Flash Series: Using Ad Testing Data To Set Up Foolproof Tests
Download as PPTX, PDF0 likes1,194 views
The webinar by Brad Geddes focuses on effective ad testing strategies to enhance PPC campaigns. It emphasizes the importance of defining segments, metrics, and confidence levels for testing, encouraging the use of both single and multi-ad group testing approaches. Key topics include analyzing consumer behaviors, localizing ads, and the significance of brand presence in non-branded ad groups.
Flash Series: Using Ad Testing Data To Set Up Foolproof Tests
- 1. #thinkppc Using Ad Testing Data To Set Up Foolproof Tests HOSTED BY:
- 2. #thinkppc Presenter • Brad Geddes – Founder of Certified Knowledge – Official AdWords Seminar Leader – Author of Advanced Google AdWords – Hero Conf London Keynote – @bgtheory
- 3. #thinkppc Join the conversation • Include the hashtag #thinkppc in your Twitter tweets. • Or use the webinar question box to send us questions.
- 4. #thinkppc Live Poll Question #1 How do you manage your PPC account(s)? #thinkppc a) I’m part of an in-house PPC team b) I do all the marketing including PPC c) I work at an agency d) I’m a consultant
- 5. #thinkppc Ad Testing is about Improvement and/or insights
- 6. #thinkppc Single Ad Group Testing
- 7. #thinkppc Multi Ad Group Testing Idea 1 Idea 2 Idea 3 Idea 4 Idea 5
- 8. #thinkppc Use Cases Single Ad Group Testing Insights within an ad group -Best for high traffic ad groups -Best for Brand ad groups -Good for other ad groups after you’ve done some multi-ad group testing for global insights Multi-Ad Group Testing Insights across a segment -Best for template created ads -Best for small data accounts -Best for market research
- 10. #thinkppc Ask the Big Questions First • Are consumers price sensitive? (discounts, prices in ads) • Does Location matter? (use geographies in ads) • Does shipping matter? (Free or Rush shipping) • Is selection important? (large, exclusive, variety, etc) • How much does quality matter? (cheap, luxury) • What’s your best CTA? (Call, Shop, Find, Discover, etc) • Should your display URL be products, geo, CTAs, or something else? • Do we need to pre-qualify clicks? (b2b vs b2c) • etc…
- 11. #thinkppc Should You Use Customizers? Testing Problem: • 804 ads in test 1 • 75 ads in test 2 Solution: Use same ad groups for your test
- 12. #thinkppc Exact Same Ad Groups in Test Static Ad is Higher Conversion Rate, Lower in CPA Customizer is Higher in CTR & CPI What metrics are important to you?
- 13. #thinkppc Should You Localize Your Ads? Lead Gen Company 1: GEOs are Clear Winners Next Step: Change a few campaigns to only target singe region with non-geo keywords and rerun test to see if there’s a difference. Then decide account organization.
- 14. #thinkppc Company 2: Is the data consistent across devices? Ideas from presenters are Ideas to Test: Not absolute facts for you to implement! Desktop Mobile Always think through results. In this case, ads were localized – landing pages were not. GEO increased CTR, but nothing else. Is that the market or the website? Solution: Redo test with localized pages
- 15. #thinkppc Should You Use Your Brand in Non-Brand Ad Groups? First Test, it appears that the brand is so strong that you should use the brand name in non-branded ad groups. It’s a good idea to test your results to ensure they are consistent.
- 16. #thinkppc Don’t Just Do One Test - Same test: different results Results aren’t 100% in yet… However, non-brand ad 1 is an ‘incentive’ ad. • This is for a travel company. Problem with many bookings is people are afraid that the price will drop and they will overpay. • Ad 1 talks to that issue with a coupon if the price drops
- 17. #thinkppc How to Start Testing
- 18. #thinkppc Determine Your Segments Basic Segments: • Your brand • 3rd party brands • Products • Informational • Long tail • By buying cycle entrance • Etc… Ad Type: • Text ads • DSAs • Image ads • Call only ads • HSAs • etc… Devices: • Mobile • Desktops • All
- 19. #thinkppc Define Your Testing Metrics Metric Best Usage CTR: Click Through Rate Max Traffic (note: CPI is usually still better) Conversion rate Most conversions from landing page CPI: Conversion per Impression Most conversions possible CPA: Cost per Conversion Lowering cost per conversion ROAS: Return on Ad Spend Hitting ROAS goals RPI: Revenue per Impression Maximizing revenue
- 20. #thinkppc What Minimum Data To Define Metric Impressions Clicks Conversions Timeframe CTR Yes Optional Yes CPA Yes Yes Conversion rate Optional Yes Yes CPI Yes Optional Yes ROAS Yes Yes RPI Yes Optional Yes
- 21. #thinkppc Minimum Data Guidelines Impressions Clicks Conversions Low Traffic 350 300 7 Mid Traffic 750 500 13 High Traffic 1000 1000 20 Well known brand terms 100,000 10,000 100 - 1000 The MINIMUM Data – It’s OK to have HIGHER Minimums|
- 22. #thinkppc Define Minimum Confidence Levels Term Type Minimum Confidence Long Tail Keywords 90% Mid data terms 90% – 95% 3rd Party Brands you Sell 90% (small brands) to 95% (large brands) Top Keywords (the ones you watch daily) 95% - 99% Your Brand Terms 99%
- 23. #thinkppc Define Maximum Data •Option 1: –3 months •Option 2: –Define maximums for each metric –Define timeframes
- 24. #thinkppc Test by Device (if appropriate) Count of Ad Row Labels Mobile All Effective Device Ad Group 1 2 All Ad Group 2 None Ad Group 3 1 2 Mobile/Desktop Ad Group 4 3 1 Mobile/Desktop Ad Group 5 10 All Ad Group 6 2 6 Mobile/Desktop Ad Group 7 1 7 Mobile/Desktop Ad Group 8 1 6 Mobile/Desktop Ad Group 9 7 All
- 25. #thinkppc Actions • If below minimum data, wait. • If above minimum, calculate confidence factors –If below minimum confidence – wait –If above minimum confidence – take action • If above maximum data and a test has no results, take action • Action = Pausing loser OR pausing loser & creating new challenger –In the case of multi-ad group testing: Replace entire string with a new pattern to test –(optional): Use insight in other areas, such as email, landing pages, etc
- 26. #thinkppc Recap • Determine what you want to test: –Single ad group –Multi-ad group • Define: –Testing metrics –Minimum data –Maximum data –Segments •Confidence levels per segment • Run Your test • When you reach minimum data: –Run confidence calculations •If winners & losers – take action •If no winners – wait –When you reach max data •Pause test and reset • When you have winners in Multi- Ad Group testing: –Apply that data to other channels / insights
- 28. #thinkppc Have more questions? Contact Us Directly • Webinar Feedback: marketing@hanapinmarketing.com ____________________________________________ Thank you for attending our webinar! #thinkppc