Google Ads management

Which Google Ads campaign type fits your goal?

Pick the campaign type by what you are trying to buy, not by what looks newest. Search buys existing demand from people typing a query. Shopping buys product visibility for retail intent. Demand Gen and Display buy attention from people who are not searching yet. Performance Max buys across all of them at once and decides the split itself.

Most accounts should start with Search, because the intent is highest and the feedback is fastest. Ecommerce accounts add Shopping early, since a product feed usually outperforms text ads for product queries. Display and Demand Gen suit brands with a real awareness problem and creative worth showing. Video sits with YouTube and is planned separately.

The mistake we see most often is starting with Performance Max because it is easy to launch. It needs conversion history and good assets to work with. Without those it spends on cheap, low-quality placements and reports a flattering cost per conversion that does not show up in the bank.

• Search: high intent, fastest learning, the right first campaign for most accounts
• Shopping: product feed does the work, essential for retail
• Demand Gen and Display: awareness and audience building, judged on assisted impact
• Performance Max: broad automated coverage, only once tracking and assets are solid

How should a Google Ads account be structured?

Structure around themes and budgets, not around single keywords. The old single keyword ad group approach is finished. Close variant matching means Google will serve your ad for related queries anyway, so splitting every keyword into its own ad group fragments the data and starves the smart bidding algorithms of signal.

A workable structure groups keywords by shared intent and shared landing page. If two queries deserve the same page and the same message, they belong in the same ad group. If they need different messages or different budget control, they belong apart. Campaigns are the level where you control budget, location, bidding and schedule, so split campaigns when you need to control one of those things separately.

Keep the account small enough to manage. A tidy account with fifteen well-fed ad groups beats a sprawling one with two hundred that each get three clicks a week. Volume per ad group matters because automated bidding needs conversions to learn from.

• One campaign per budget or bidding decision you need to control
• Ad groups grouped by shared intent and shared landing page
• Enough traffic per ad group for bidding to learn from
• Separate brand and non-brand campaigns so reporting stays honest

How do match types actually behave now?

Match types are looser than their names suggest. Exact match no longer means exact. It covers close variants including reorderings, misspellings, plurals and queries Google judges to have the same meaning. Phrase match covers the meaning of the phrase rather than the literal word order. Broad match uses the whole context of the account, including your landing page and other keywords, to decide relevance.

That means the search terms report is now more important than the keyword list. What you bid on is a suggestion. What you actually buy is whatever Google matched. We read search terms weekly on new accounts, because that is where the waste hides and where the next winning keyword usually appears first.

Broad match can work well, but only with tight conversion tracking and smart bidding to steer it. On manual bidding or thin conversion data it behaves like a leak. We usually start tighter, prove the conversion signal, then open up where the economics justify it.

What does a real negative keyword strategy look like?

Negatives are not a one-off cleanup task. They are an ongoing filter that shapes what your budget buys. The starting point is a list of obvious commercial mismatches: free, jobs, salary, DIY, wholesale, second hand, complaints, and any competitor terms you do not want to pay for.

Beyond that, the work is routine. Each week you read the search terms report, mark anything that will never convert, and add it at the right level. Account level negatives for whole categories you never want. Campaign level for things that belong in another campaign. Ad group level for steering traffic between close themes.

Two details matter. Negative match types work slightly differently to positive ones, and negative phrase and exact do not cover close variants, so misspellings need adding separately. And negative lists shared across campaigns save hours once you run more than a couple of campaigns.

• Build a starter negative list before launch, not after the first invoice
• Review search terms weekly early on, monthly once stable
• Use shared negative lists across campaigns to avoid repeated work
• Add misspellings explicitly, since negatives do not catch close variants

What actually moves Quality Score?

Quality Score has three parts: expected click-through rate, ad relevance and landing page experience. Of those, expected click-through rate carries the most weight, and it is the one most people ignore while they fiddle with keyword density.

The practical route to a better score is tighter alignment. The query, the ad headline and the landing page headline should say recognisably the same thing. When someone searches for a specific service and lands on a generic homepage, relevance and landing page experience both suffer, and you pay more per click for the privilege.

Quality Score itself is a diagnostic, not a target. Chasing the number is a distraction. Use the three component ratings to find where the mismatch sits, fix that, and let the score follow. A better score shows up as a lower cost per click for the same position, which is where the money is.

Manual bidding, target CPA or target ROAS?

Choose the bidding strategy by how much conversion data you have. Manual or maximise clicks suits a brand new account with no conversion history, where you are buying data rather than optimising. It gives you control while the account has nothing to learn from.

Target CPA suits lead generation once you are recording a steady flow of conversions, because every lead is worth roughly the same. Target ROAS suits ecommerce, where order values vary and you care about revenue per pound rather than count. Maximise conversions or maximise conversion value with no target set is a reasonable middle step while you learn what your real cost or return looks like.

Whatever you pick, change targets gradually. Large swings in a target reset the learning period and stall performance for days. Moving a target by ten to fifteen percent at a time keeps campaigns stable while you find the level where volume and profit balance.

• New account, no conversions: manual or maximise clicks, briefly
• Steady lead flow with similar lead values: target CPA
• Ecommerce with varying order values: target ROAS
• Adjust targets in small steps to avoid restarting the learning phase

Getting conversion tracking right, including offline sales

Every bidding decision Google makes depends on your conversion data, so tracking is the first thing we check and the first thing we fix. That means one primary conversion action that reflects real business value, secondary actions recorded but not bid on, and a check that the numbers reconcile with your CRM or order system rather than just appearing in the interface.

Enhanced conversions improve accuracy by sending hashed first-party data, such as an email address the customer already gave you, alongside the conversion. It recovers measurement that browser restrictions would otherwise lose. It requires a privacy policy that covers it and correct consent handling, which we set up rather than skip.

For businesses with a sales process, offline conversion import is the bigger win. You send the qualified lead or closed deal back to Google against the original click. Bidding then optimises towards deals rather than form fills, which usually changes which keywords look good. Accounts that make this change often find their cheapest leads were their worst ones.

Performance Max: the trade-off worth understanding

Performance Max trades control for coverage. You supply asset groups, which are collections of headlines, descriptions, images, video and audience signals, and Google decides where and to whom they run across Search, Shopping, YouTube, Display, Discover, Gmail and Maps.

The transparency cost is real. Reporting is thinner than in standard campaigns, placement detail is limited, and it can quietly absorb traffic that your Search campaigns would have won more cheaply. Brand terms are the classic case, which is why we usually exclude brand from Performance Max so the reported return is not just your own customers coming back.

It works best when you feed it properly. Multiple asset groups split by product or service theme rather than one giant group. Real video assets rather than auto-generated ones. Clean conversion data with sensible values. Search theme and audience signals to point it in the right direction. Done that way it can be a strong performer. Launched empty, it is an expensive way to buy your own brand traffic.

• Split asset groups by theme so reporting means something
• Add your own video rather than accepting the auto-generated one
• Exclude brand terms so the reported return reflects new demand
• Use search themes and audience signals as steering, not as targeting

The mistakes that waste the most budget

The same handful of problems account for most wasted spend. Search partners and Display expansion left switched on inside Search campaigns, quietly buying low-quality traffic. Location settings set to presence or interest, so you pay for people in other countries who once looked up your city. Broad match running against poor conversion data.

Then there are the reporting problems. Counting every conversion when one enquiry gets counted three times. Bidding on a thank-you page view that fires on refresh. Judging campaigns on last click when the assisting campaign gets no credit. These do not just misreport, they actively misdirect the bidding.

Finally, the structural ones. No brand and non-brand split, so a healthy-looking blended cost per acquisition hides a poor non-brand result. Ad rotation left on the wrong setting. Landing pages that load slowly on a phone, where most of the clicks come from. We check all of these before we touch a bid.