Spectrum, Twelvefold Media’s managed service designed to target ad messages in real-time based on understanding consumers’ intent around the content they’re consuming, now is offering a self-service version of the platform. (See The Semantic Web Blog’s earlier coverage of the platform here.) With Spectrum 3.0, trading desks, clients and marketing cloud companies can use Spectrum’s listening and indexing capabilities — algorithms for determining why someone is reading a piece of content at that moment in time and for scoring millions of URLs daily — with their own bidding rules. It will continue to offer Spectrum as a managed service for always-on and spot campaigns, as well.
The wizard that used to be behind the process of understanding the mindset of the content to target – based on a series of data inputs stemming from Spectrum’s advanced understanding of natural language on the page, from which targeting schema are created – now operates in the background, so users aren’t required to enter in keywords or phrases to go up against. For instance, a smartphone vendor, leveraging an article on Apple slashing iPhone 5c orders, can add that URL to the system to go up against Spectrum’s big index of the visible web to find relevant pages like that one, says Mike Campbell, VP, product at Twelvefold.
Spectrum’s forecasting methods act to understand in real-time how consumption happens; learning what the opportunity is by watching pages be created and consumed and monitoring them for impressions as they come through. “That [Apple] news is a great opportunity for a competitor to run an ad,” he says. “Clients and agencies always wind up saying that they want ads to run ‘on pages like these,’ and this answers the ability to find that at an incredible scale.”
It’s one thing to know the news that Apple is cutting back on these orders, says CEO Dave Hills. But there’s more underneath such headlines, such as information about what people might not like about the phone or software glitches. To that end, Spectrum is boosting its ability to dissect content into subtopics, “to respond even better with different ads that optimize how they address that particular audience,” he says. That way, “you can press into the subtopic and create yet another content target that is even more finite. That’s powered by our forecasting capability.”
Spectrum also aims at helping self-service users understand if content targets are maintaining their relevance over time, by providing the capability to score pages on an absolute basis over time. “Content creation and consumption undulates so quickly, so now each month they can look at the target to see if it goes from an 8 to a 6.5 or whatever, and dig into what topics there are that have changed and adjust the content target automatically,” Hills says.
The company notes that it’s been steadily working to improve its natural language processing technology to help with content targeting. Regular keyword targeting doesn’t take into account, for example, that a word like ‘xylophone’ is rare and so would count as a stronger signal and a good match for content tracking. “We take into account relativity to how frequently a word is used in the English language generally, and we have to do it that way to understand phrases on a page vs. individual words, and you have to understand phrases to understand mindset,” says Hills.