Think BR: Taking the guesswork out of online targeting

With so much online data available, it’s perhaps surprising that many advertisers use ‘modelling’ or guesstimates to create target audiences .

Yet many of today’s targeting platforms use this strategy, implying behaviour across tens of thousands of consumers based on the known actions of just a few.

It’s an approach more closely associated with mass marketing and might seem disappointing given the enormous potential for one-to-one targeting that the internet affords. 

However, much of our industry is still to rise to the challenge with platforms that are unable to harvest and handle behavioural data at scale or in significant volume.

It’s important to understand why modelled data is good at delivering reach, but falls short when it comes to finding a point of conversation between brand and buyer when someone is in-market.

In today’s highly fragmented media world, targeting is getting more complex, technologies more numerous and data readily more available.

However complexity can hinder and with so many options available, as an advertiser, it can be difficult to know where to start.

What seems to be increasingly lost amidst all the available data and measurement options is one simple but fundamental point: targeting is all about reaching people.

This should be the starting point and most pivotal element when developing audience buying strategies.

Real data vs modelled data

 

One of the greatest challenges with online advertising is how to match the right content or offer, with the right visitor, at the right time.

For advertisers, effective marketing is less about finding an audience and more about finding a point of conversion. In short advertisers want to see a positive yield from the time, resources and money they put into their advertising.

Reaching the right audience is only the first step to making this happen. Advertisers need to reach the right consumer, the one most likely to take the desired action. 

While lookalike modelling can be valuable and does have a role to play, even taking into account clever analytics and a skilled data analyst, modelled data can never be as powerful as actual information, the reality is there is no substitute for real data to define an audience accurately.

That’s because modelled data is based on ‘best guesswork’ and when used online, is particularly good for developing reach by helping to expand the size of an audience by finding ‘lookalikes’ with similar characteristics.

However, ultimately targeting is about relevance and precision, this is where real data comes into its own, because it is based on fact.

Customer profiles or modelled data, by their very nature, are less accurate and operate on the assumption that lookalikes behave the same was as real buyers. This can limit potential because modelled data:

• Excludes anyone that doesn’t fit the definition but may be in the purchasing funnel
• Includes people who do fit the definition, but who aren’t in the purchasing funnel

Real data can be used to understand and define precise audiences, based on a consumer’s online actions and activity over a specified period of time.

In addition real data takes into consideration recency and frequency to determine if/when the consumer is in the purchasing funnel.

This allows advertisers to deliver content that is relevant and appropriate and most importantly affords them an opportunity to reach the consumer at the most critical time.

There is no doubt that while modelled data can add valuable insight, its weakness is its inability to identify if the target is in the market to purchase a product or service, which can result in campaign wastage and increase spend for less return.

Ultimately, the accuracy of any model is actually driven by the proportion of actual data used to derive it. The smaller your sample size and the less real data used, the more guesswork is needed.

As online marketing and audience targeting matures, the need for relevant and real data is increasing. And as the need for this data increases so the need for better ways to collect, measure and analyse the data emerge.

The results will see advertisers better able to define who they are targeting and, perhaps more importantly, why.

Using real data, advertisers are able to effectively identify target audiences while they are active and instantaneously deliver the right message, to the right consumer at the right time. 

The constant evolution means that audience segments that are based on ‘best scenario’ lookalike targeting will continue to give way to solutions that are based on real and actionable data.

Stuart Colman, MD, AudienceScience

This article was first published on brandrepublic.com