Your PPC Account Is Optimising Toward the Wrong Thing

Google's bidding algorithms are genuinely good at finding more of whatever you tell them you want. That is the problem. If you tell them the wrong thing, they will efficiently, relentlessly deliver more of it.

Almost every underperforming paid search account we inherit has a tracking defect at its root, not a bidding or creative one.

Mistake 1: counting the enquiry, not the sale

The most expensive error. The conversion fires when someone submits a form, so the algorithm optimises for form submissions — and gets very good at finding people who fill in forms and never buy.

You end up with a falling cost per lead, a rising lead volume, and flat revenue. The account looks like it is improving right up until someone compares it against the sales ledger.

The fix: track the outcome that has commercial value. Where the sale happens offline or weeks later, import offline conversions from your CRM so the platform learns which clicks became customers. This single change halved cost per enrolment on a recent education engagement.

Mistake 2: double-counting

Tags fire twice. A tag deployed both directly and through a tag manager. A thank-you page that counts a conversion on every refresh. A single-page application firing on route change and on load.

Inflated conversion counts make everything look better than it is, and make genuine improvements impossible to detect. Audit what fires on each key page and count the actual network requests, not what the container claims is installed.

Mistake 3: no negative keyword hygiene

Broad match without negatives is an invitation. Accounts we audit routinely show a substantial share of spend on queries with no purchase intent whatsoever — job seekers, students researching, people looking for free alternatives, and competitor brand terms nobody chose to bid on.

Pull twelve months of search terms, sort by cost, and read them. It is tedious and it is usually the highest-return hour of the engagement.

Mistake 4: attribution set and forgotten

Last-click attribution systematically undervalues everything that happens before the final touch. If your customers research over weeks, last-click will tell you brand search is brilliant and everything that created the demand is worthless — and you will cut the campaigns that were actually generating it.

Use a data-driven model where volume supports it. At minimum, understand what your current model is doing before making budget decisions on its output.

Mistake 5: ad and landing page mismatch

The ad promises a specific thing. The landing page is the generic homepage. Quality Score falls, cost per click rises, and conversion rate falls — three penalties from one lapse.

Each ad group should arrive somewhere that matches its query specifically. This is usually cheaper to fix than any bidding adjustment and returns more.

Mistake 6: judging too early

Statistical significance requires volume. Pausing a keyword after eleven clicks and no conversion is not optimisation, it is noise-chasing.

Set a minimum data threshold before acting — a defined number of clicks or a defined spend — and hold to it. Constant intervention also resets the learning phase on automated bidding, which costs you performance directly.

A working order of operations

1. Fix tracking. Nothing downstream is trustworthy until the conversion definition is right.
2. Clean the search terms. Stop paying for queries you would never want.
3. Match landing pages to ad groups. Cheapest conversion rate improvement available.
4. Restructure into tight, intent-based groups. Then let automated bidding work.
5. Then, and only then, adjust bids and budgets.

Most accounts are optimised in exactly the reverse order, which is why so much effort produces so little movement.

The uncomfortable question

Ask what a customer is worth to you, and what you are currently paying to acquire one. If either number is unavailable, the tracking work is not a technical chore — it is the entire foundation of knowing whether the channel works at all.

Our PPC team starts every engagement with a tracking audit, because optimising against a broken signal is worse than not optimising at all.