Pinterest’s expanded targeting options are supplied by its Taste Graph that has indexed more than 100 billion pins.

Pin­ter­est is increas­ing the num­ber of inter­est-based ad-tar­get­ing options offered to adver­tis­ers from rough­ly 400 to more than 5,000, thanks to the social network/search engine’s updat­ed index of more than 100 bil­lion pins.

Pinterest’s expand­ed set of inter­est-tar­get­ing options will roll out over the next few weeks and be avail­able through Pin­ter­est Ads Man­ag­er and for cam­paigns pur­chased through the adver­tis­ing API or direct sales team.

Enhancing Pinterest’s ‘core data asset’

Pow­er­ing the expand­ed inter­est-based options is Pinterest’s Taste Graph, an index of more than 100 bil­lion pins and cor­re­spond­ing meta­da­ta, such as the boards those pins have been saved to and the search terms peo­ple used before click­ing on a pin.

The Taste Graph is Pinterest’s core data asset. It’s how we actu­al­ly con­nect these 200 mil­lion peo­ple [who use Pin­ter­est each month] with 100 bil­lion-plus ideas on Pin­ter­est by actu­al­ly mod­el­ing how the world’s tastes and inter­ests evolve over time,” said John Mili­novich, Pinterest’s prod­uct lead for Taste Graph.

In the past, Pin­ter­est had used Taste Graph to make organ­ic pin rec­om­men­da­tions and cre­ate rough­ly 400 inter­est-based tar­get­ing options that were seg­ment­ed across 29 top-lev­el inter­est cat­e­gories. But last year, after the com­pa­ny acquired mobile ad-tech firm URX co-found­ed by Mili­novich last year, it put Mili­novich and team to work on enhanc­ing Taste Graph’s abil­i­ty to parse more spe­cif­ic inter­ests and offer adver­tis­ers more gran­u­lar inter­est tar­get­ing.

Pre­vi­ous­ly, if a brand want­ed to reach fash­ion-for­ward women, it could have select­ed the “haute cou­ture” or “run­way fash­ion” inter­est options; now it could pick out the more spe­cif­ic “french chic style” option. Sim­i­lar­ly, a brand look­ing for plant-based grill mas­ters might have been sat­is­fied with com­bin­ing the “bbq grill” and “healthy recipes” inter­est options, but now it can choose the more to-the-point “veg­e­tar­i­an bar­be­cue” option.

In addi­tion to being able to iden­ti­fy more nuanced inter­ests, Pin­ter­est has worked to make Taste Graph bet­ter able to under­stand how people’s inter­ests change over time. For exam­ple, if some­one uses Pin­ter­est to research kitchen remod­el­ing ideas, that user may not want to con­tin­ue see­ing organ­ic and paid pins relat­ed to home decor long after the remod­el is com­plet­ed.

How Pinterest’s expanded interest set will roll out

Pin­ter­est is already apply­ing this more time-aware inter­est pars­ing to refine the tar­get­ing of exist­ing cam­paigns that use its old­er set of inter­est cat­e­gories.

The new set of 5,000-plus inter­ests should roll out com­plete­ly before the end of Octo­ber, Mili­novich said. As the expand­ed inter­est list becomes avail­able, Pin­ter­est plans to noti­fy ad buy­ers through email and in-app noti­fi­ca­tions.

Those inter­ests will be pre­sent­ed in two ways in Pin­ter­est Ads Man­ag­er, sim­i­lar to the exist­ing pre­sen­ta­tion. Pin­ter­est will dis­play a drop-down menu of top-lev­el inter­est cat­e­gories, and ad buy­ers will be able to click on these entries to see a list of relat­ed sub-inter­ests. Addi­tion­al­ly, ad buy­ers will be able to use the search bar to query inter­ests, which will auto­mat­i­cal­ly appear as char­ac­ters are typed. Even­tu­al­ly, Pin­ter­est expects to make all 5,000 inter­ests brows­able through the drop-down menu and to auto­mat­i­cal­ly sug­gest inter­ests based on a campaign’s oth­er para­me­ters as well as the per­for­mance of a brand’s oth­er cam­paigns, said Mili­novich.

While test­ing the new inter­est cat­e­gories, some adver­tis­ers’ click-through rates increased by more than 50 per­cent, and the price some adver­tis­ers paid per click decreased by more than 20 per­cent, accord­ing to Mili­novich.

Algorithmic sourcing, human editing

Pin­ter­est is using an algo­rithm to expand its inter­est-based tar­get­ing options at a sen­si­tive time. Last week, Face­book came under fire for its algo­rithm cre­at­ing tar­get­ing options that enabled ads to be tar­get­ed to peo­ple who list­ed “Nazi­Par­ty” as their employ­er and “Jew hater” as their edu­ca­tion­al field of study. How­ev­er, Pin­ter­est was mind­ful of the mess an unchecked algo­rithm could make even before Face­book illus­trat­ed the point.