailored for the audience quality. Also now you have this level of control over your remarketing bidding, you may spot correlations between other dimensions (age, gender, location) which you’ll now be able to segment into more remarketing campaigns and increase the bid for – if desired – and increase the traffic from these users.

KEY THINGS TO REMEMBER

Remarketing campaigns should serve a measureable, specific purpose; if your segments have been derived from this objective - and support the overall goal – your campaign will perform well. Such as the example mentioned above (re targeting previous bookers from the past year), establish what you’re trying to achieve with your remarketing campaign (e.g, bring custeomers from last year back to the website) then from there design your segments (e.g, users who purchased between 365 and 730 days ago).

100 users per audience. While taking it to this level (in regards to segmentation) is optimal, if you don’t have at least 100 users who say, came from Google / CPC over the last seven days, who visited this one product page and 5 other pages on-site, you won’t be able to have an audience for just those users. So you need to scale your segmentation in-line with your websites traffic volumes. Though you don’t have to be as big as Amazon to have amazing remarketing audiences, but it helps.

Back to blog