What makes an ad campaign successful? In the online world, clients expect to see a return on their media spend; but how do you gauge the performance of a branding and awareness campaign? Believe it or not, this is difficult to do. If you see as many online campaigns as I have, and you were trying to optimize based solely on raw ROI numbers, your search campaign would probably consist of low-cost brand keywords and your display campaign would consist of low-cost, high-impression properties such as advertising networks. That may be the most efficient way to deliver the numbers, but is it the best way to run a campaign? Is it possible to step away from the numbers and still run an effective campaign?

Numbers, numbers, numbers...

Measurability is digital media’s ace in the hole. Where else but online can an advertiser get such instant gratification? Yet by the same token, measurement turns an online campaign into an ever-changing entity, optimized on the fly  based on the numbers… numbers, numbers, numbers . We use our online tracking tools to quickly weed out ineffective media placements and creative, and increase the levels of better-performing campaign components. But on the flip side, measurement can be detrimental to innovative campaigns based on video or social networking. While these new media can have positive effects on brand awareness or customer engagement, they may not show it when measured by conventional online metrics. And therein lies the problem: although innovation is crucial to the interactive landscape, it presents challenges to clients trained to judge a campaign’s success by backend metrics alone.

An alternate approach

Here’s what I’d suggest: set aside a percentage of your budget that you know will consistently hit your target metrics. Let’s say 80 percent. That way, you know you will be efficient with the lion’s share, but you will also have the flexibility, with the remaining 20%, to try out new properties, run a branding campaign or place ads on sites that make sense for your target. Metrics aside, there are often places you must be in order to be relevant, places where your target audience lives online. You could also handle the metrics differently by property. I mentioned earlier that ad networks always rise to the top in terms of efficiency, owing to high volume and low cost; but these metrics are driven mainly by post-impression activity. Post-impression activity occurs when a user sees an ad, does not click, but visits the site later and takes action – registers or makes a purchase, for example. Although there was no click, the impression is credited for the action. It is therefore quite possible to have a low click-through but a high conversion rate. It is safe to say that the “value” of an impression will be less on an ad network than on a more targeted site. The theory behind post impression is that the banner leaves a lasting impression on the user, who visits the site later as a result. In my opinion, though, for some properties what we are seeing is merely the billboard effect. How can we assume the user’s action was triggered by the banner ad he saw? I strongly believe that post-impression attributions should be higher for more relevant sites, where that “lasting impression theory” makes more sense. Adjusting the attribution percentage by site could level the playing field among all sites in the campaign. For example, ad networks would only get “credit” for say 20% of post-impression activities, while a more relevant property might get credit for 80%. Truth to tell, measurement is very complex, because we can’t simply hold every piece of creative or every site to the same metrics.

To conclude: do we, as interactive agencies, shoot ourselves in the foot by demonstrating our measurement chops? Would we run more effective campaigns if we just didn’t know what was driving the metrics? Or if we didn’t share as much data with our clients? I think media plans are often constructed with data and numbers weighted too heavily in the process. In the online space, we’re constantly looking at market research data, historical campaign data, demographic data, behavioral data, etc.

But what would a campaign look like if we just went with our gut?