Doubling budget = 1/4 impressions and clicks

[kevinfishburne]

I've been using AdWords since 12/01/07 with great results and have been very pleased. I noticed that when modifying ads, adding new keywords, etc., that Google gets crazy for about 24 hours and the stats are terrible. After 24 hours they build back up to normal levels. This time I'm stumped, and I don't think it is a coincidence with an external event (competitor's higher bidding, etc.), but rather a problem with AdWords.

I use 27 keywords and one ad group of six ads, all selling different desktop PC models. All link to the same landing page, but jump to a different HTML anchor for each specific model PC. That way they see the specific PC in the ad, but don't have to click anything else to see the other models.

At 11:45 PM on 03/24/08 I doubled my monthly budget from $200 to $400. I examine statistics with one week granularity, and wanted to establish a one-week baseline for budget doubling versus my old $200 budget. If there results were acceptable I'd leave the $400 budget in place and continue to refine other changes on a weekly basis. Below are the results, before and after the budget doubling. Keep in mind, -no- changes were made other than the monthly budget and campaign-wide max CPC. The max CPC was always higher than I've ever actually used for my keywords ($0.90 - $13.33).

February 1 - 29

$200 Monthly Budget
$6.66 Daily Budget
$6.66 Max CPC
Budget Optimizer On

1,478 Clicks (60/day)
1,905,792 Impressions (65,717/day)
0.07% CTR
$0.13 CPC
$190.52 Cost ($6.57/day)

March 1 - 24

$200 Monthly Budget
$6.66 Daily Budget
$6.66 Max CPC
Budget Optimizer On

996 Clicks (42/day)
1,154,607 Impressions (48,109/day)
0.08% CTR
$0.16 CPC
$158.46 Cost ($6.60/day)

March 25 - 26.5 (3:00 PM - 3 Hours)

$400 Monthly Budget
$13.33 Daily Budget
$13.33/$0.90 Max CPC
Budget Optimizer On

19 Clicks (13/day)
27,055 Impressions (18,037/day)
0.07% CTR
$0.15 CPC
$2.87 Cost ($1.91/day)

On 03/14/08 I added keywords 26 and 27, which may explain the differences between stats set 1 and 2. For the last set of stats, the results are based on 1.5 days for today and yesterday. They were taken at 3:00 PM, and I calculated the daily results by counting today as 0.5 days since Google could have been 3 hours behind in reporting. There are two values for max CPC because I first used $13.33, then changed it to the recommended $0.90. Either way no keywords have ever been higher than $0.20 in practice.

I understand that I can manually set my CPC, remove some keywords since there are so many (27), etc. I've read a million forum posts, watched the Black Book DVD, and all that. What I'm wondering is:

Why would I get less impressions and clicks after doubling the budget of my already-successful campaign?

Should I continue to wait things out to see if it changes?

Is this typical of budget changes?

What are the odds this is actually the result of insanely-coincidental external factors?

Thanks so much.

[edited by: engine at 9:46 pm (utc) on Mar. 26, 2008]
[edit reason] No sigs, thanks, see TOS [/edit]

[vol7ron]

I would say that I'm curious to see how this turns out. As with all lines of business, there are external factors that drive web browsing. This could be a result of weather, new technology, other more interesting web topics, tax returns, x-mas money, easter gifts, etc.

Because there are changes in browsing trends I would say that 1-2 days is not enough time to really consider the stats. Especially when comparing to a month long average. You should evaluate some web-traffic logs from internet statistic sites to see what part of the day most people browse the web for your topic. If you've taken statistics courses, you might want to do some t-tests to compare your proportions and means or possibly even some ANOVA testing to compare the means. This would be the best way to see if your data is significant. My guess is that it is not.

At the very least you should evaluate this after a week. This would include every day of the week and weekend, however I would assume that a week and a half to two weeks would be your best bet for comparisons. When you do this, could you please post your findings -- I'm curious to see the results.

[kevinfishburne]

After a second day of dismal results, it looks like two things were occurring:

First it appears that the budget optimizer was not automatically increasing my bids to my maximum CPC set in my campaign settings. It was keeping it at the old value, which interestingly had also been being enforced improperly. The old value was $6.66, but the actual max CPC being allowed by the budget optimizer was around $0.25. When I changed my budget from $200 to $400, I changed the max CPC from $6.66 to $13.33. The budget optimizer ignored this. The problem with lack of impresssions was caused by the max CPC bid being set too low by the budget optimizer.

Second it appears that for whatever reason the minimum bids necessary for many of my keywords to get my ad on page one have increased significantly. Since my max CPC had stayed the same I was getting less impressions despite the higher budget.

These two things combined practically nuked my daily impressions, and Google's error in not properly adjusting my max CPC also resulted in my daily budget not being met.

To resolve the issue I set my max CPC to $10. I don't think the specific number made the difference, but just changing the value (for the third time) seemed to snap Google out of its error and allow it to start raising the keyword bids. They're now automatically set around $0.40. The change in results was almost immediate.

It is only the second day of semi-acceptable results and the fourth day of weekly statistic collecting, so Tuesday morning I'll have more complete results for the week. I'll post again with those results. Thanks everyone.