Cambridge personality research general presentation feb 2012
Ad TargetingusingPersonalityDataEvidence and Theory2011firstname.lastname@example.org
Cambridge Personality Research Personality and behavioural data >6.5 mn individuals Model for predicting behaviours, preferences and individual traits Online tool: www.preferencestool.com In use with global media agencies World leader in research
people and their actions are inter -connected in a Giant Global graph Websites Likes People Emails Searches Products / Services Workplaces
to describe an individual’s location isto predict her traits and behaviour
tools we use for describing people 100 dimensions of mapped behaviour exclusively from Cambridge University Research models personality mapping standard – the “big 5” trait dimensions openness conscientiousness extroversion agreeableness neuroticism social demographics
technique Matrix of > 35 million connections between objects behaviours & people Extract >100 best components of patterning Any 25 of our components explain enough variance to make a reliable prediction
accuracyWe predict which of thesetwo people is connectedwith BMW.Our accuracy is 93% Accuracy for other variables between 67 and 93 %
relevancewe gather data across the websearches browsing logs tweets shoppingrecords mobile sensors Facebook profilemodel applicable in all situationstargeting by psycho-demographics personalisesearch results add user descriptors demographicand personality predictions user understanding
value in ad targeting Value proven in Facebook ad targeting Best personality-based groups are stable and emerge early in campaigns Method integrated to ad-serving platforms Personality-based target groups score CTR up to 100% higher than Agency methods Conversions can rise by 45% CPA can be more than halved
example: CTR (food brand) CTR: Preference vs AgencyTheme of keyword group 3 0.19% 0.12% 2 0.13% Preference keywords: CTR 0.10% Agencys keywords CTR 1 0.12% 0.06% CTRA leading Agency’s Facebook fanning campaign compared its in-house keywordgeneration system against personality-based keyword lists generated byCambridge Personality Research Preference Tool in a trial of 3.5 mn impressionsDecember 2011. Keyword lists were generated for 3 campaign themes 1. Family, 2.Cooking, 3. Online Shopping. Taking Click Through Rate CTR as themetric, Cambridge Personality keywords outperformed agency by between 30and 100%. Enough to double the brand audience.
example: CPA (foodbrand) Cost per Action: Preference vs Agency Theme of keyword group C 0.51 0.89 preference agency B 0.82 1.08 A 0.80 1.92 $ per Social ActionA leading Agency’s Facebook campaign compared its in-house keywordgeneration system against personality-based keyword lists generated byCambridge Personality Research Preference Tool across 3.5 mn impressions inDecember 2011. Keyword lists were generated for 3 campaign themes A.Family, B. Cooking, C. Online Shopping. Taking Costs per Social Action as themetric, Cambridge Personality keywords outperformed the Agency keywords bybetween 24 and 58%. Enough to double the campaign ROI
example: conversions(insurance brand) Conversion rates: Agency vs Preference (3 top-performing segments) Stegment number 3 0.32 0.18 2 0.33 0.18 Preference 0.35 Agency 1 0.19 conversion rate: clicked, then entered competitionA leading Agency’s Facebook competition-entry campaign compared its in-housekeyword generation system against personality-based keyword lists generated byCambridge Personality Research Preference Tool in a full campaign of 42 mnimpressions in November 2011. The top three performing segments of each methodare compared. Taking conversion as the metric, target groups defined byPersonality using the Preference Tool outperformed target groups defined by theAgency by an average of 45%. Note that CPM was, however, 30 % higher onaverage for these top-performing groups.
current applications predict business-critical behaviour likeliness to repay credit card debt quantifypersonality of brands, products, competitors and audiences brand insight - positioning and media strategy recommender engines and apps ifyou like this music, you’ll probably like this [music, or other product eg car] ad targeting