Keys to a Successful Landing Page Andrew Eisner Vice President, Client Services.

Download Keys to a Successful Landing Page Andrew Eisner Vice President, Client Services.

Post on 28-Dec-2015

212 views

Category:

Documents

0 download

TRANSCRIPT

  • Keys to a Successful Landing Page

    Andrew EisnerVice President, Client Services

  • Optimost BackgroundOptimost helps customers improve their conversion rates online through real-time testingCompany founded in 2001First on-demand multivariable optimization solutionMarket leader most diverse client base

    Entire company and platform built with mindset of multivariable optimizationMethodology: developed with experts from Harvard Business School, Columbia Business School, Yale School of ManagementTechnology: content management, analyticsFull-service: dedicated client services team to support test design, production, and analysis

  • Over 150 Customers

  • Key Concepts

  • Not Just One Decision.

  • Many MinidecisionsYESNOYESNOYESNOYESNOYESNOYESNOYESNOYESNO

  • keyword group: legal musickeyword group: burn musickeyword group: buy musickeyword group: radioNot Just A Landing Page

  • Radio KeywordsWinning Bullet PointsLosing Bullet PointsEXAMPLE: Winning bullet points varied by keyword segmentRadio KeywordsBuy Music KeywordsBuy Music KeywordsLegal Music KeywordsLegal Music Keywords

  • EXAMPLE: Adding price information to bullet points increased response in Yahoo segment, even though it had no impact in non-PPC segment

  • Dont Need To Do Complete Redesign

  • Control Winner +16.2%

    Vs.EXAMPLE: Minor Changes Make A Major Difference

  • ControlWinner +36.7%EXAMPLE: Minor Changes Make A Major Difference

  • EXAMPLE: Minor Changes Make A Major DifferenceControl Winner +16.2%

    Vs.

  • What Works (and Doesnt Work)

  • CopyAmount of Copydo you need it all?Are Benefits Really BenefitsHeadline/Header Construction

  • EXAMPLE: Keeping instructions simple worked best

  • EXAMPLE: Removing No Spam improved performance

  • EXAMPLE: Satisfaction Guaranteed is important benefitSubscriptions +11.1%

  • EXAMPLE: Changing headline into question improved responseVs.baseline

  • Vs.baselineEXAMPLE: Small header changes can make a difference

  • LayoutOrdering of CopyKeep It Simple

  • EXAMPLE: Order of Copy Blocks had substantial impact on response ratesClicks to Order -10.6%ControlClicks to Order +8.4%

  • Vs.UCTR = + 35.4%baselineEXAMPLE: Order of Copy Blocks had substantial impact on response rates

  • EXAMPLE: Simple Layout Performed Best

  • ImagesPeopleCelebritiesCEOs, Writers, etc.Endorsements

  • EXAMPLE: Barry Williams Had a Positive ImpactC = 4Confirmation = Cart =

  • EXAMPLE: Authors Picture Had Negative EffectBaselineClicks to Order = +3.1%

  • EXAMPLE: Endorsements outperformed Free messaging

  • What Works (and Doesnt Work)

  • Submit ButtonsBe Cool At Your Own Risk CopySizeLocationFlashing Buttons

  • EXAMPLE: Click Here buttons did significantly betterVs.baseline

  • EXAMPLE: Red order button was successfulVs.baseline

  • EXAMPLE: Animated button did very well in promo group, but poorly in generic groupVs.baselineUAR promo group = +7.8%UAR generic group = -5.2%

  • How to Optimize

  • The Traditional ProcessA/B Testing

  • The Traditional Process A/B Testing

  • The Traditional Process A/B TestingNOTE: for illustrative purposes only - not real data!!!!

  • Why A/B Testing Is Not Optimal-10%+20%+10%NOTE: for illustrative purposes only - not real data!!!!

  • Why A/B Testing Is Not Optimal0%+20%+10%NOTE: for illustrative purposes only - not real data!!!!

  • Why A/B Testing Is Not Optimal Serial A/B TestingMonth 1: ImageMonth 2: HeadlineMonth 3: Form LayoutMonth 4: Submit ButtonMonth 5: Endorsements

  • A Better Solution:Multivariable OptimizationImages UsedLayout; including Right/LeftMain headlineCopy and OrderOf Sell Points

    Endorsements

    Form Headline CopyForm LayoutSubmit ButtonsInclude Phone Number?

  • A Better Solution:Multivariable OptimizationMain Headline:Lose 10 Pounds by April 20Want to Lose 10 Pounds?Get the Perfect Diet for YOUGet Your Free Diet ProfileCopy and Order of Sell Points:What makes eDiets different? - 24/7 Support - Convenience and Privacy - No meeting to attend - No expensive foods to buy

    What makes eDiets different? - Customized meal plans and shopping lists - 24/7 Support - Convenience and Privacy - No meeting to attend - No expensive foods to buy

    Top 5 reasons to join eDiets - Customized meal plans and shopping lists - 24/7 Support - Convenience and Privacy - Personalized fitness plans and workouts - Expert health and fitness advice

  • Potentially Millions of Possible Permutations

  • Much Fewer Needed for Reliable Results

  • Much Fewer Needed for Reliable Results 35% 29% 29% 31% 30% 30% 34% 38% 30% 29%NOTE: for illustrative purposes only - not real data!!!!

  • Multivariable Optimization

    Winner+15%NOTE: for illustrative purposes only - not real data!!!!

  • Multivariable Optimization

    WinnerIndividual Effects-4%+15%+9%-3%-4%-3%+5%+15%NOTE: for illustrative purposes only - not real data!!!!

  • Multivariable Optimization

    WinnerIndividual Effects-4%+15%+9%-3%-4%-3%PredictedWinner+5%+15%+29%NOTE: for illustrative purposes only - not real data!!!!

  • Why Isnt This Done In Real World?

    Real WorldInternetCreative cost would be prohibitiveTechnology can result in huge efficiency in creative developmentGetting large enough sample would be prohibitiveFinding target audience is easy (and can be performance-based)

    Guaranteeing balanced sample size is difficultTechnology easily manages thisTimeline for response loop is too longReal-Time TrackingInstant Iterating

  • Benefits of Multivariable OptimizationAbility to test multiple variables and values simultaneously, without increasing sample sizeTimeline significantly reducedCan test Non-Obvious factorsFind best solution out of much higher quantity of versionsActual data on best out of 200 rather than best out of 2Ability to predict the best out of 2 million rather than best out of 2Learn where to focus developments efforts going forwardDiscover how much each variable contributes to performance

  • Methodology Is CriticalTest Population Evenly DistributedDay of WeekTime of DaySource of TrafficChoose Values in Logical, Methodical FashionMultivariable Testing: Cant Choose Versions At RandomIs It Statistically Significant?

  • Final ThoughtsBe Adventurous!Limited Downside and Unlimited Upside

    There Are Always Ways to ImproveUntil youre getting 100% conversion rates, you can do better!

  • Thank You! www.optimost.com andrew.eisner@optimost.com

Recommended

View more >