Google Ads for online shops: Where the budget disappears — and how to get it back

If an online shop isn't profitable with Google Ads, it's rarely due to the budget. It's almost always because part of the budget is being wasted on things that don't affect the account balance: clicks from people searching for something else; ads for products no one buys; campaigns whose success can no longer be measured because the data connection is lost along the way. None of this appears as a line item in the report. It only shows up in the results.

This article reveals five places where money regularly disappears from online shops—and what you can do about it. All five can be checked without an agency; most can be done in an afternoon. The result is an account check that you can repeat in your calendar.

To clarify: This isn't about the quality of your product data—we'll be publishing a separate article about product data feeds this week. Our article on consent, tracking, and measurability explains how to set up tracking and consent in a technically sound way . This article focuses on the account itself: its structure, management, and what you're actually paying for.

The search term report is the most honest report in the account.

The most important distinction in Google Ads is also the one most frequently overlooked: A keyword is what you bid on. A search term is what someone actually types. The space between these two lies in the range where Google decides whether your ad is shown—and it's within this range that most of the wasted ad spend occurs.

The search term report makes this flexibility visible. Google describes it as follows: It shows "the performance of your ads when delivery was triggered by search queries on the Search Network." Google itself provides an example: Someone searching for "red roses" will see your ad if you have booked the keyword "roses."

For a shop, this is where the truth lies. This is where you find the search queries you're paying for unintentionally: people looking for repair instructions instead of a spare part; people who mean your competitor; people who include "used," "DIY," "reviews," or a brand name you don't even carry. Each of these clicks costs the same as a click from someone with the intention to buy.

The process is straightforward and effective: Open the search term report, set the time period to the last 30 to 90 days, sort by cost , and review from top to bottom. Anything that cost money and didn't generate any revenue is evaluated: Is it a valid query for which only the ad or landing page wasn't a good fit? Or is it simply the wrong intent? In the latter case, the term should be excluded. In practice, a recurring 20-minute appointment is sufficient—initially weekly, later every two weeks.

Keywords to exclude: the trap that hardly anyone knows about

Negative keywords are the counterpart to booking: As Google writes, they allow you to "exclude search terms from campaigns" in order to "focus on the keywords that are relevant to your customers." So far, so good. The unknown half costs money.

The help section for negative keywords contains a sentence you should write down: "For negative keywords, no matches with similar variants are detected." In other words: While Google is generous with variations, typos, and singular and plural forms of your paid keywords, the opposite is true for negative keywords—they are letter-by-letter exact. Excluding "shoes" doesn't automatically block "shoe." Excluding "repair" doesn't block "repair." And a typo in the negative only excludes the typo itself.

This explains a pattern we repeatedly see in inherited accounts: A well-maintained exclusion list exists, yet the same unwanted requests continue to appear—albeit with slightly different spelling. The exclusion wasn't incorrect; it was simply too narrowly defined. Those who understand this rule create exclusions in all necessary variations: singular and plural forms, the most common typos, and compound and separate spellings.

Two further limitations are also listed there and are good to know. For very long search queries, the exclusion isn't unlimited: If your excluded term appears beyond the sixteenth word, the ad may still appear. And for display and video campaigns, there's a limit of 1.000 excluded keywords per account—anyone trying to work with endless lists will hit a wall. Search campaigns offer significantly more flexibility, but even there, a list that no one can keep track of isn't a control mechanism, but rather an archive.

Practical advice from working with shop accounts: Maintain two lists. An account-wide list of things that are fundamentally unsuitable for your business—"free," "used," "instructions," "jobs," "returns," and the names of competitors you don't carry. And for each campaign, a more focused list that refines your intent. Update the account-wide list only once and infrequently; the campaign lists grow organically from the search term report.

Performance Max: You control the input, not the display.

Few campaign types have caused as much unrest as Performance Max. The reason lies in a gap in expectations: Those accustomed to search campaigns want to control keywords, ads, and placements. Performance Max doesn't offer this—and that's not a flaw, but its very design.

According to Google Ads Help, such a campaign runs across all channels simultaneously: "YouTube, the Display Network, Google Search, Discover, Gmail, and Maps." It is controlled via objectives and asset groups —which Google defines as "collections of creatives targeted at a specific theme or audience." Audience signals further guide the automated process.

E-Commerce News - Tips & Tricks Consent, tracking and measurability without data chaos in the online shop

This shifts the focus: instead of adjusting bids and keywords, you decide on the input . And this is precisely where shop budgets disappear. Three points are worth considering. First, the objectives. If a campaign is optimized for "conversions" and everything counts as a conversion—newsletter signup, contact form, purchase—then the automatic optimization reliably prioritizes the cheapest approach. For a shop, the only meaningful metric is revenue, not the number of events. Second, the asset groups. A single group for a mixed product range leads to Google combining images, text, and products that don't belong together. Splitting groups by product category is the most effective adjustment available in Performance Max. Third, the assets themselves. Google explicitly recommends providing enough creative assets so that the campaign can be delivered on all platforms. Anyone who only uploads a landscape image simply won't appear on the platforms that require portrait or square formats.

And a fourth, inconvenient point: Performance Max also includes search queries that a search campaign could serve—including queries for your own brand name. Anyone running both campaigns simultaneously should know which campaign is generating which demand, instead of wondering about fluctuating numbers.

This leads to the crucial question that determines the results of any shop account: How do you allocate your budget between campaign types? There's no one-size-fits-all answer, but there is a reliable order. First, you serve existing demand —people who are already searching for your product or brand. This is the most cost-effective demand in the account and belongs in clearly defined search campaigns where you can see the results. Only then do you expand demand using Performance Max, which reaches additional people who aren't yet actively searching. Reversing this order and allocating most of your budget to automated campaigns while your search campaigns are maxed out at their daily limit means you're paying for reach before you've covered your guaranteed demand.

In practical terms, this means: First, check if your search campaigns are limited by your budget—Google displays this information in your account. As long as there's unused demand on the table every day, every additional euro spent on Performance Max is a costly mistake. This is precisely the first step we take when managing shop accounts , before adjusting anything else.

Your own brand: the most expensive dispute in your account

Few questions are answered incorrectly in shop accounts as reliably as this one: Should you bid on your own brand name? The honest answer is: It depends — and the costs of the wrong decision are real in both directions.

Bidding on your own brand when no one else is advertising there and your shop is already ranked number one means buying clicks you could have gotten for free. Conversely, not bidding when competitors or marketplaces are advertising your brand name cedes the top spot to them, giving it to the very people who are actively searching for it—the most valuable demand of all.

The decision can be measured, not debated: Search for your own brand name in an incognito window and see who is advertising above your listing. If no one is listed, brand advertising is usually unnecessary. If someone is, it's a defensive measure. And if you do use it, it belongs in a separate campaign with its own budget—otherwise, the reports will mix the cheap brand clicks with the expensive customer acquisition efforts, and the overall result will look better than it actually is. This mixing of results is the most common reason why an account works on paper but not in practice.

No consent, no data — and no data, no optimization

E-news for online retailers - read your way to wealth Product data feed before the season: Google Shopping, marketplaces and the typical sources of error

The last point is the most invisible, yet it affects everything else. Modern Google Ads campaigns are largely driven by conversion data: what is measured is optimized. What isn't measured doesn't exist for the automated system.

And in Germany, data collection is subject to a clear condition. Section 25 of the TDDDG (Telecommunications Data Protection Act) states it unequivocally: “The storage of information in the end-user's terminal equipment or access to information already stored in the terminal equipment is only permitted if the end-user has given their consent on the basis of clear and comprehensive information.” The two exceptions in paragraph 2 are narrow: the mere transmission of a message—and what is “absolutely necessary” for an explicitly requested service to be provided at all. Advertising and analytics purposes are not included.

In practical terms, this means that some of your purchases will never appear in your account because consent is lacking. This is legally correct—but it's a reason to know your own numbers. If Google Ads shows significantly fewer orders than your shop's backend, this isn't initially a campaign problem, but rather a measurement gap. Those who aren't aware of this gap will throttle campaigns that are actually performing well.

Three things are therefore worth checking before adjusting the budget: Is the consent signal actually reaching the advertising tags, or is everything simply being blocked? Is no advertising script actually loading before consent is given—checked in an incognito window, and not accepted? And what is the gap between orders in the shop and conversions in the account? This figure represents your measurement gap, and it should be included in every report you read. We have described separately how to set this up technically soundly without ending up with data chaos.

The account check: 90 minutes, six glances

The following six steps have proven to uncover the majority of wasted resources. First: Sort the search term report for the last 90 days by cost and review the fifty most expensive lines without revenue. Second: Create exclusions for the resulting variations—single and plural, common typos, compound and separate words, because similar variations are not automatically recognized here. Third: Review what counts as a conversion in your account and deselect anything that isn't revenue; for online stores, the revenue value should be the target metric, not the number of events.

Fourth: Compare asset groups in Performance Max against your product range—one group per category, not one for everything—and check within each group whether images are stored in all required formats. Fifth: Search for your own brand name in an incognito window and decide whether brand advertising is defensive or self-deception; if defensive, then use it in a separate campaign. Sixth: Compare orders in the shop backend against conversions in the account and note the difference—this is your measurement gap, and it explains more than any bid adjustment.

Anyone who reviews these six points once a quarter will need significantly fewer interventions in between. And they will recognize the difference between a campaign that doesn't work and one that simply isn't being measured correctly.

Conclusion: Not more budget, but less waste.

The five features mentioned in this article have one thing in common: they don't cost extra money; they actually give you a return. The keyword report shows you exactly what you're paying for. Exclusions only work if you know they're letter-by-letter accurate. Performance Max is driven by user input, not bids. Brand advertising is either defensive or an expensive knee-jerk reaction. And without reliable measurement, even the best automated optimization is useless.

If you take one thing away from this post: Open your search term report this week, sort by cost, and read the first fifty lines. That half hour is the cheapest optimization you can get in Google Ads—and it will tell you more about your account than any dashboard. If you'd prefer to go through this with someone who has seen many shop accounts, you can find the framework for that on our Google Ads support page.

Where does your budget disappear to?

Which search query surprised you the most when you last looked at the search term report? We are particularly interested in cases where a single excluded term measurably changed something.

And to everyone using Performance Max: Did splitting your asset groups benefit you—or was it a waste of time and effort? Let us know in the comments. For specific account inquiries, please contact us via the contact page.