Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Attribution case study | Ad:tech NY 2012 | Encore Media Metrics

2,496 views

Published on

Attribution case study presented at Ad:tech NYC in 2012 by Steve Latham, ceo of Encore Media Metrics

Published in: Business
  • Be the first to comment

Attribution case study | Ad:tech NY 2012 | Encore Media Metrics

  1. 1. Attribution Case StudyImpactful InsightsNovember 2012Steve Latham Bradley MayEncore Media Metrics KSL Media@encoremetrics @kslmedia
  2. 2. Background•  Agency: KSL Media•  Client: Retail Gasoline Brand•  Display Media (20+ vendors) –  Platforms: PC-Based + Mobile –  Placements: Branding + Direct response –  Formats: Video, Rich Media, Flash•  Paid and Natural Search
  3. 3. Objectives•  Campaign Goals –  In-Ad Engagement •  Brand metrics (awareness) •  Take action: Download mobile app, view videos –  On-site Engagement •  Download content and mobile apps •  View video spots •  Find a station •  Register•  Holistic View of Media Performance –  Conversion paths –  Role of each Channel and Platform –  Performance by Publisher and Placement
  4. 4. Conversion Path Insights •  Converters utilized numerous channels •  Among Converters who were exposed to ads: –  Served 6.7 ads –  Visited 3.2 times before converting –  Paid + Natural Search accounted for 28% of Visits Sources of Visits in Conversion Paths =====> Display Direct Natural PaidConversion Paths IMPs Visits Referring (Web) Nav Search SearchAll Visitors 6.7 3.2 1.0 1.0 0.4 0.4 0.3Relative Contribution 62.9% 37.1% 31.3% 31.3% 14.0% 13.8% 9.6%Includes Converters who were exposed to ads, grouped in natural clusters via machine-learning algorithm.Based on Modeled data (after excluding outliers and excess impressions served):
  5. 5. Frequency Analysis •  Optimal frequency was 6.7 Impressions •  Actual frequency was 35 –  Publishers: 12 –  DSP: 67 ImpressionsServedtoVisitorsbyFrequencyTier NumberofVisitorsperFrequencyTier350,000"" 9,000"" All"Others" DSP" All"Others" DSP"300,000"" 8,000"" 7,000""250,000"" 6,000""200,000"" 5,000""150,000"" 4,000""100,000"" 3,000"" 50,000"" 2,000"" 1,000"" !"""" !"""" " " " 5" " 10 o"9" 20 "19" 30 "29" 40 "39" 50 "49" 0" 9" 0" 9" 0" 9" 0" 9" 0" 9" >1 9" " 1" 2" 3" 4" 0" 10 "to"5 20 o"19 30 o"29 40 o"39 50 o"49 9 00 "9 "to "to "to "to t to " " " 5" " 10 "9" 20 19" 30 29" 40 39" 50 49" 0" 9" 0" 9" 0" 9" 0" 9" 0" 9" >1 9" " t t t t 1" 2" 3" 4" 0" 10 "5 20 "19 30 "29 40 "39 50 "49 9 to 00 " " " " "9 "to "to "to "to "to to to to to to
  6. 6. Channel Performance •  After attributing credit for Assist Imps and Clicks: –  PC Display Ads comprised 18% of Actions (excl. mobile) •  Exceeded Paid + Natural Search –  Display CPA fell by 83% vs. last click Ac#onsByChannel120,000 Last Click Attributed100,000 80,000 60,000 Cost%Per%Ac+on%By%Channel% $50 Last Click Attributed 40,000 $45 $40 20,000 $35 $30.54& 0 $30 Direct Nav Org Search Referrals Email Paid Search Display Mobile Display $25 $20 $15 $10 $3.05& $5 $2.54& $2.77&$2.94& $0.87&$0.86& $0 Email Paid Search Display Mobile Display
  7. 7. Vendor Performance•  After attributing credit for Assist Imps and Clicks: –  5 Winners (vs. only 2 using Last Click) –  4 Challengers –  4 Laggards Without Attribution, 7 of 9 vendors would have been cut from the plan Converters Assist Actions CPA:.Last CPA: Efficiency Vendor Assist.Clicks Rating (Last.Click) Impressions (Attrib.) Click Attributed Oppty Ad#Network#C 6 4 589 ############ 150 $3766.30 $150.42 56% Laggard DSP 2,593 4,248 23,095 #########9,237 $35.55 $9.98 44% Winner Publisher#G 241 73 4,460 #########1,002 $69.50 $16.71 n/a Winner Publisher#H 26 10 2,081 ############ 538 $4715.19 $227.89 55% Laggard Publisher#J 1 0 614 ############## 44 $9266.41 $211.51 42% Laggard Publisher#MA 10 2 895 ############ 314 $804.44 $25.58 52% Winner Ad#Network#ME 19 1 5,117 #########1,922 $2403.05 $23.75 44% Winner Publisher#MO 16 3 694 ############ 268 $529.79 $31.60 44% Challenger Publisher#P 139 211 8,225 #########2,318 $343.60 $20.61 73% Winner Ad#Network#R 32 3 2,217 ############ 594 $802.91 $43.26 30% Challenger Publisher#TE 63 19 2,495 ############ 542 $387.27 $45.04 36% Challenger Publisher#TA 12 37 112 ############## 72 $2021.62 $335.54 34% Laggard Publisher#Y 482 8 2,979 #########1,788 $142.23 $38.33 31% Challenger Totals 3,646 4,624 53,858 #######18,859 $141.59 $27.37 45%
  8. 8. Vendor (Publisher) Performance•  Ranked Vendors based on 3 criteria•  Recommended Actions and Freq. Cap for Each Cost$Per On$Site$CPA Cost$Per Action$toWeb$Display Passive (Attrib) Active$Int. take Recommended(Frequency( Audio/VideoAd$Network$C Laggard Challenger Winner Pause Publisher$Y$ 4.6$$DSP Winner Winner Laggard Increase Publisher$TA$ 9.6$$Publisher$G Winner Winner Increase Publisher$TE$Publisher$H Laggard Challenger Pause Ad$Network$R$ 7.9$$ Publisher$P$ 8.4$$Publisher$J Laggard Pause Publisher$MO$ 5.6$$Publisher$MA Winner Winner Increase Ad$Network$ME$ 5.4$$Ad$Network$ME Winner Winner Increase Publisher$MA$ 5.9$$Publisher$MT Challenger Laggard Optimize Publisher$J$Publisher$P Winner Winner Winner Optimize Publisher$H$ 8.3$$Ad$Network$R Challenger Laggard Pause Publisher$G$Publisher$TE Challenger Laggard Pause DSP$ Ad$Network$C$Publisher$TA Laggard Pause 8.4$$Publisher$Y Challenger Challenger Optimize 0$ 2$ 4$ 6$ 8$ 10$
  9. 9. Key Takeaways•  Start with a Holistic view –  Not all Channels and Placements serve same purpose –  Groups of publishers may have different objectives•  Display Ads are effective in creating awareness –  Accounted for 18% of Attributed Actions (excluding mobile) –  Performance varied dramatically by vendor –  Found significant over-serving to same users•  Search played a Supporting role•  Mobile ads were very efficient for Mobile actions –  App downloads and video views at very low CPA•  Frequency must be monitored closely
  10. 10. Additional Insights•  There is no “silver bullet” –  Must look at numerous metrics and factors•  It’s still Art and Science –  Requires oversight and subjective reasoning•  Frequency: too much can hurt performance –  Builds case for fewer, higher quality placements•  Offline promotion can skew on-site results –  Excluded specific Actions when modeling results•  Fragmentation = Obfuscation –  e.g. hard to see how Mobile gets lift from PC display
  11. 11. How Insights are Applied•  Evaluate and Segment Media Vendors –  Performance by Platform and Format –  In-ad (branding) vs. On-site (response)•  Optimize Media Plan –  Allocate budget to top performers –  Manage frequency more effectively –  Improve efficiency and buying power•  Improve partnership with Clients and Vendors –  Better accountability and transparency
  12. 12. Contact Info bmay@kslmedia.com Steve@EncoreMetrics.com @SteveLatham @kslmedia @EncoreMetrics http://KSLmedia.com http://EncoreMetrics.com http://Attribution101.com

×