6K members online now
6K members online now
Use AdWords conversion tracking and reporting to measure your results. Have a question about Google Analytics? Post it here, on the Google Analytics Community!
Guide Me

Why does removing state targeting reduce city targeting

Follower ✭ ✭ ☆
# 1
Follower ✭ ✭ ☆

I have a very old Adwords account with data that goes back to 2007 and we had another account that went even further back. I am Adwords certified.  Just trying to eliminate simple answers. 


I have always targeted the state of Texas. I added city targeting when they added the ability to see stats based on cities and let you adjust bids at the city level. 


Last month I made some changes that increased my CTR and conversion rate dramatically. The problem is that it increased my spend dramatically. I looked in my DB and noticed that most of my clients came from Dallas, Houston and Austin. Yesterday I went in and removed Texas and all cities except Dallas, Houston, and Austin. 


I came in the morning and my impressions went down 88% and I did not make my change until several hours into the day so it is actually much more than 88%.  If I were to go back to last Monday and look at my campaign settings under location and subtract all non Houston, Dallas, Austin traffic from the total I would have reduced my impressions by 35%.


If it was just a little off I would chalk it up to normal up and down but this is beyond normal up and down. What this tells me is that Google knows what city somebody is in for reporting but not for showing ads. The system that reports how much traffic I got for a city is not the same system that decides where my ad should be shown. Why can't they use the same system for both? Is it because it is hard to get that data quickly when an ad is shown. It takes 2 days to get the numbers in my report so I wonder if they use that time to pinpoint location more accurately. 


The system is broken and not working as advertised or is working as planned to force me to show my add for all of TX to get traffic in those 3 cities.  I'm pretty sure I can work this out with negative locations but I hate that I have to do a complicated work around to fix this problem.

Re: Why does removing state targeting reduce city targeting

Participant ✭ ✭ ☆
# 2
Participant ✭ ✭ ☆

Hi TX-ppcguy,


Thanks for sharing your story, this is really interesting. I have a few questions for more context.


My first questions are:
-- You mentioned most of your clients were from Houston/Dallas/Austin but have you looked at the dimensions report to see if that's where your AdWords conversions were coming from?

-- It sounds like you calculated the traffic you should have lost based upon the size of the cities you were targeting but your traffic volume wasn't necessarily proportional to city size.  Did you check out the dimensions report to see which locations most of your visitors were coming from?


If the answers to both questions above are yes, then I have a few more questions:

-- Can you share more details about the optimizations that you made to improve CTR and Conversion Rate? 

-- Are you targeting people "in my location" or "people in and searching for my location"? Same with exclusions, are you targeting "people in my excluded locations" or "people in and searching for my excluded locations"?
-- Do You have any exclusions?
-- Were any of the cities that you removed from targeting near the Dallas, Houston and Austin areas?
-- Have you looked at search queries to see if many queries mentioned locations?


What I'm thinking is that if your traffic volume decreased as a result of the geo-targeting change it could be because your Dallas, Houston, and Austin visitors were searching for keywords that may have shown intent toward one of your now-removed cities, or their location could have been misinterpreted by Google as having been in one of the now-removed locations.


OR people in the now-removed locations could've been searching with queries geo-modified for Houston/Dallas/Austin.


I'm really interested to hear more detail! Hope this helps to get down to the bottom of it!



Re: Why does removing state targeting reduce city targeting

Follower ✭ ✭ ☆
# 3
Follower ✭ ✭ ☆
Wow that is a lot of questions. Let me know if I leave something out.

1. No Exclusions
2. "in my location" and "people in and searching for my location
3. Some of the cities were close like I removed San Antonio and kept Austin.
4. I use the Nielsen DMA regions when they exist
5. I have not looked at Dim tab for this

I have changed the account now to show the 3 cities and all of Texas and then excluded out everything not in those 3 cities. I used Nielsen DMA regions and congresional distrcts to exclude large rural areas.

My current conclusion is that Google really does not know where most people are at search time. They figure this out later for your reports. It will be interesting to see what happens with my new test. The reason I say this is removing Texas should not affect the number from Dallas & Houston. Why would adding Texas reaise my impressions in Dallas & Houston? LIke I said above I have "searching for my location" turned on so people outside of Texas should be able to find my ad if they type in terms that include Houston.

One question I have is does my bid modifer for Houston really affect all the Houston traffic?

As far as what I did to improve CTR the main thing I did was create mobile ads with less text that fit on cell phones. I made shorter site links as well. I also increased mobile bids and kept on top of my average position. I wanted to keep them around 1.2 because below 3 does not get shown or seen much because of how mobile phones show ads. I also wrote a script that looks at my keywords bids and moves them up and down basesd on average position. I also stayed on top of the negative keywords. I had not worked on this account for some time because I was working at another company. Even though I have looked at this account on and off for years this is the first time I am working on it full time.

Re: Why does removing state targeting reduce city targeting

Visitor ✭ ✭ ✭
# 4
Visitor ✭ ✭ ✭
Hi PPCguy,

Thanks for the info, I know I asked for a lot. My next question is how are all of your campaigns set up? Are they set up at the city level?

The reason I ask is because I'm wondering how you know that traffic decreased in those specific cities without looking at the dimensions tab? What reporting are you using to determine the traffic change?

Thanks again for all of the detail!

Re: Why does removing state targeting reduce city targeting

Follower ✭ ✭ ☆
# 5
Follower ✭ ✭ ☆
All Traffic went down 90% I have two campaigns one includes geo terms and one that does not. An example would be "texas keyword keyword" or "Houston keyword keyword" and the other campaign would be just keywords with zero geo terms at all.

The vast majority of my terms include a single root and just different ways of saying the same thing. My root term is a type of law.

We were spending $13,000 a month and my CTR improvements made it $20,000 a month. My total spend went went down 88% and I was comparing two Mondays. There has never been a deviation between two consecutive Mondays other than holidays in the history of the account. I had Texas targeted and several cities individually targeted with no exclusions.

I remove everything other than 3 cities which have traditionally shown 65% of my impressions. The next day all of my traffic went down 88% instead of 35%. I would not have thought much about it other than the huge drop. I have talked to Google support and they have no idea what happened. I wish I could talk to somebody higher up but was not offered a chance when I asked.

I will be able to prove one way or another next week after I get more data. I am doing several tests.

Re: Why does removing state targeting reduce city targeting

Participant ✭ ✭ ☆
# 6
Participant ✭ ✭ ☆
Hey PPCguy,

I think you need to take a look at your dimensions report and see where your visitors were coming from before you updated targeting. The decrease in traffic may have been coming from other locations. It would make sense that you would lose 90% of traffic when you remove the state of Texas in favor of 3 cities. The most likely thing that has happened was that you were getting a lot of visits outside of the cities you are now targeting. The dimensions report will tell you for sure.

If the dimensions report confirms that you did lose traffic from those three cities, it could still be that you had people in those cities who Google has misinterpreted their location - this can often happen. My last company was based out of Toronto and when I was connected to the network, it often looked like I was in Toronto. Networking is a little over my head but I do know there are a lot of ways that someone can appear to be in a different location than they are. There's also the occasional traveler with their location preferences set to their home city.

Last but not least and certainly not unlikely is that people in the three cities you are targeting were searching for one of the locations that are now removed OR people from now-removed locations were searching with a term that indicated that they were interested in the cities that you are targeting.

Hope this helps!

Re: Why does removing state targeting reduce city targeting

Follower ✭ ✭ ☆
# 7
Follower ✭ ✭ ☆
I still don't think you understand. Even if what you say is true that proves that the stats in the location tab are junk. Today is the first day that I have access to that data since it takes 2 days to get. Here is what I found for last Oct 27.

Here are the stats you asked for.

1. 2.7% of my traffic that day was from location of interest
2. Houston was 24% of my impressions
3. Dallas was 23% of my impressions
4. Austin was 18% of my impressions
5. blank was 14% of my impressions

These are the exact number I got from the stats on the location tab of my campaign. Simply removing all my targeting other than those 3 cities should have only reduced my spend by 35% not 88%.

That is too big of a drop to be explained by any other conclusion that Google does not know where most people are when an ad is displayed. This actually makes sense because it would take forever for a search result to load if Google used all the ways it knows to find out where someone is. They store the raw data and spend two days crunching that data to show us the actual location.

The conclusion is that there is no way to accurately target a specific location or exclude a location for tha tmatter. I can further prove this by adding TX back and excluding all but my 3 cities and traffic does not change much. I still believe that Google only know where ~10% of people are at the metro city level during search time. I would not be suprised if I were to add the US my Houston and Dallas numbers would go up even more. I think that Google has a better idea where people are when the Dim reports are generated but not when an ad is served.

Re: Why does removing state targeting reduce city targeting

Participant ✭ ✭ ☆
# 8
Participant ✭ ✭ ☆
Hi PPCguy,

This is helpful. May I ask a few more questions?

The stats your provided above, are they for Oct 27, 2014 or 2013?
When did you make the adjustment to your location targeting?
Could you share a table that shows what locations your visitors were coming from the week before you changed targeting and then also the week after?

I have to say, I have worked with clients extensively to create local campaigns with a lot of success. Location targeting isn't perfect for all of the reasons I mentioned above but those aren't typically due to Google's errors. I don't think the drop you are seeing is due to lack of geo-data.

Re: Why does removing state targeting reduce city targeting

[ Edited ]
Follower ✭ ✭ ☆
# 9
Follower ✭ ✭ ☆

It was 2014. I made the changes early on Monday Nov 3 in the morning. I only have one day of the giant drop and that was Nov 3. I added TX back to as soon as I saw the problem on Nov 4th in the morning. 

I'm sure that the location targeting accuracy is very dependent on your audience. If your going after people that are more tech savoy or higher income your going to have more accurate location targeting. My audience is low income blue collar workers. Also about 40% mobile.

I made this list with a pivot table in Excel. These are metro regions not city.
Houston TX 24%
Dallas-Ft. Worth TX 23%
Austin TX 18%
(blank) 14%
San Antonio TX 6%
Waco-Temple-Bryan TX 3%
Harlingen-Weslaco-Brownsville-McAllen TX 1%
El Paso TX 1%
Lubbock TX 1%
Tyler-Longview(Lufkin & Nacogdoches) TX 1%
Odessa-Midland TX 1%
Corpus Christi TX 1%
Abilene-Sweetwater TX 1%
Amarillo TX <1%
Beaumont-Port Arthur TX <1%
San Angelo TX <1%
Laredo TX <1%
Victoria TX <1%

My conclusion may not be correct for many advertisers but for my particular situation it is true. Google does not know where my particular audience is located. The only way you can prove me wrong is to take your local campaign that is targeted at the metro city level and add your entire state. I promise you that your city stats will skyrocket as well as your entire budget. If you target just a city or even a metro area you are missing out on a ton of traffic in that city.

Another way to prove this is to add your entire state then exclude everything other than where you want your ads and your impressions will increase by a very large percentage. If Google can't exclude people what makes you think they can include them.

Re: Why does removing state targeting reduce city targeting

Follower ✭ ✭ ☆
# 10
Follower ✭ ✭ ☆
Well it appears this was a known problem and was talked about at SMX. Opt out seems to be working better for me than opt in. I will need another week to know for sure.