For anyone who has worked with online banner ads as a medium for a long time, there are things that we think we know about banners. We see the patterns emerge over and over but we can’t actually prove things categorically. I believe a recent campaign at the ad network I work for took a step closer to proving that banner ads have a significant and measurable effect on brand and product awareness.
To give you a bit of background, the client is Tourism Fiji (the beautiful tropical island), the agency is PMA Communications Group, and the execution was standard banner units. The sites for the campaign were selected using comScore behavioural analysis and the target market was the U.S., even though the campaign was booked in Sydney, Australia.
The campaign employed the standard banner measurements of impressions and clicks and these numbers were solid but not overly remarkable. What made the campaign break out was using social media monitoring to measure the level of discussion and interest around the brand. For this reporting, we used Statsit, a leading regional social media monitoring and insights company. For the purpose of the measurement, only forum and blog mentions were measured. We disregarded data from Twitter, as we did not consider it substantial in reflecting online awareness for the purposes of this campaign.
The campaign began in July 2010 and had a number of bursts before ending in October 2010. No TV ads or significant other mass marketing was run concurrently with the banner campaign.
Here are our key findings:
Banner ads create awareness evidenced in social media discussion. By the end of the campaign, the blog mentions and forum entries talking about Fiji were up 110 percent over the established baseline.
The awareness created was lasting. The level of discussion did not immediately fall back to its previous range once the impressions were reduced or stopped.
The awareness created was compounded. As the impression levels increased, so did the awareness. The awareness grew in a compound fashion building on the awareness levels of earlier impressions.
There is a time lag between impressions and online discussion. We can see a clear time lag between the spikes in the impressions and the corresponding increases in awareness. Interestingly, these lags seem to lessen as the compounding effect takes place.
I have included the graph showing the dates with the corresponding number of banner ad impressions run over those dates and the number of online blog and forum posts including the word ‘Fiji’.
We can see that this campaign strongly suggests there is a direct and measurable correlation between banner ads and an increase in substantial online discussion. We can see the relationship between volume increases in banners and the effect that has on awareness. We can also see that the awareness does not immediately fall away when the ad volume decreases and that awareness has a compounded effect.
However exciting this may be, I am not going to claim a complete victory yet. We intend to run further campaigns with measured exposure to the banner ad being an additional variable. Hopefully we can have a compelling update soon.