Where Google Ads Budget Actually Leaks

Most accounts do not waste money in dramatic ways. There is rarely one campaign obviously burning cash. The waste is spread thinly across settings nobody has looked at since the account was built, and it usually adds up to a meaningful share of the monthly budget before anyone notices.

This is a walkthrough of where to look, in the order that tends to find the most money fastest. It deliberately does not cover the search terms report or negative keyword lists — those deserve their own treatment and you can find both elsewhere on the blog.

Open the settings before the keywords

Nearly everyone audits keywords first. That is the wrong order. Keyword problems cost you a percentage of a campaign. Settings problems cost you the whole thing.

Open each campaign and go to Settings. Work down the page.

Networks. On a Search campaign you will find two checkboxes: include Google search partners, and include display network. Search partners covers sites and apps outside Google that show search results. Display expansion lets a search campaign spend on banner inventory. Both are ticked by default on a lot of campaigns, and neither is inherently bad — but you should be able to say what each one earned. Add the Network segment to your campaign table and look at cost and conversions per network separately. If display inventory has consumed budget and produced nothing over a meaningful period, untick it.

Locations, and specifically the location options. This is the single most common leak we see. Under the location list there is a collapsed section called Location options. The default setting reaches people who are in, regularly in, or who have shown interest in your targeted area. That last clause is the expensive one. It means somebody sitting in another country reading about your city can trigger your ad. Switch it to the presence-only option — people in or regularly in your targeted locations — unless you have a specific reason to want the wider net.

Then check what actually happened. The predefined location reports distinguish between the locations you targeted and the physical locations users were in. If those two lists disagree badly, you have found real money.

Ad schedule and devices. Segment the campaign by day of week and by hour, and separately by device. You are not looking for small differences. You are looking for a block of hours or a device that takes a visible share of spend and returns nothing. A tradesperson who never answers the phone after 7pm is buying clicks after 7pm for no reason.

Check what the conversion column is counting

A conversion number that flatters you is worse than no number, because it makes every other decision wrong.

Go to Goals > Conversions > Summary and read the list properly.

• What is actually in there? Page views of a contact page, clicks on a mail link and scroll depth are not enquiries. If any of those are set as primary conversion actions, your cost per conversion is fiction and any automated bid strategy is optimising towards fiction.
• Count setting. Each conversion action has a count setting of one or every. Lead generation usually wants one — a person who submits the same form three times is one enquiry. Ecommerce wants every.
• Double counting. Plenty of accounts import conversions from Analytics and fire a Google Ads tag for the same event. The result is every enquiry counted twice, a cost per lead that looks half what it is, and bidding that overpays accordingly.
• Primary versus secondary. Only primary actions feed bidding and appear in the Conversions column. Use secondary for the things you want visibility of but do not want to bid towards.

Fixing this changes nothing about your spend on the day you fix it. It changes every decision you make afterwards.

Structural leaks

Brand mixed in with everything else. If searches for your own company name sit in the same campaign as searches for your services, the brand terms — cheap clicks, high conversion rates, people who were coming to you anyway — will make the campaign averages look fine while the part that has to win new customers loses money. Split them. You need to see non-brand performance on its own before you can judge it.

Campaigns competing with each other. The same keyword in two campaigns does not double your chances; Google picks one. What it does do is split your data so neither campaign accumulates enough signal to bid well, and it makes reporting harder to read. Consolidate.

Campaigns permanently limited by budget. The status column will say so. This is not always waste, but it means Google is choosing which of your auctions to enter and you are not. If a campaign has been capped for months, either raise the budget or cut the targeting so the money goes somewhere you chose.

Broad match on manual bidding. Broad match hands query selection to Google’s models. Those models steer using conversion data. On a campaign with manual CPC or a clicks-based strategy and little conversion history, broad match is close to unmanaged spend. Either give it the conversion data and the bid strategy it needs, or use tighter match types.

The recommendations you did not make

Open the Recommendations page and then find the auto-apply settings behind it. Google can be configured to apply changes to your account on your behalf, including adding keywords and broadening match types. Several of these are on by default in newer accounts.

There is nothing sinister here, but “apply automatically” and “apply after we have checked it” are very different arrangements, and account owners are often surprised to find which one they are on. Turn off anything that changes targeting or match type without review. Leave the recommendations themselves — they are useful as a prompt, just not as an autopilot.

While you are there, ignore the optimisation score. It measures how much of Google’s advice you have taken, not how well your account performs.

What to do with the list

By the end of an audit like this you will have a page of findings of very different sizes. Order them by how much money each one touches, not by how easy each one is.

In practice the order usually comes out as: fix conversion tracking, fix location settings, split brand from non-brand, review the networks, then everything else. The first two often take an hour between them.

Then leave it alone for a few weeks. Changing eight things at once means you learn nothing about which one worked, and every bid strategy you touched restarts its learning in the meantime.

When the leak is not in the account

Sometimes the account is fine and the money still disappears. That usually means one of three things: you are buying searches from people who are researching rather than buying, your landing page loses people the ad has already paid for, or enquiries arrive and nobody follows them up quickly enough.

None of those are fixed in the campaign settings, and an audit that only looks inside Google Ads will miss all three. If your click-through rate is respectable and your conversion rate is not, the problem has moved to the website and that is where the next hour should go.

If you would rather have someone else do the reading, that is what a paid search audit is for — and a decent one tells you when to spend less, not just where to spend more.