Google shows ads to people outside your geotargeting if they deem the search to be “relevant”.
What? Really?
Yep, I read about this recently and immediately called our AdWords people, and they confirmed it. AdWords parses all searchers’ queries- if a searcher is outside your geotargeting, but their search query appears to be relevant, they may still serve your ad to them.
For example, say I want to serve ads about Myrtle Beach Hotels only to people in North Carolina – I’m creating a campaign specifically for North Carolinians with ads about a gas credit to save money on the drive down to Myrtle Beach. With AdWords’ current set-up and the query parsing exception, if someone in California searches for one of my keywords, like “myrtle beach hotels”, AdWords may still show them that ad. This happens despite the fact that the ad is for a gas credit that no one in California would ever use. Ridiculous.
Here’s some background info:
- “How to Prevent Adwords Ads from Appearing Outside your Geotarget” – Admittedly this is pretty inflammatory writing, but…
- What AdWords Help says. Check out the second bullet point in that answer.
One of our AdWords reps told me that the percentage of my clicks from outside by geotargeting should be small. So I checked, and the results are below. I’d be interested to see your results, if you review your geotargeted accounts.
Complicating this research is the fact that geotargeting is set at campaign level, but the geographic report in AdWords can only be run at the account level. So, I checked the three clients we have whose campaigns are all geotargeted. I may use Omniture and GA to check other clients later.
Results
In one client, fully 30% of the clicks came from outside the geotargeted area. And 30% is not just a few clicks. For another client, 20% of my impressions are going outside the geotarg area and the CTR for those impressions is zero. That will lower quality score because it lowers the account CTR. I’m taking that to my rep.
| Client | Geotargeting | Outside Geotargeted Area |
Spend Outside Geotargeting |
Problem |
| #1 | NC/SC only | 20% of impressions, zero clicks (zero CTR) |
0% | Lowers account CTR and quality score, raises overall CPC |
| #2 | SC only | 51% of impressions, CTR is 1/8th those in the geotargeted areas |
13% | Lost spend on bad clicks, again quality score problem will increase overall CPC |
| #3 | 31 Eastern states | Only 0.3% of impressions and good CTR |
Minimal | None, but still disconcerting |
The first client reviewed was geotargeted only to NC and SC because it’s a medical service for which people are unlikely to travel much further. In a 2 month period, there were 669 clicks total. There was only 1 click but 10,738 impressions from other areas. The total of impressions from all areas was 51,526; so 20% of my ad impressions were shown to areas I didn’t request. Is AdWords’ insistence on showing outside relevant areas dragging down my quality score? Because 20% is substantial- but because the quality score algorithm is secret, I can’t project how much less CPC would have been with the better quality score I would have received from the more relevant impressions I requested.
Second client is a cable company that only services South Carolina. This does not and cannot serve clients outside of the state, so there are no relevant ad viewers outside SC. Out of 1.139m impressions, only 558k are in my geotargeted area. These (many in content network) get an overall 0.54% CTR. There were 581k from outside where I chose, and these only received a 0.06% CTR. Again, having 51% of my impressions come from outside my geotargeted area and getting 1/8th the CTR on those drags down my quality score and increases the CPC on the clicks from GOOD areas. There were 376 clicks from outside my geotargeted area for $1026.10 spend, which was 13% of the total spend in this time period. Google increased their earnings by 15% in this case on the anti-geotargeted impressions and by an unknown amount on my requested impressions.
The third client is a golf destination, and we targeted 31 states east of the Mississippi. This one turned out as my account rep predicted. For some reason, they showed 0.3% of my impressions in Quebec, which seems more reasonable. But my geotargeting here was much wider than in the other clients. Google may have shown impressions in these areas even if I had geotargeted only to NC. In this case, I have no big complaint, because the CTR for those Quebec clicks was good, and we know we get Canadian golfers wanting to come down to Myrtle Beach.
Discussion
Despite the third case, the larger problem remains: if I’m smart enough (or have good data from my analytics) to geotarget more specifically for better results or for specific campaign goals, I should be able to do so. Google either thinks their algorithm is smarter – and clearly it’s not – or they care more about making money than about helping me reach my advertising goals, or this is an antiquated approach they need to update.
In everyday life, I drive a stick shift because I like to control the machine. I don’t use automated bid rules or third party software to optimize AdWords. I do it by hand. I do it better and faster that way. I like the control because I know what I’m doing.
What Google is doing here, making query parsing an exception to geotargeting, is worse than turning a manual transmission into an automatic. An automatic would be fine for the convenience or if it got you better gas mileage, right? But in this case, their machine is set to maximize their KPI’s (more clicks and dollars), not mine (more relevant impressions, better CTR, better quality score, lower CPC, better CR, better cost per lead, better ROAS).
They use the rationale that query parsing delivers relevance, but the impressions and clicks are clearly not as relevant in two out of the three cases above. It’s a clever rationale, because we know relevance is the foundation of everything Google does… but it doesn’t make sense when the CTR is so dramatically lower and when they know that lower your CTR increases CPC and thus their earnings- they win either way- they get higher CPC on your relevant clicks, and they may get more clicks overall.
Generally I’m a Google supporter, but this policy goes over the line.
The advertiser loses here, and I think this is a bad approach for Google over the long term. They may be the best and in some ways the only pay per click solution right now, but if they abuse that position, advertisers will jump to the next competitor faster with less loyalty to Google.
Solution
In the meantime, the AdWords rep suggested I follow the solution in Dan’s blog. More specifically:
If you think that query parsing does not generate quality leads, then you can follow the steps outlined in the blog article and select the individual regions within the state or country that you are targeting. This would reduce the query parsing (but not eliminate it) because the more granular the targeting the less likely that someone is using that in their search term.
That is, you have to select all the metro areas, not just states. Extra work for us. And as he says, does not completely turn our AdWords system back into a stick shift. How about just a button that turns off geotargeting-oriented query parsing?
Brian Carter is the Director of Search Engine Marketing for Fuel Interactive, an interactive marketing agency in Myrtle Beach, South Carolina. He is responsible for the SEO, PPC, SMM, and ORM programs at Fuel and its partner traditional agency Brandon Advertising & PR.
