Hart Vice President of Media and Analytics Kira Clifton said that with the introduction of SmartBid, a smart bidding system based on artificial intelligence, Basis Technologies transformed from a trading partner to an ad-technology partner that helped to more quickly optimize campaigns.
“Whenever you do biddable media there’s a learning component,” Clifton said. “You can poll two people to see who’s going to be president, but you need enough data to gain a degree of confidence to be statistically accurate, and in biddable media that could take up to three weeks.”
There is a learning curve when introducing AI into campaigns and bidding, but SmartBid pulls data from the client’s previous campaigns to speed up the ability to optimize more quickly, she said.
Hart had a campaign running and then turned on SmartBid, the bidding feature in Basis Technologies’ demand-side platform (DSP) that leverages AI.
advertisement
advertisement
“It’s just one campaign, but the goal was to gain conversions, and we wanted registration-type data,” Clifton said. “Within the first couple of weeks, we saw the conversion rate increase more than 2.4x, and the cost per conversions fell by 75%.”
Clifton said that for every 100 people who previously saw the ad converted, with the use of SmartBid, the number rose to a 4.5 conversion rate. The campaign was for a client in the education space, but she believes it could improve clients in other spaces, too.
Basis upgraded its technology in February to run an AI engine, which enables faster processing of data to improve results.
AI typically is only used in the campaign’s own results to inform optimization, but the company’s latest upgrade enables a brand’s campaign to use the knowledge gathered in aggregate from all campaigns running through Basis. It may become a full-scale option for clients in the future.
Clifton views the ability to increase conversion rates as a challenge. “We do have data and can get smarter about what works,” she said. “Being stewards of clients’ money and treating it as it is my own, I want to know it’s working. I think of it as buying clothes and cost per wear.”
The challenges are centered around “pulling the correct levers.” There are hundreds, she said.
“I care about methodology and the human element,” she said. “It’s like a Sudoku puzzle. Your brain can see the challenges and you need to solve them. I want to understand the ‘and if’ statements, but AI is a little more difficult to learn, because you don’t have those same ‘and if’ statements.”