The Complete Google Ads Guide for 2026

Want certified experts running your account? Explore our Google Ads management service →

The Search Terms Report Is Where the Money Is

If you do one thing to an underperforming Google Ads account, read the search terms report. Not the keyword report — the search terms report, which shows what people actually typed rather than what you told Google to target. The gap between the two is where most wasted spend lives.

Broad match combined with automated bidding will find queries you never intended to buy. Some of those are valuable discoveries. Many are not, and they persist because nobody looks. A single afternoon spent building out negative keywords from that report is routinely the highest-return work available in an account.

• Build negatives at the right level. Campaign-level negatives for things you never want; ad-group-level for steering traffic between groups. Mixing them up creates conflicts that are hard to debug later.
• Watch for close variants. Google’s interpretation of exact match has broadened considerably. “Exact” no longer means what it once did, and the search terms report is the only place you will see the difference.
• Check your brand terms. Bidding on your own brand is sometimes justified defensively, but it is often subsidising clicks organic would have won free. Test pausing it and watch total brand traffic rather than paid conversions alone.
• Look for competitor names. These often appear unintentionally through broad match, convert poorly, and cost a premium.

Feeding the Automation Properly

Google’s bidding automation is genuinely capable, and it optimises toward whatever signal you give it. Give it a weak signal and it will efficiently maximise something you do not want.

The most common failure is optimising toward form submissions of wildly varying quality. If a newsletter signup and a request for a quote both count as a conversion, the system will learn to buy whichever is cheaper — and that is the newsletter signup. The account will report an improving cost per conversion while the sales pipeline dries up.

Fix this by defining conversions that reflect commercial value. Feed qualified lead status back from your CRM where possible, use offline conversion imports for longer sales cycles, and assign values that reflect actual worth rather than counting every action as equal. Automation given good signals outperforms manual bidding comfortably; given poor signals, it fails faster and more expensively than manual ever did.

Performance Max Without Losing Control

Performance Max consolidates inventory across search, shopping, display, YouTube and Discover into a single campaign, and trades visibility and control for reach. Used carelessly it becomes a black box that absorbs budget and reports impressive-looking results built largely on branded and remarketing traffic that would have converted anyway.

• Separate branded traffic. Without brand exclusions, Performance Max will happily claim credit for people already searching for you. This inflates reported performance dramatically.
• Use asset groups deliberately. One asset group per genuinely distinct audience or product set, not one covering everything. The signals get muddled otherwise.
• Supply real creative. Auto-generated assets fill gaps but rarely perform as well as properly produced copy and imagery. This is where campaign quality is actually decided.
• Watch the shopping feed. For retail accounts, feed quality drives Performance Max results more than any campaign setting. Titles, attributes and images do the heavy lifting.
• Keep a search campaign running. Exact-match search on your core commercial terms gives you visibility and control that Performance Max will not.

Landing Pages Decide Cost Per Acquisition

The account is only half the system. Doubling landing page conversion rate halves cost per acquisition without touching a bid, and it is frequently the cheaper lever — yet the majority of optimisation effort goes into the account because that is where the interface is.

Match the page to the ad specifically. Sending every campaign to the homepage wastes the intent you just paid for; someone who clicked an ad for a specific service should land on that service, with the offer from the ad visible without scrolling. Check load speed on mobile using field data rather than a lab score, keep forms to the fields you genuinely need, and make the next step unmistakable.

Then judge the whole system on cost per qualified enquiry rather than on platform-reported ROAS. Platform figures use the platform’s own attribution model, which is generous to the platform by design. Reconciling against your own revenue data usually produces a different and more useful number.

Account Structure That Stays Manageable

Google Ads account structure has moved away from the highly granular single-keyword ad groups that were standard years ago. Automated bidding needs conversion volume to learn from, and splitting a small number of conversions across forty ad groups starves every one of them.

Consolidate enough that each campaign accumulates meaningful data, and separate only where you genuinely need different budgets, different bidding targets or different geographic settings. A useful test: if two ad groups would take the same bid strategy, the same budget and the same landing page, they probably should not be separate.

Separate what has genuinely different economics. A product line with a fifteen percent margin cannot support the same cost per acquisition as one with sixty, and putting both under a single target guarantees that one is overspending while the other is underserved.

What to Review, and How Often

• Weekly: search terms report and negative keyword additions. This is the habit that compounds.
• Fortnightly: conversion tracking sanity check. Tags break silently after deployments and nobody notices until the month-end report looks strange.
• Monthly: cost per qualified enquiry reconciled against your CRM, not platform-reported conversions. Also creative refresh — fatigue is real and measurable.
• Quarterly: whether the account structure still matches the business, whether targets still reflect margin, and whether any campaign has quietly stopped earning its budget.

Resist the urge to change bids daily. Automated strategies need a learning period, and interrupting it repeatedly produces worse results than leaving it alone. The discipline that separates well-run accounts from poorly run ones is mostly about frequent observation and infrequent intervention.