LONDON: Retailers can boost online campaign margins by up to 900% - that is the extraordinary claim being made for a new predictive analytics engine that has been four years in the making.

Online retail specialist Summit worked with team of academic statisticians to develop Forecaster, which CEO Hedley Aylott described as a "crystal ball" capable of showing where and how to spend online ad budgets to deliver the greatest profit.

It takes vast amounts of internal and external data – from sources such as Google, transaction history, weather, location, stock availability and current TV ads – and then applies complex statistical models to predict the likely sales from a given marketing spend.

To achieve the predicted sales, the platform then recommends how to adjust the bid level and messaging for paid search, display and product listing ads so the profit from each channel is maximised.

Aylott reported that in trials with major retail clients Forecaster had increased margins from campaigns by as much as 900%.

He pointed to several unique features, including the number of customer buying triggers and the way it used TV.

"We 'fingerprint' TV ads by client and their competitors," he explained. "When a TV ad appears, we instantly make changes to the client's online ads."

So, for example, when a competitor advertises a tablet on TV, Summit can then increase bids for their client's tablet ads for a set period of time in order to capture people who go online to buy after seeing the TV ad.

Summit said the biggest increase in search volume occurs within ten minutes of an ad, particularly at peak viewing times between 8pm and 9.30pm. Further, 30-second ads were shown to increase search traffic by 11% more than 20-second ads.

It cited a campaign for catalogue retailer Argos, using the Forecaster system, which increased paid search traffic by 33%, sales by 31%, lifted conversion rates by nearly 5% and the average order value by 18%. Ultimately, the cost of each sale fell 14%.

Data sourced from Summit; additional content by Warc staff