A good way of illustrating what’s possible is using an example from TUI – one of the world’s biggest travel groups, which incorporates Thomson.

TUI offers a very diverse range of holiday packages which brings with it a unique set of problems: notably how to market this increasingly diverse product range, to an increasingly wide demographic, through an increasingly noisy web.

To tackle this, TUI experimented with personalising its online messaging to individual consumers based on a technique involving predicting what people might be interested in based on their previous behaviour across a range of online activities.

These activities including browsing, searching, sharing, purchasing and knowledge of their location, all of which are known as ‘signals’.

A logged-out world

These days, unless you’re Facebook or have a paywall, it’s becoming much harder to get people to actually log in to your site.

This represents a problem for TUI because around 90% of site visitors are anonymous at any one time.

Although there’s more visibility on returning visitors in terms of their search behaviour or on-site activity, very little is still known about them as individuals.

The solution was to use standard display ads but overlay them with dynamic elements that can change the colour, content, product, and messaging, depending on the individual user.

Essentially, by tailoring overlaid content to individual users, TUI was able to match up its broad product range with its huge site audience of individual users to achieve ‘mass personalisation.’

Making ads more relevant

This personalisation was made possible using the ‘signals’ technology I mentioned earlier which predicts the type of product that an individual site visitor might be interested in, having mapped and scored their previous online behaviour.

It can take that data and programmatically deliver the appropriate ‘overlay’ content boxes in real-time.

Varieties of Thomson ads

This approach allowed TUI to go a step further still, not only just tailoring content to unique users but also to the environment in which they found themselves.

Take for example a single user who has shown an intent to travel. At work, we might choose to serve that user a message saying, “Tired of working? Plan & book your next vacation with us now.”

A few hours later, that same user might be at home with his family, browsing on a mobile or tablet. At that point it might be more appropriate to tailor the ad-text to something like, “Take your family to Florida this year!”

That’s fundamentally the same ad, for the same brand, using the same campaign – but with a completely different message.

The ad can also be adapted based on where a person is in the process of making a purchase so that the most effective ad can be served to move them closer to purchase.

For example, someone who’s been researching Florida activities (e.g. golf) but not booked a hotel might see an ad message along the lines of, “Check out Florida’s best golfing hotels,” containing golfing imagery and hotel prices for golf resorts.

When you begin to add day-part analysis, screen shifting formats, tonality, colour scheme, demographic and offers, you can begin to build up truly contextual messages based on predicting what the user might want to see at any given time.


As a result of this campaign, TUI experienced a 17% increase in conversions compared to a standard “non-overlay” ad approach – i.e. one that didn’t personalise.

The click-through rate was 21% higher than normal and TUI generated 546 extra lands on its site – through 90 different destinations being shown.

TUI’s success was essentially down to finding relevant individuals who had shown an intent to travel, and engaging with them at the moment they were ready to select their holiday – before they had visited the Thomson website.

This engagement included live pricing information and a call to action based on the intent signals and a link to a specific landing page.

TUI’s results, impressive as they were, actually pale in comparison to some others. Western Union, for example, saw a 97% uplift in site conversions using overlay ads.

Western Union was particularly clever in using an “action count message” to prove how popular the service was by showing how many people had registered in the last hour.

Music distribution platform, SoundCloud saw a 28% uplift in clicks for a campaign promoting a new album from the Virgin Records stable. The SoundCloud ad allowed people play a specific song from the album.

Citrix’s GoToMeeting campaigns using overlays have seen Cost Per Action half that of their target (£30 vs. £60).

As an average, site visit uplift using these techniques is about 15%, whereas travel sees a 33% average uplift across a mixture of site visits, conversions and clicks.

As the Adobe study alluded to, marketers are increasingly realising that more relevant and personal ads work better, however, it can seem overwhelming or fanciful for those operating in a company with huge product ranges.

However, if TUI can do it fairly easily across 90 different destinations, it can be smooth sailing for anyone.