1. Chinwag Facebook Marketing Conference London 6 October 2011 Stephen Haggard Director, myPersonality Cambridge model FOR BEHAVIOUR PREDICTION Predicting clicks and conversions by personality-groups
2. PATTERNS IN PREFERENCES & BEHAVIOUR SIMPLIFY Preference patterns to 100 dimensions CALIBRATE to personality data (> 6 mn FB questionnaires) CONFORM to “big five” personality model Psychometrics Centre, Cambridge University mapping web behaviour to personality
3. In theory…. Behaviours on FB will be psychologically patterned Those patterns may be predictable by personality We have a model that correlates web data to personality Could it help… Support brands with weak socio-demographic dimensions ? Improve performance metrics ? Microtarget FB individuals by intent not profile ? Test… 3 x FB campaigns with agency i-spy labs summer 2011 Using i-spy’s Upcast Facebook ad management tool Optimisefb targeting by behaviour & preference pattern
4. First trial: gadget sales Client: high street electronics retailer Aim: sales of popular portable gadgets to holidaymakers 20 “extreme” personality profiles to 9mn impressions Highest and lowest* performing personality-based segments * Segments served with over 200,000 impressions only are counted
5. Top personality and social-demographic segments SECOND TRIAL: TV SET salesRUN-UP TO FOOTBALL SEASON
6. Conversion looks interesting Conversion seems 25 % better in personality targeting Sample is small, so findings are tentative
7. Aim: 16-17 y.o. to “like” insurance brand: Uni room contents Personality segments on 50% of campaign (by impressions) Conversion and CTR data: clear gain for personality But disadvantage on CPM. Personality took 8% more clicks but 12% more budget overall Third trial: brand campaignclick to enter competition * Considered only segments with over 0.5 mn impressions. “Best” and “Worst” assessed by conversion rate.
8. OPTIMISATION OF Key target GROUPSinsurance prospects Age 16-17 Personality based targeting in the key segment improves CTR generally, but CPA improves in the best segments only. 8% saving in CPA 400 % gain in CTR * Best by CPA
9. Personality-based FB optimisation… Cannot - Wave a magic wand. Individual differences remain a marketing challenge; social-demography may be okay for some FB campaigns Is already proven in FB marketing with Cambridge model - Predicting preferences by applying personality filters Forecasting behaviour (especially conversion) Integrating personality management to campaign techniques Will be - Available soon in automated FB ad management programmes Further improved by analysis and research at Cambridge Extended to other channels eg search, mail TENTATIVE CONCLUSIONSCAMBRIDGE MODEL FOR behaviour Prediction www.mypersonality.org stephen@mypersonality.org
Notes de l'éditeur
Personality does predict behaviouron FB around ad-response (segments were “blind” to campaign content, but still produced a spread of results)The top performing segments emerge early, are stable, and not at cost of inflated CPCCan outperform CTR of conventional targeting significantly (0.018 was CTR average in campaign, best personality group is 22% better at CTR = 0.022)
Client: same high-street electronics retailer Total 50 MN impressions to football fans20% of spend allocated alternative personality based segments. Research approach: not aimed to maximise sales, just to maximise learning.Social-demographic portion of campaign: 0.034 average CTR. Personality groups average matched this
Side-note. Value of personality as a different slice: the sales were 50% to F segment; social-demography had excluded F, but this personality type might be a valuable client.
Hard market – 16-17 financial product salesPersonality segments overlaid on social-demog criteria: measuring the value added by personality filterPersonality gives advantage CTR – 60 % improvement on S-D segments. Stable resultHigher advantage on conversion – 85 % improvement on S—D segments. Stable resultNote spread: top segments But cost issue.
Across all top performing personality segments, CPM was slightly higher in personality segmentsBut in key target sector client asked agency to emphasise, there is optimisation for the best personality segment CPA In the 16-17 target group overall, personality segments on average got massively higher CTR, slightly lower conversion, slightly higher CPAIn the best performing segments of the 16-17 target group, personality segments got lower CTR but equal conversion and a slightly lower CPA
Optimisation panaceas are suspect – statistical view of any kind optimisation is: lots of incremental gains over time. Personality will be no different. But it clearly offers marketers some gains, and we at Cambridge will be involved in moving it forward. We invite marketers to work with us on using personality-strategies to their advantage.Try us out on facebook, take our personality test App there, play with our free facebook tools such as likeaudience, see some stuff at mypersonality.org, get in touch with me.