This is from a joint presentation between myself and Martin Lundin, Head of Digital Marketing, Sony Ericsson at AdTech London, 2007:

How to Make a Good Impression: A user-centric approach to advertising.

The digital landscape has changed and so have the ways of measuring it. By seeing the consumer as an individual, technology allows brands to speak to consumers directly at their point of engagement and then allow them to choose how and when they want to proceed. By watching and learning user behaviour behind the scenes and with the assistance of profiling and retargeting, brands can now personally speak to individuals with the message most relevant to them and then gently influence them through the purchasing funnel. We will look at how in a recent campaign, we used this combination of passive learning and pro-active nudging to move consumers along that product awareness path.

Download the AdTech presentation here (3.87 MB) 

The article:

When online advertising was young, and advertisers had no choice of ad formats, when it was time to justify online ad spending and show tangible results, the choice was clear. When standard banners ads ruled, all viewers could do was click, and thus clicks became the established metric – someone saw an ad and did something. This metric was complemented by frequency of exposure.

Intrusive advertising like overlays or pop-ups became a better way of gaining higher clicks, as they presented the message right in the middle of what the user was doing.

The challenge of any ad designer is to get the viewer’s eyes off the web content and onto the ad. Advertisers have tried intrusive ways of making this happen – but users have become hardened and irritated by poor use of these formats. Now, it’s getting harder to find new techniques. Especially given today’s wide-reaching campaigns running internationally across multiple sites, demand is growing to establish standard ad formats and metrics across all sites to ease production time, speed campaign rollout, and facilitate ROI.

Video is a highly effective way of engaging users, as the television has shown us. This medium derives its powerful influence because it does not rely on visual alone, involving rather both sight and sound. With TV, you can’t measure effectiveness by direct response, at least not in the immediate sense. People don’t generally see a TV ad and pick up the phone – unless it is a charity ad or perhaps the shopping channel. Television advertising aims to place something deep within a viewers psyche, so that when faced with two brands in a shop later on, they naturally reach for the one to which they have been more exposed.

The web for the most part has been silent, and this has until now been a key restriction of transferring video online. It has offered only half the solution by engaging only half the senses – and few people ever turned the sound on.. Where a video has been developed specifically for an online ad it allows the user to interact with the ad and determine the outcome. This medium adds another sensory experience to the mix – touch – something only clever, direct response print pieces had been able to claim.In online, if click is the only natural measurement, the cost of that click becomes paramount. Google built a business case around that concept. However, click is transitionary – it is only a fraction of the definitive action, such as a purchase. But at least it offered some kind of tangibility to a response.

If we look at three key spenders on television to build brand; entertainment, automotive and financial – we see the first two effectively using display advertising online, but the latter has only recently seen benefits of online branding, favoring search as a justifiable way for measuring success. Why because of the click!In order to fully understand metrics and measurement we need to look at what journey the user takes – the click stream – assuming people are indeed clicking. With more and more personalised content online, Web 2.0 shows that users are building their own web environment and are more in control of their domain. With the advent of widgets and gadgets, users are now also on the desktop. The need to click and browse to find content is less as users now expect the content to be sent directly to them. This means we have to find new ways of engaging with users wherever they are and offer them the chance to interact and be in a position to carry out the purchase there and then. iTunes does this well. There is no need to leave the music programme, to open the browser to find and buy the music that you want. It is now all a seamless process.

Let’s try and break it down into simple terms. Say I was a farmer. Firstly I would need to seed the message. I may need to prepare the ground so it is fit for purpose, to sow that seed in position or scatter it out, so I would want to measure how many of those seeds landed, then find someway of measuring how many of those took root, and then how many I can harvest and ultimately how many of those turned into something useful – like bread.The metrics we need to look at are, for a start, impressions – how many ‘potentials’, but then we need to also look at reach and frequency – how many ‘actuals’ and how many times, and their influence.

One cannot assume every impression will be effective immediately or even measurable by response. What we currently do is to measure every click-through or interaction against the number of impressions. We do not measure the number of unique users who have in fact seen the ad. So, if only 50% of the impressions were seen by unique users, we want to determine if the first exposure or the second had more effect. And then measure that response. If we measure that 1% of impressions were clicked on, we still cannot tell if the majority of the response was by people who were seeing the ad for a second time – which would actually mean that 2% of the impressions were clicked on by unique users.

How many impressions, how many eyes?

Not everyone exposed to the ad will notice the ad or do “something” with it. Some people will not notice it at all, some will make a mental note for later, some will do something there and then. By determining that 2% of uniques is not enough. Of the 50% of people who saw the ad, 40% did not even notice it, 40% filed it away mentally for later and 20% stopped and looked by interacting. We tend to find that even though clicks are well below 1% now, 10-20% of people do however perform mouse roll-over.

When you start to drill down and find that 10% of impressions is Interaction rate or 20% of actual users ‘touched”, the questions we ask are what did they do and for how long did they do it? By measuring the number of responses against unique’s or even against unique interactions will give higher percentages if that’s what floats your boat, but it certainly gives a far more accurate picture of what people are doing. Then there is the need to really learn and understand what they are doing and just like any web analytics tool, look at the drop-off points and plug it suitably. Did the user turn sound on or video on? Did they open up another panel? Did they type in their email address for further information, but stopped when asked to provide their date of birth?

The time spent engaging with the creative, either sub-consciously from time spent on page (Ad Duration) or consciously via actual interaction (Dwell Time) also need to be brought out and standardised to ensure brand exposure is clearly defined, measured and attributed to a brand recall response much later in the user journey.

Armed with this information the creative agency is able to build better creative’s. The media agency can better analyse which sites are providing specific results and understand how to increase campaign effectiveness.

By analysing automotive campaigns an interesting trend is that email can better at producing data capture results than news sites, as users who were ready to type an email address moved into an ad and typed their name. It does not provide better clicks, after all if I am in middle of typing a mail, why would I want to click away, but rather distract me for a few secs, let me play and respond then get back to what I was doing… the objective here is creating a response where the user is and measuring it effectively. For the client would they prefer X number of email addresses or X number of clicks? Ah yes, objectives – to drive traffic or drive results?!

Analytics are far more advanced than most people realise. Most campaigns contain a wealth of data already but people are not using the data or do not know how. If you cannot prove effectiveness per site, media agencies cannot know if their decisions were correct and are unsure how to develop these analysis skills in the first place. Creative agencies might assume one size fits all, and not yet know the art of finesse – that email data capture should gain slightly higher prominence, but that on a homepage video a silent video of less than 5 seconds would be more applicable – why? Because all that data is at hand, and yet the questions are never asked. Then we scrape around in the dark in trying to help clients build better campaigns and get excited by creatives that look pretty, are innovative, win awards and do what for delivering tangible results for the client exactly? As the web moves towards an entertainment source, “sound on” by default will change with in-stream video and will ultimately be the only medium that actually engages three of the five senses simultaneously; sight, sound and touch.

Basic psychology is that the more senses or emotions involved, the deeper that drills down so that somewhere down the line it will cause a continual reminder. This is good news for advertisers who want to use brand recall to make the hand go towards one product over another at the crucial point of the purchase decision. Some may still remember “a finger of fudge is just enough to give the kids a treat”. So we may ask – what will tomorrow’s kids remember 20 years from now when they didn’t just see and hear, but actually had the chance to play and interact?

Recent studies have shown that impressions are 4:1 more powerful at driving conversions then actual clicks. Of those impressions, those with most impact were utilising rich media in practically every case. It shows us that even when exposed to the ad, people aren’t always ready to make a response right there and then. Perhaps they were reading the headline news and did not want to click away? But that impression counted – it worked hard at driving home something deep within a user, by engaging all those senses simultaneously that are measured later on by the conversion response. Bottom line: the click is going to be seen less and less, especially as the two channels, web and TV, combine. You can easily find plenty of great creative examples with appalling click results. When you look at the creative the reasons are clear; the content of the ad answered everything that the user needs to know there and then – so why would I need to click on it anyway? The question is did the ad really fail? New ways of measuring effectiveness have to emerge and be taken seriously and new ways found of making every impression count. This is the creative objective as opposed to just trying to get clicks. No doubt however that a request will come through today to answer the question ‘what is the benchmark for click-through of an MPU of an automotive client in France please?’


Trying to place this in some kind of order:

  • Impressions -> Unique’s (Reach) -> Frequency of exposure -> what did they do (overall / at each stage)
    • Interacted (rolled-over) -> Time (Dwell) -> Expanded -> found info -> played -> respond (Data capture)
    • Watched video -> turned sound on -> replayed -> multiple videos
    • Clicked-thru -> drop off (non-intentional) -> found info -> played -> respond (Data capture)

As you can see. there is a whole load of data in middle sections that can be understood as pre-click – and questions validity of click at all if users are given all the answers to their questions where they are. You can also see that click is not the end of the process, and fair comparison is to find out what they do post-click…