How to tell if you are getting screwed on ad spend

Every ad account eventually asks the same uncomfortable question: is this money working, or is it just leaving? There are exactly two ways the answer comes back "leaving," and they are worth telling apart because the fix is different for each. The first is a person, an agency or freelancer, who is not actually doing the work you are paying for. The second is a platform's own automation, Performance Max, Advantage+, automated bidding, auto-apply recommendations, running with nobody checking its results. Both leave real fingerprints. Both are checkable in an afternoon, without needing to be a PPC expert. This is that checklist.

Leak one: branded search taking credit for demand you already had

Branded search is anyone typing your company name, or something close to it, directly into Google. Those searchers already know you. They found you some other way, word of mouth, a previous ad, an article, a friend's recommendation, and they were coming to you regardless of whether an ad showed up. If your account is bidding on your own brand name without a clear reason, or if a Performance Max campaign has no brand exclusion applied, you are very likely paying for conversions that would have happened for free.

This is not automatically a scam. Some businesses have a legitimate reason to protect branded search from a competitor bidding on their name. The problem is when nobody can tell you which situation you are in, because nobody has looked. Pull the "search terms insights" report inside Performance Max, or a standard search terms report on a Search campaign, and see what share of matched queries look like your own brand and close variants. A high share, paired with a real cost per click on those terms, is money that likely would have converted for a fraction of the price, or for free through organic search.

Leak two: low-intent display and video spend riding along with real intent

Both Performance Max and Advantage+ are built to spread a single budget across many kinds of inventory at once: Search alongside Display, YouTube, Discover, Gmail, and Maps on Google; Feed alongside Reels, Stories, and the Audience Network on Meta. The bidding model does not inherently prefer Search or Feed just because those tend to convert better. It goes wherever it can win an impression at a price that fits the target it was given, and cheap, low-intent placements are often easier to win.

The result, left unchecked, is a budget that quietly drifts toward video views and display impressions with a real cost attached and a much lower actual return, while the top-line "conversions" number still looks fine because it is being propped up by the smaller Search or Feed share doing the real work. The fix is not to distrust these campaign types. It is to open the channel or placement breakdown at least once a month and look at the split directly, instead of trusting the blended total.

Leak three: asset groups and ad sets nobody has touched in months

Stale creative does not throw an error. It does not trigger an alert. It just quietly underperforms inside a system built to average results across everything running at once, which makes a slow decline easy to miss for months. An asset group built around last year's pricing, a discontinued product, or messaging that no longer matches your current offer will keep serving exactly as configured until someone opens it and notices.

Check the last-modified date on your active asset groups and ad sets. If the answer is "I don't know" or "a long time ago," that is the leak, independent of whatever the performance numbers currently show. A campaign type built for automation still needs a person feeding it current material.

Leak four: conversion tracking that is quietly lying to the algorithm

Automated bidding, in Performance Max, Advantage+, or a standard automated bid strategy, optimizes for exactly one thing: whatever event your tracking setup tells it counts as a conversion. If that setup is broken, duplicated, or measuring the wrong action, every automated decision built on top of it inherits the mistake at scale. A pixel that fires twice on a confirmation page, a lead form that counts a page view as a submitted lead, a cross-domain checkout that drops the session partway through: all of these produce a conversion count the algorithm treats as ground truth, and none of them show up as an error anywhere in the interface.

The only real check is comparing platform-reported conversions against your actual CRM or order system, at least once a quarter, ideally monthly for a high-spend account. A gap between what the platform reports and what your business actually closed is not a rounding error. It is a sign that every automated decision for the period in question was optimizing for the wrong number.

Leak five: an agency, or a person, who is logging in instead of managing

The human version of this problem looks different but produces the same result. A monthly report with a chart and a few green arrows is the easiest deliverable in this business to produce and the least useful one for judging whether anyone actually looked at the account. Most of what fills that report now comes straight out of the platform automatically. It does not tell you whether a person reviewed the channel breakdown, checked the change history, or caught a stale asset group before it cost real money for a quarter.

Ask for the account change history directly instead of a summary of it, and ask for a plain-language explanation of one specific decision made in the last month: which lever, why that one, what was expected, what actually happened. An agency doing the work will have both answers ready. One that is mostly watching automation run will point back to the report.

A 30-minute audit you can run this week

1. Pull the channel or placement breakdown for your largest automated campaign and note the Search-or-Feed share versus everything else.
2. Check whether a brand exclusion is applied, and if not, check what share of matched queries look like your own brand name.
3. Check the last-modified date on your active asset groups or ad sets.
4. Compare last month's platform-reported conversions against your actual CRM or order count.
5. Open the account change history and see what a human being actually did in the last 30 days.

None of these five checks require deep platform expertise. They require twenty to thirty minutes and a willingness to open reports that are easy to ignore when the top-line number looks fine. Do them once, and you will know within the hour whether your money is leaking from unsupervised automation, an unsupervised person, or neither.

If you have run through this checklist and found something you do not know how to fix, or would rather have someone run it against your account and act on what they find, that is exactly what The Discoverability Company's advertising management service is built for: hands-on Google Ads and Meta management, account audits, conversion-tracking cleanup, and turning what the reports show into an actual decision. Pricing is public, $350 a month or 10% of ad spend, whichever is greater, no setup fee, no contract minimum. Details on the advertising management page.