Digital Doubles Are a Good Alternative to Algorithms
With the ubiquity of artificial intelligence these days, algorithms are seemingly making almost every decision for us, often to the detriment of the customer experience, Forrester Research says in a new report.
“Algorithms are reaching a breaking point, so tuned to businesses’ objectives that they are degrading CX,” says Stephanie Liu, a Forrester senior analyst and lead author of the report. “Algorithms are optimized for business outcomes such as increasing revenue, maximizing clicks, or reducing costs,” goals that are often in conflict with what consumers want.
For that reason, Forrester suggests using digital doubles, which the firm defines as consumer-owned, algorithmically driven AI agents; consumers will be able to inject their personal context, preferences, and goals into third-party algorithms to shape personalization on their own terms. With these digital doubles, consumers can have better experiences when interacting with companies, while companies can get better zero-party data and enjoy marketing benefits, according to Forrester.
Liu says that algorithms have a number of other drawbacks as well, including the following:
- They can produce or reinforce systematic biases due to how they were initially constructed or trained, particularly when they use bad or incomplete training data.
- They can promote low-quality content and fake news when they optimize for ad revenue. This includes clickbait headlines and programmatic advertising that is often unchecked.
- They can spread AI-generated spam when they respond to search queries. AI-generated content is feeding AI-generated summaries, with predictably low-quality results, according to Forrester.
“The antidote will be the digital double,” Liu says.
Unlike algorithms, digital doubles are consumer-owned and consumer-driven, according to Liu. They will interact with companies on consumers’ behalf based on their unique needs and preferences.
And the companies that use them will also benefit. For starters, they can build revenue from selling through an app store or ads that could be tied to the digital double, or some combination of both, Liu says.
Other business benefits include the following:
- Digital doubles will enable marketers and CX professionals to focus on customer journeys across brands and platforms.
- Marketers and CX professionals will be able to do more with less data, and research will be more pointed as companies will already have key insights immediately available. Therefore, companies can be more strategic in exploring what they don’t know about their customers.
- Companies will be able to add better-quality data and have more confidence that these look-alike models more closely resemble their current customers. Currently companies use a blend of inferred and observed data to create target audiences that they suspect resemble their current customers.
- Companies can recognize similarly looking digital double requests (e.g., consumers who shop for similar products across similar retailers). This analysis enables companies to understand consumer goals, context, values, and needs.
- Digital doubles also democratize product delivery because they level the playing field for smaller or younger companies competing with much larger, more established ones. Instead of search engine algorithms dominating discovery, discovery will be defined by the companies themselves.
Liu also expects digital doubles to provide new opportunities for companies to differentiate based on seamless convenience.
Because of the business benefits, including more accurate data to drive identity, personalization, and customer understanding, Liu expects many companies, particularly retailers, to encourage consumers to use digital doubles, perhaps through loyalty programs.
While the benefits are clear, the timeline for implementation is a little hazier. Liu said she initially thought it would take three to five years for digital doubles to emerge, but some of her colleagues at Forrester think that consumers could start using them in as little as two years.