How to Audit a Google Ads Account for Wasted Spend

Most wasted Google Ads spend comes from four places: search terms you never intended to bid on, conversion actions that count the wrong thing, campaign types with settings you did not choose, and budget sitting on keywords that have never produced an enquiry. This is the audit we run on every account we take over, in the order we run it — eleven steps, ordered so the fastest savings come first and each step's findings stay usable for the next. If a full pass is not realistic today, the abbreviated four-check version near the end finds most of the waste in half an hour, and the scoring list after it is for tracking an account over time.

Work through it before changing any bids. Optimising an account with broken tracking makes the data worse rather than better.

Where does the money actually leak?

Almost always in the gap between the keywords you bought and the searches you paid for.

You bid on "web design Beirut". Google, on broad match, may also show your ad for "free website builder", "web design jobs", "how to design a website" and "web design course". Every one of those is a click you paid for from someone who was never going to hire you.

That gap is invisible from the keywords tab. It is only visible in the search terms report, which is why that is step one.

Step 1: Read the search terms report

Where: Campaigns → Insights and reports → Search terms. Set the date range to the last 30 days and sort by cost, descending.

Read the top 50 by spend. For each, ask one question: would I want this person to call me?

Three categories emerge, and each has a different fix:

• Relevant and converting. Leave alone. Consider adding as an exact-match keyword in its own ad group.
• Relevant but not converting. Do not pause yet — this may be a landing page problem rather than a keyword problem. Note it and check the page it lands on.
• Irrelevant. Add as a negative keyword. This is the money you get back today.

Common irrelevant patterns worth having as negatives from day one: free, cheap, jobs, salary, course, tutorial, template, download, how to, DIY, and the names of any large competitor you are not trying to conquest.

Step 2: Check what your conversions actually count

Where: Goals → Conversions → Summary.

This is the check that most often changes how an account performs, because an account optimising toward the wrong action will do it efficiently.

Look at each conversion action and ask what it counts. Watch for:

• Page views counted as conversions. A "thank you page view" is reasonable. "Any visit to the contact page" is not — you will buy visits to the contact page.
• Duplicate conversions. The same enquiry counted by both a form submission and a page view, doubling your reported results and halving your apparent cost per lead.
• Actions marked Primary that should be Secondary. Only what you actually want more of should be Primary. Newsletter signups and PDF downloads usually belong in Secondary.
• Nothing at all. Accounts running without conversion tracking are optimising toward clicks, which Google will deliver.

If tracking is broken, stop the audit and fix it first. Everything downstream depends on it, and conversion tracking has its own diagnostic order.

Step 3: Check match types and negatives

Where: the Keywords tab, with the Match type column added.

Broad match is the default and the most expensive one to get wrong. It hands Google discretion over which searches you pay for. That works when the algorithm has plenty of good conversion data to learn from, and burns money when it does not.

Three practical rules:

• New accounts start on phrase match. Move to broad only once conversion tracking has been reliable for a month or more.
• Every campaign needs a negative list. If your negative keyword list is empty, that alone explains a large share of your waste.
• Watch for keywords competing with each other. Two ad groups bidding on near-identical terms raises your own cost per click.

Step 4: Look at the segments — device, location, time

Where: any campaign → Segment, or the Locations and Ad schedule tabs.

Three splits that regularly reveal waste:

Device. If mobile costs twice as much per conversion as desktop, that is usually a landing page problem, not an audience problem. Check the page on a phone before adjusting bids down. Adjusting the bid hides the symptom and keeps the cause.

Location. Check the setting under Locations → Location options. The default in many accounts is "presence or interest", which shows your ads to people interested in your area rather than in it. For a local business that can mean paying for clicks from other countries. "Presence" is usually what you meant.

Time. If you are a restaurant closing at midnight, ads running at 4am are buying clicks nobody can act on. Ad schedules are underused.

Step 5: Open up Performance Max

PMax is the campaign type most likely to be spending money you cannot see, because its reporting is deliberately thin.

What to check:

• Brand terms. Without brand exclusions, PMax will often take credit for people searching your own business name — traffic you would have got for free. Add brand exclusions and watch what happens to reported conversions.
• The placement report. Campaigns → Insights and reports → When and where ads showed. It is less detailed than a normal display placement report but shows the worst offenders.
• Asset group performance. Weak assets drag the whole campaign, and PMax will not tell you plainly which ones.
• Whether it should exist at all. PMax needs conversion volume to learn from. On a small account with a handful of conversions a month, a well-built search campaign usually outperforms it and is far easier to diagnose.

Step 6: Match the landing page to the ad

Not a settings check, and often the largest single gain available.

Google scores how closely your landing page matches the ad and the search behind it. A better score means you pay less per click for the same position — so a mismatched page costs you twice, once in conversion rate and once in click price.

The test: read the search term, then look at the page it lands on. Does the page answer that exact question in the first screen? If someone searched for one specific service and landed on your homepage, the answer is no. Why landing pages beat homepages covers the mechanics.

Step 7: Account structure

Where: the campaign list, then Ad groups, with the Keywords column visible.

Structure problems do not waste money as visibly as a bad search term, but they make every other problem harder to see and slower to fix.

• Ad groups covering too many themes. The test: can you write one ad that is genuinely relevant to every keyword in the group? If not, the group is too broad, and Ad relevance will show it.
• Campaigns split for no reason. Separate campaigns are for things needing separate budgets, locations or schedules. Splitting by theme alone belongs at ad group level.
• Single-keyword ad groups everywhere. A fashion a few years ago, and mostly counterproductive now — it starves each group of data.
• Branded and non-branded in the same campaign. Brand terms are cheap and convert well. Mixed in with cold traffic, they flatter the campaign average and hide that the non-brand half is failing.
• Orphaned campaigns. Paused-then-unpaused campaigns from a previous manager, still spending. Sort the campaign list by cost and check you recognise everything in it.

Step 8: Read the ads and assets

Where: the Ads tab, with the Ad strength column added.

• How many ads per group? Fewer than two and there is nothing to compare. Three responsive search ads is plenty.
• Ad strength below "Good". Not a ranking factor in itself, but a reliable flag for thin or repetitive headlines.
• Pinned headlines everywhere. Pinning every asset removes the testing that makes responsive ads work. Pin the one thing that must always appear — a price qualifier, a location — and let the rest rotate.
• Missing sitelinks, callouts and structured snippets. These are free extra space on the results page and free extra reasons to click. An account without them is buying a smaller ad for the same money.
• Call assets with no schedule. A phone number showing while you are closed generates missed calls, which cost the same as answered ones.
• Ad text that does not repeat the keyword. The closer the ad language matches the search, the better the relevance score, and the less you pay.

Step 9: Audiences and remarketing

Where: Campaigns → Audiences.

• Observation vs Targeting. Set to Targeting, an audience restricts who sees your ads — often applied accidentally, and it can throttle a campaign to near-zero impressions. Observation is usually what was meant.
• No remarketing list at all. The cheapest traffic available is people who already visited. If no list exists, nothing is being collected, and you cannot build a list retroactively.
• Remarketing with no frequency cap. Following the same person around the internet forty times a week does not make them likelier to call.
• Customer lists never uploaded. If you have a list of past customers, it is both an audience and an exclusion — useful for repeat business, and useful for excluding people who already bought.

Step 10: Bidding strategy

Where: Campaign → Settings → Bidding, then Tools → Change history.

• Smart bidding on thin data. Below roughly 30 conversions a month, the algorithm is guessing. Manual or enhanced CPC is more predictable until volume arrives.
• Target CPA set too high. It reads as permission to pay far more per click than you intended. Set it near your actual historical cost per acquisition, not your hoped-for one.
• Target ROAS on an account that cannot measure revenue. If conversions have no values attached, a revenue target is optimising against nothing.
• Strategy changed recently. Check the change history. A CPC shift right after a bidding change is the strategy working as designed, not a problem to solve.
• Portfolio strategies nobody remembers creating. These sit above campaigns and quietly override what you set at campaign level.

Step 11: Budget and allocation

Where: the campaign list, sorted by cost, with Search impression share and Search lost IS (budget) added as columns.

• Budget-limited campaigns that convert well. "Limited by budget" on your best campaign is the easiest win in the account — that is proven demand you are declining.
• Full budget on campaigns that never convert. The opposite problem, and more common.
• Shared budgets hiding the picture. Convenient, but they make per-campaign spend hard to reason about during an audit.
• Impression share lost to rank rather than budget. A different problem with a different fix: rank means relevance and bid, not money.
• Seasonal budgets nobody reset. A raise for a busy period that stayed raised.

The 30-minute version

If a full pass is not realistic, these four checks find most of the waste in most accounts:

1. Search terms report, last 30 days, sorted by cost. Read the top 20. Add negatives. (10 minutes, immediate saving.)
2. Goals → Conversions. Confirm each action counts an enquiry, not a page view, and that nothing is double-counted. (5 minutes.)
3. Locations → Location options. Change "presence or interest" to "presence" if you are a local business. (2 minutes.)
4. Campaign list sorted by cost. Check you recognise every line, and look for "Limited by budget" on anything profitable. (10 minutes.)

Scoring the account

Score each category out of 5, where 5 means no action needed and 1 means it is actively costing money. What a 5 looks like:

• Search terms — reviewed within the last month, negatives applied.
• Conversion tracking — one action per outcome, no duplicates, everything firing.
• Match types — deliberate rather than default, with a maintained negative list.
• Segments — location set to presence, schedule matching your opening hours.
• Performance Max — brand excluded and placements reviewed, or not running at all.
• Landing pages — each ad group points at a page that answers that search.
• Structure — one theme per ad group, brand split from non-brand.
• Ads and assets — two or three responsive search ads per group, sitelinks and callouts present.
• Audiences — remarketing collecting, set to Observation rather than Targeting.
• Bidding — strategy matched to the conversion volume actually available.
• Budget — nothing profitable sitting at "Limited by budget".

Anything scoring 1 or 2 goes on the fix list. The total out of 55 is only useful as a before-and-after measure — the individual low scores are what you act on.

Common audit mistakes

• Changing things while auditing. Read the whole account first. Half the findings turn out to share one cause, and you will fix it once instead of six times.
• Auditing on too short a window. Seven days of data on a small account is noise. Use 30 to 90 days.
• Trusting the conversion column before checking what it counts. Every downstream judgement inherits that error.
• Restructuring on day one. It resets the algorithm's learning and costs roughly two weeks of performance for a benefit you have not yet proven.
• Comparing to published benchmarks. Industry averages blend branded and non-branded traffic across wildly different markets. Compare your ad groups against each other instead.
• Auditing the account but never the landing page. Half of Quality Score lives on your website, not in the account.

What do you actually change first?

In this order, because it maps to how quickly the money comes back:

1. Add negative keywords from the search terms report. Same-day saving.
2. Fix conversion tracking if anything looked wrong. Everything else depends on it.
3. Fix the location setting if it was on "presence or interest" and you are local.
4. Pause keywords with significant spend and no conversions over a meaningful period — at least 100 clicks or a month, whichever comes first.
5. Then, and only then, adjust bids.

Resist reorganising the account structure on day one. It is satisfying and it resets the algorithm's learning, which costs you two weeks of performance for a benefit you have not yet proven you need.

What if the account is small?

Most advice assumes budgets that most businesses do not have. On a small account:

• Fewer campaigns, not more. Splitting a small budget across six campaigns starves all of them and none ever accumulates enough data to optimise.
• Ignore most automated bidding. Smart bidding needs conversion volume. Below roughly 30 conversions a month, manual or enhanced CPC is often more predictable.
• Check weekly, change fortnightly. Small accounts are noisy, and reacting to noise is its own form of waste.

How we work

We audit before proposing anything. Management is 20% of monthly ad spend with a $200 minimum, invoiced separately from the budget — and you own the account and pay Google directly.

If you want the audit run on your account, ask. We will tell you what we find, including if the answer is that it is already well run.

More on this in practice: Google Ads in Lebanon, and why your cost per click went up if that is the specific thing that brought you here.