By Dan Cooney, managing director, Crimtan Ireland.
It is important not to lose sight of the fact that technology is programmed by people and is a systematic enhancement of human competencies. Even sophisticated, machine-led technologies still require a human to oversee activities. This is true of RTB (real time bidding), which still needs to be ‘manned’. In fact, while RTB has enhanced online advertising, it lacks the ability to process at its optimum potential without human assistance.
Today, many organisations depend too heavily on machines and are ceasing to manually oversee pre-programmed processes. Businesses frequently make the mistake of setting programmatic to carry out campaigns by inputting targets and bidrates. While this will produce results, the outcome will be less than optimal. The complexity of online marketing requires not only technological prediction ability, but also the flexibility and response of a human-led approach.
Can you tell the dancer from the dance?
There is no denying the advances machines have provided to the online advertising industry, but there are many external variables that can affect success rates, which machines won’t register. These variables are often referred to as an outside context problem (OCP) and can affect conversion rates during RTB.
Take the weather for example; a DIY retailer may expect to see a peak in sales of barbecues during the summer months, but an unusually wet weekend could see sales slow. In this instance, unless a machine was adjusted to take account of this OCP, it would be unable to detect the cause of decreased conversion rates.
This is where human interaction can add a significant advantage over machines, as humans can recognise potential issues and manually overide the program to maximise revenue and ROI.
Ultimately, the best outcome lies in a collaborative approach that takes into account four key areas:
1) Target Practice
Programmatic is usually programmed to achieve maximum CPA and ROI. However, the bidder will focus on retargeting – which is a flawed approach due to the exaggeration of ROI, even when attribution is applied. Retargeting requires conscious human evaluation – taking into account relevance, accuracy and value.
2) Target Audience
RTB will automatically target the audience that statistically presents the highest potential conversion, regardless of whether or not they are the intended audience. This may not matter for direct response campaigns (DR), where the audience is less relevant, but as programmatic moves beyond DR it is important to remember that not all sales are equal when revenue is accounted for, and the target audience and desired outcome need to be judged against the intended audience – and this means manual input into programmatic campaigns.
3) Creative Intelligence
Bidders do not understand creative messaging beyond a simple understanding about how different creative impacts conversion rates. It is therefore important for marketers to maintain a tactical control over the delivery of relevant messages to the appropriate target audience, which can be enhanced with elements of machine learning.
4) Plan and Deliver
Planning a campaign’s pace of delivery is also important. For a machine, a sale is a sale, and any euro of ROI is equal to another, but most businesses do not operate in this way. If a business front-loads budget during a high demand period, it could find that demand exceeds availability and worse still, during low demand later in the year, the budget is spent and the business could find itself struggling to deliver revenue. Although this example is simplified, it provides insight to the necessity of pacing campaigns and making manual adaptations when needed.
Ultimately humans and machines need to work together for effective marketing optimisation. Not only will businesses achieve online marketing goals, but human understanding will greatly enhance the performance of programmatic in future campaigns.
This article appeared on Digital Times.