When an account is burning money, the owner usually assumes someone set the bids wrong. In the accounts I have audited, that is almost never the cause. The bids are a symptom. The money is leaking through decisions made when the account was built, and it will keep leaking no matter how many hours someone spends adjusting bid adjustments and ad schedules on top of it.
The difference between structural and tactical waste
Tactical waste is a bad bid, a weak headline, a day of the week that underperforms. It costs you a few percent and it is self-correcting once someone looks at the data. Smart Bidding handles most of it now.
Structural waste is different. It is waste that the system cannot see, because the structure tells the system that the waste is success. A campaign optimizing toward a conversion action that counts newsletter signups will get very good at buying newsletter signups. Nothing in the interface will flag this. The graphs will go up. The bank account will not.
That is the defining feature of structural waste: it looks like performance. Which is why it survives for years.
Where the money actually leaks
These are the patterns I see over and over, roughly in order of how much they cost.
- The wrong conversion action is set as primary. Page views, PDF downloads, phone number clicks that nobody follows through on, form loads instead of form submissions. Everything downstream of this is corrupted, including every automated bidding decision.
- One campaign carrying products or services with wildly different margins. The algorithm optimizes for conversion count or a blended return. It will happily buy fifty low-margin sales instead of ten high-margin ones and report a great month.
- Geography that does not match where you can deliver. The default location setting has historically included people merely showing interest in your area. For a local service business that quietly buys clicks from people who will never be customers.
- Brand and non-brand mixed together. Brand searches convert cheaply because those people already decided. Blended into the same campaign, they make cold traffic look profitable and hide the fact that your acquisition is not working.
- No exclusion between campaign types. Performance Max and Search competing for the same queries, Shopping overlapping with PMax, remarketing audiences catching people who were going to buy anyway.
- Landing pages that were never built for the query. Sending every keyword to the homepage is a structural decision, and it caps your conversion rate permanently.
How to find your own structural leaks
Give this an afternoon. You do not need tools beyond the Google Ads interface.
- Open the conversions table and read every conversion action. For each one, ask: would I be happy to pay money for exactly this event? Set anything that is not a real inquiry or a real sale to secondary. Do this first, because it changes everything else.
- Segment spend by campaign and write the business purpose of each one in a sentence. If you cannot, that campaign has no purpose. Any campaign whose sentence is just its name is a candidate for rebuilding.
- Split brand from non-brand and look at the two numbers separately. Filter the search terms report for your company name. Subtract that spend and those conversions from the totals. The remaining number is your actual acquisition performance.
- Check location settings on every campaign. Set targeting to presence rather than presence or interest unless you have a reason not to.
- Run the search terms report over the last ninety days sorted by cost, and read the top hundred lines. Not to add negatives yet. Just to see what you are actually buying. Most owners are shocked by this exercise.
- Compare cost per conversion by product or service line. If they are all lumped into one campaign, this is your evidence for splitting them.
The tactical work that does still matter
I am not arguing that day-to-day management is pointless. Some tactical work pays for itself many times over, particularly negative keyword maintenance, asset testing in responsive search ads, and watching for the query drift that broad match and AI-driven matching produce over time. What I am arguing is about sequence. Tactical optimization on top of a broken structure compounds the error. You get better and better at buying the wrong thing.
There is also a class of problems that looks structural and is not. Rising click costs in a competitive market are not a structural flaw. Neither is a genuinely hard product with a long sales cycle. Do not rebuild an account because the market got more expensive. Rebuild it because the account is measuring the wrong thing or mixing things that should be separate.
What to do next
Before you touch a single bid, do two things. Confirm that your primary conversion action is an event you would happily write a check for. Then separate brand from non-brand and look at the cold traffic numbers on their own. In my experience those two checks explain the majority of the gap between what an account reports and what the business actually feels.
If the answers make you uncomfortable, that is useful information. It means the fix is available and it is not a bidding fix. Rebuilding structure is a week of work, not a year of it, and it is the only kind of change that keeps paying.