Paid search management restructured around profit

What changed in Google Ads and why does it matter?

Two shifts have made the 2018 playbook actively harmful. The first is automated campaign types — Performance Max and, more broadly, AI-driven campaign management — which have absorbed a large share of standard-search inventory and removed much of the control that manual campaign structure used to provide.

The consequence is not that you should use these campaign types or avoid them. It is that the old lever set — dozens of tightly segmented campaigns with manual bidding and exhaustive negative keyword lists — has been substantially absorbed into the platform. Accounts still structured that way now carry management overhead without proportional control.

The second shift is in bidding. Exact-match control has eroded as Google infers intent from user signals. Broad match with strong conversion value data can now outperform exact match for many advertisers — which inverts the standard advice about match types that most textbooks still carry.

What does account structure actually do?

Structure is not an aesthetic preference. It determines three concrete things: how quickly the algorithm has enough data to make good decisions, how cleanly you can attribute value back to intent, and how much budget you can safely commit before knowing whether it works.

1. Group by intent, not by keyword similarity. Someone searching "marketing agency pricing" and someone searching "marketing agency" are at different stages and need different landing pages and different offers. Splitting by shared words rather than underlying intent produces a blended conversion rate that tells you nothing.
2. Keep budgets interpretable. Too many campaigns each with a small budget means none of them accumulate data fast enough for automated bidding to work. Consolidation is usually better than proliferation.
3. Use negative lists as strategy, not maintenance. A negative list documents a decision. If nobody can explain why a term is excluded, it is probably costing you.
4. Keep brand and non-brand separate. Blended brand and non-brand reporting hides the only question that matters: is non-brand acquisition working?

Is Smart Bidding worth it?

Smart Bidding optimizes toward a conversion goal you define. It works when you have sufficient volume and a goal that genuinely correlates with revenue. It fails when the goal is a proxy that does not correlate with value — optimising toward form fills when low-value leads convert those forms.

How do you match ads to search intent?

Intent matching is where most wasted spend lives, and it is a copy-and-structure problem rather than a bidding problem. Four intent tiers appear consistently in commercial queries, and each needs different handling.

The diagnostic question is simple: if this person searched that, what would be the most useful page on your site for them? When the honest answer is "not our service page", routing that traffic to a service page is what produces the conversion rate nobody can explain.

What attribution can you realistically trust?

Platform attribution is a directional signal, not ground truth. Several effects make it structurally optimistic about paid search.

• View-through inflation. Some conversions are credited to an ad because the ad was merely viewable, not because it caused anything.
• Last-click bias. The final click takes credit for journeys that began elsewhere.
• Ad blockers and ITP. A meaningful share of conversions never fire a tag, and they are not randomly distributed — they skew toward technical and privacy-conscious audiences.
• Consent mode modeling. Denied-consent traffic produces modelled figures that behave differently from observed ones.
• Cross-device gaps. A phone click followed by a laptop purchase is frequently invisible.

Treat platform-reported return on ad spend as a modelled estimate and label it as one. Where you can import offline conversions — actual booked jobs, actual revenue — do it, and it will usually change which campaigns deserve budget. The measurement detail is in the GA4 guide.

When should you not run paid search at all?

More often than agencies admit. Paid search amplifies an existing offer; it does not rescue a broken one. It is a poor investment when any of these are true.

• Conversion tracking is unreliable. You would be optimising toward noise.
• There is no defined offer. Paid search against "we do marketing" cannot work.
• Gross margin cannot support the cost per click in your category.
• There is no sales capacity to handle the leads. Unanswered leads are pure cost.
• The landing page does not match the ad. You are paying premium prices to increase bounce rate.
• The business is pre-revenue with no proven demand. Test demand cheaply first.

None of these are permanent disqualifications, but all of them are reasons to spend the next fortnight on something else. If you want the arithmetic worked through for your business, start with the break-even calculation.