Programmatic is a term surrounded by noise. It goes along with persistent calls for more data, and more insight from across the industry. And along with that data, more management platforms -  generally more stuff.

With this stuff has come the early association of poor-quality, irritating advertising that is so ineffective as to induce users to download ad-blockers. It feels a little out-of-control – too powerful for human hands. This is because humans, consumers, are deeply complicated and bring with them a set of constantly shifting needs and demands. In a sense, what has happened is that we've discovered our own complexities, and these show no sign of falling into line any time soon.

Cognitive advertising sounds impressive. It relates to thought where programmatic relates, well, to programmes. There's not much of us in it. As IBM puts it: 'in the flood of data that surrounds us, there is a story that not everyone can hear.'

Last week, Warc spoke to Ross Webster, Managing Director of International Sales at IBM's The Weather Company about how a new approach to media buying is redefining the discipline. 

'Cognitive marketing platforms already exist,' he said, nodding to Persado, the automated copywriting programme, which 'works on the basis that emotion is key to inspiring engagement in marketing.'

The platform, its website says, 'generates language that inspires action.' With the world's largest database of over a million tagged emotional and motivational words and images, it scores its content against response data from over 40 billion impressions. This allows it to predict which messaging variations and combinations have the highest potential for success. The platform is also capable of adapting this information to touchpoint.

But Webster says the Weather Company wants to take it further: '[we have] recently launched Watson Ads, an industry-first technology in which consumers will be able to interact directly with IBM Watson through advertising.'

 A user can ask a simple question (for a human), such as: "What shall I eat for dinner tonight?"

'Through machine learning, language abilities and natural learning capabilities, the advert will understand consumers' intent, provide accurate responses, sort through flavours and ingredients using algorithms including taste and texture combinations, weather, time, your location, even the contents of your fridge.'

It is about understanding the user, 'vital for every marketer's success.' Taking the raw data through to intelligence has not been easy for marketers; IBM are trying to make these tools available and usable. The more Watson learns, the leaner and more effective the system.  

'For the consumer, the ad can make suggestions and give valuable insight; for businesses, it can provide marketing opportunities that were previously unavailable.'

In July, Mediapost reported IBM's findings following months of experimentation. Using Watson to buy digital ad-space, IBM claimed, reduced cost-per-click by 35% on average, and up to a 71% in some cases.

However, chatter across the ad-world urges caution. In his Admap prize winning essay (available to Warc subscribers here), Oliver Feldwick recommends building serendipity into personalisation, allowing people to discover brands by chance. The exchange must provide value to the individual, and that is, in part, the thrill of discovery.