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.

The truth about google keyword mapping


Published on

At SMX East, WebMetro presented an eye-opener for SEMs on Google's keyword mapping algorithm. Senior Vice President Michael Behrens shared how online advertisers can take back control of their Google PPC campaigns.

Published in: Business
  • Be the first to comment

  • Be the first to like this

The truth about google keyword mapping

  1. 1. Uncovering the Truth about Google Keyword Mapping
  2. 2. Multiple Options… Which Keyword to Choose Keywords Search Query• For A Typical Account ? – Almost every search query will match more than one keyword• If ‘wrong’ Ad-Group matches – May show sub-optimal ad-copy or landing page!• If more than one keyword is triggered over a short period of time – Makes bid management harder, more keyword chaos• Google publish rules for how keyword are matched to queries… 2
  3. 3. How Google Determines which Keyword wins the Auction? The Original Way After the 2008 UpdateUse keyword matching query exactly Use keyword matching query exactly (exactly matching) (exactly matching)Use most restrictive match type (for Use most restrictive match type (for same keyword) same keyword)Use longer keyword (Match Length) Use longer keyword (Match Length)**Ad Group Only Use keyword with highest Ad-Rank (Quality X Bid) 3
  4. 4. Multiple Mappings: Same Query Mapping to Different KeywordsVisualization of queries (pink) andkeywords (blue).Red lines show multi-mapped queries. Queries triggering more than one keyword Result in chaotic account behavior Severely reduce accuracy of portfolio modeling and bidding Create data integrity problems for analysis 4
  5. 5. Multiple Mappings: Example 2 Multi-Mapped Queries 5
  6. 6. Multiple Mappings: Example 3Two of the top campaigns have severe mapping issuesproviding little control and not allowing the bidmanagement platform to effectively execute models. 6
  7. 7. Campaign vs. Ad Group Multiple MappingCross Campaign Query Matching Cross Ad-Group Query Matching 7
  8. 8. Multiple Mappings: Quantifying the Frequency Keyword Counts by Query (30 Days) 60% 50% 40% 30% 20% 10% 0% 2 3 4 5 6 7 8 Non ‘Expected’ Multi-Match Queries 8
  9. 9. So to determine the severity of the problem and how much ispotentially caused by the Ad Rank Rule we built our own searchengine with old school match rules 9
  10. 10. Building a New Search Engine (Old School Rules)1. Decide sensible criteria for ‘ideal’ matching Rule (1) – Use keyword matching query exactly Rule (2) – Use most restrictive match type Rule (5) – Use longer keyword (match length)2. Program a new matching algorithm Use open source search engine Apache3. Run search query stream through algorithm Can take a long time, speed up using multi-core CPU4. Pivot table results with KPI’s Excel the results 10
  11. 11. Mapping Definitions in the New Search EngineDefinition DescriptionExpected Expected keyword was matched to query (as per rules 1, 2 & 5)MatchTypeLength A longer matching keyword (and/or narrow type) was foundExactMiss An exact match keyword was foundExactlyMiss An exactly matching keyword was found (broad or phrase)NarrowerType The same keyword with a more restrictive type was foundNarrowerTypeBMM As above but specific to broad match modified typeSession Keyword was matched through session based matchingDuplicate Duplicate keywords were foundAmbiguous More than one keyword matched the query equally wellUnclassified Unable to classify match definition 11
  12. 12. New Search Engine ResultsDefinition Cost Cost % Conv CPA CTR • Only 53% of the time did our new rulesExpected $ 108,441 53.13% 604 179.5 6.1% match Google’s actualMatchTypeLength $ 48,566 23.79% 113 429.8 14.1% match selection.Ambiguous $ 13,936 6.83% 37 376.6 8.9%ExactMiss $ 9,185 4.50% 30 306.2 5.3% • Match Type Length appears to be the mostDuplicate $ 8,812 4.32% 32 275.4 8.7% frequently over-riddenSession $ 7,665 3.76% 13 589.7 23.8% ruleUnclassified $ 4,152 2.03% 8 519.0 4.3% – CPA HigherNarrowerTypeBMM $ 1,842 0.90% 3 614.0 12.8% – Google’s Yield (CTR) HigherNarrowerType $ 1,190 0.58% 2 595.0 19.6%ExactlyMiss $ 323 0.16% 15 21.5 14.1%Grand Total $ 204,112 100.00% 857 238.2 7.1% 12
  13. 13. New Search Engine Results: Example Match Type Length MissKW ServedBest MatchQueries In this example, a shorter keyword took precedence over a longer keyword
  14. 14. How can I take back control…Use Negative KeywordsForce matching through negatives. The use of negatives is a great way to put up Google roadblocks and assist theiralgorithms in selecting the right ad group by not allowing the wrong selectionTightly Themed Ad-GroupsMaximize Ad Group match length rule. This will not help with creative as this rule only applies to the ad group, butsometimes the bid is the biggest challenge for search marketers when it comes to inconsistent mappingRemove Duplicates/AmbiguousSometimes a problem we uncover during account audits is the use of duplicate keywords. The problems is difficultenough to manage without having to double your challenges by giving Google another eligible keyword.Maintain Campaign BudgetsThis can be deadly and can easily be isolated by running and IS Loss to Budget Report. Using budget capping to controlinvestment by campaign is sub optimal from a pure mathematical perspective and is a huge catalyst to cross campaignmulti mapping 14
  15. 15. Thank youQuestions/Comments/Ideas…. 15