# Why Does AdWords Go Overbudget?

Anyone who has administered an Google Ads account for any length of time probably wonders why the account’s spend never quite matches what you’ve set for a limit. Here’s a little look into why Google goes over your budget occasionally.

How Many Days Per Month?
One thing that will impact your monthly account spend is understanding how a budget is divided.  Say you have \$1,500 to spend on Google Ads per month and you to calculate your daily budget by dividing \$1,500 by 31.  For a month like October with 31 days that’d be just fine, but in a month like February your account won’t hit the monthly spend because there are only 28 days.  There are two options to deal with this:

1. Recalculate your daily budget every month based on the number of days in the month.  Monthly budget ÷ number of days = daily budget
2. Enter a daily spend based on the average number of days per month.  This strategy will average out your spend over the year and hit your total available budget for the year.

Calculating Average Number of Days Per Month
Finding the average days per month isn’t hard. 365 days per year ÷ 12 months = 30.42 days per month.  Go ahead and use 30.4.  Divide your monthly target budget by 30.4, enter the result in Google as your daily limit and let it ride.  On months with 28, 29, or 30 days you won’t spend your entire monthly budget, and on months with 31 days you’ll go slightly over.  It’s ok.  If your monthly budget is determined by a yearly max budget, you’ll be just fine.

Remember, Google Will Go Over Budget & There’s Nothing You Can Do About It!
Google will go over budget, sometimes as much as 20%!  If your yearly Google Ads budget is set-in-stone you’ll need to check year to date (YTD) numbers periodically to make sure you don’t bust the budget before December.  I recommend auditing them quarterly to make sure they are on budget.  If you have popular keywords and it seems like you’re going over budget fairly often, you probably need to audit monthly, and possibly reduce your daily budget.  In this case I’d calculate how much you’re going over and then reduce the budget by that amount. During you monthly or quarterly audit, calculate how much has been spent year-to-date, subtract that from your total budget and find out how much is left.  Then figure out how many more months (or days if you want to be precise) are remaining in the year, and adjust your daily spend accordingly.  If you’re right on track in December, it’s best to be safe and reduce December’s daily spend by 10-20% to make sure you don’t accidentally go over.

In the example above, the cookie jar represents your Google Ads account, and the people taking cookies are Google’s distributed servers.  The servers are decentralized and when things get going really fast it’s hard for them to keep up with what each other are doing.  One server sees the number of cookies dwindling and takes two cookies because it thinks they are available because it doesn’t realize that there are three other servers waiting to take cookies.  It’s not a perfect way of explaining the concept, but it gets the point across.

Glad you asked.  Google had real growing pains in developing a highly-available, extremely stable database for Google Ads.  Because the database can’t be locked every time a transaction happens they use a management method called optimistic locking.  Google’s latest database is named F1 and in a paper Google describes it as:

F1 is a hybrid, combining the best aspects of traditional relational databases and scalable NoSQL systems like Bigtable. F1 is built on top of Spanner, which provides extremely scalable data storage, synchronous replication, and strong consistency and ordering properties. F1 inherits those features from Spanner and adds several more [including] Optimistic transactions.

I made “optimistic transactions” bold because when it comes down to it, that’s why Google goes over budget.  (Detailed Wiki Explanation)  This same kind of database management is used by airlines and is why they sometimes oversell flights.  There is no way a pessimistic database could handle all the requests from all the travel agents, airline employees, and websites if every record had to be locked when someone was looking at a particular flight.  Under pessimistic locking, when one person is booking a ticket on a plane, nobody else in the world could book a seat on that flight at that moment (heck, they couldn’t really even check how many seats were available!). Overselling does happen and it does cost airlines money, but nowhere near as much as it would if they could not book multiple tickets at the same time.  If your flight is oversold it’s because a lot of people bought tickets really quickly for whatever reason.

Why Are They So Optimistic When It Costs You Money?