Name what "not working" means: no impressions, no suitable clicks, no useful enquiries, no records or no viable work.
Each of those has its own route through the ten checks below. Verify the evidence before you believe it. Then fix the break you can see before you talk about spending more.
Start with the symptom and what changed
Pin down the commercial goal, the date range and the comparison period before opening a single setting.
Then ask which of three histories you're in. The campaign never worked. It launched recently and hasn't had time. Or it worked and then stopped. The third needs a timeline.
Google's Change history sits under the Campaigns menu. It lists the changes made to the account, campaigns and ad groups for the past two years. Each entry shows the user, the Google Ads system or the API tool that made it. The history maps those changes on a timeline against impressions, clicks, conversions and cost.
Change history only records what happened inside the account. Put website releases, form or booking integration changes, offer changes, staffing and seasonal demand beside it. A change that lines up with the decline is a lead to investigate. It isn't proof that the edit caused the drop.
Symptom | First evidence to inspect | Check |
|---|---|---|
No impressions or almost none | Account notices, statuses, dates, billing and policy | 1 |
Impressions but few clicks or little spend | Impression share, bid strategy status, restrictions in combination | 2 |
Conversions in reports, no enquiries in the inbox | Conversion action definition, counting and a test submission | 3 |
Clicks and spend, nothing useful | Search terms against outcomes, exclusions, locations, schedule | 4 and 5 |
Budget gone early or bids doing odd things | Spend concentration, targets, automation and applied recommendations | 6 |
Clicks but no enquiries | Ad promise, destination, form, phone and booking route | 7, 8 and 9 |
Enquiries but no customers | Response, qualification, capacity and acquisition cost | 10 |
Before trusting any of those symptoms, check the report itself:
the right account
the right campaign
the date range
the filters
the account time zone
the reporting delay
"Zero" is sometimes a filter.
When the campaign worked and then stopped, write the timeline down before you change anything.
Date: [when the symptom began, from the performance data]
Symptom: [what changed, in which metric, by how much]
Account change: [the Change history entries near that date, with who or what made them]
Website change: [releases, form, booking or tag changes near that date]
Business change: [offer, price, hours, staffing, stock or season]
Evidence to verify: [what would confirm or rule out each candidate]
Check 1: Can the account and ads serve?
Symptom: no impressions, or a campaign that spent yesterday and nothing today. Google's low traffic diagnosis page starts where this check does. An account that is suspended, under a security review or carrying a billing issue won't serve. Even an enabled campaign needs active ad groups, assets and ads inside it, with no policy issues.
Read the account notices and the statuses of the campaign, ad group and ad. Then check the start and end dates and the schedule. Remember that Google's ad status page says most reviews are completed within 1 working day. More complex reviews can take longer.
Action: resolve the stated billing, verification or policy issue through the official account route, and nothing else. A new account to dodge a suspension or a disguised destination breaks Google's circumventing systems policy. Google suspends for that without warning. A cycle of appeals with nothing fixed just burns time.
An "Eligible" label on the campaign does not establish that every ad and keyword can serve. When one search term matters, use the Ad Preview and Diagnosis tool. Google says it helps identify why a specific search term may not serve. Don't keep searching for the ad, and don't click it.
Recheck: statuses clear, a preview shows the ad, impressions resume within the review window. A first time setup that never got this far belongs elsewhere. Our guide to setting up Google Ads for a small business covers it.
Check 2: Can the campaign reach useful demand?
Symptom: eligible ads, little traffic. The same Google page lists the usual constraints. The point is that they act together:
Very low bids and a small budget.
Match types and negatives that narrow the reach.
Location and audience settings that exclude the people searching.
Overlapping campaigns competing in the same auctions.
Insufficient conversion data for an automated strategy.
Weak ad quality and the other advertisers in the auction.
Check the bid strategy status and its targets first. A tight CPA target can throttle a campaign before any keyword gets a chance. So can an ambitious ROAS target, a narrow audience set to Targeting or a short schedule.
Broadening everything blindly buys unsuitable traffic. Google's own suggestion is to switch location from Presence to Presence or interest. Weigh that against whether the business can serve the people the switch would add.
Treat every missed impression as a budget problem
Separate loss attributed to budget from loss attributed to Ad Rank before choosing an action
That's a conceptual comparison, not a result. Impression share is measured against the auctions the ads were eligible for, not the whole market. Google's impression share page splits the lost share into budget and rank. Our guide to Google Ads metrics that matter covers how the share is estimated and when it updates.
Google's Ad Rank page says Ad Rank is calculated every time someone searches. The inputs are the bid and the auction time quality of the ad and landing page. They also include:
the Ad Rank thresholds
the competitiveness of the auction
the context of the search
the expected impact of assets
The Quality Score column in the account is a summary diagnostic, not that formula. A low score is a prompt to look rather than a cause to fix. Our guide to Google Ads Quality Score explains why.
Action: loosen the specific constraint the evidence names. Confirm the offer and the economics support the extra reach before you buy it. Recheck: impression share and lost share move in the expected direction over the next reporting cycle. More budget is not a cure for a rank or eligibility problem.
Check 3: Do conversions mean real outcomes?
Symptom: the reports and the inbox disagree, in either direction. Reconcile the business's actual enquiries for one date range against the campaign goals and conversion actions. Read each action's source, its counting method and whether it is primary or secondary.
Google's primary and secondary actions page says primary actions are reported in the Conversions column. They're also used for bidding when their goal is selected. Secondary actions are reported in All conversions and not used for bidding.
The one exception is an action inside a custom goal. It's used for bidding whichever label it carries. An imported CRM lead and the native form tag for the same submission must not become two leads.
Business outcome and qualification rule: [the real world event and what makes it count]
Conversion action source and campaign goal: [website, import, calls, and the goal it sits in]
Primary or secondary status: [and whether a custom goal overrides it]
Duplicate and counting checks: [one per enquiry or every, imports against native tags]
Test firing, storage and delivery: [the labelled test and where it landed]
Date range, reporting lag and reconciliation: [period, lag allowed, systems compared]
Action: run an authorised, labelled submission through Tag Assistant from the Goals summary. That checks the tag fires when the conversion happens on the live site. A verified action then shows as having no recent conversions until genuine ones arrive.
Tag firing, successful storage and delivery to the inbox or CRM are three separate pieces of evidence. A direct diagnostic test need not appear as an ad attributed lead.
Then allow for lag. Google's conversion lag page says conversions are reported by the date of the query. That makes recent cost per acquisition look inflated until the delayed conversions land. Don't bypass visitor consent to make a test pass.
Recheck: a labelled test appears once, in the right action. And the reconciled count matches the business records within the lag. The implementation lives with our guide to Google Ads conversion tracking.
Check 4: Are searches and match controls suitable?
Symptom: spend on searches the business cannot serve, or useful searches that stopped arriving. Read the costliest visible queries beside the offer they imply and the qualified outcomes behind them. The search terms report omits low activity queries. So it's evidence rather than a census.
Exclude the clear business mismatches with the narrowest match type and scope that remove them. Keep the valid price, research and free quote demand. Inspect the account, shared list, campaign and ad group exclusions together. That's because a negative at any of those levels can block a positive keyword within its scope.
Action: fix the specific control, not the philosophy. Broad match is not inherently wrong and exact only is not a universal cure. Zero conversions from a handful of clicks is not proof that a query is irrelevant.
Recheck: the excluded spend stops, the protected searches keep serving and the qualified outcomes hold after lag. Our guide to Google Ads negative keywords covers selection and conflicts. For the reach controls, see our guide to Google Ads match types.
Check 5: Are location and audience settings right?
Symptom: clicks from places the business does not serve, or none from the places it does. Compare the stated service footprint with three things. Two are in the account: the campaign's location targeting controls and the reported user and matched locations. The third is the actual job or customer addresses from the business records.
Google's location options page calls Presence or interest the default and recommended option. The page also says 100% accuracy is not guaranteed in every situation. Presence targeting can suit a local service. But a visitor planning a trip can legitimately be somewhere else. A suburb name in a query does not establish where the job is.
Then separate the campaign and network segments before reading any blended figure. Check whether an audience left on Targeting is quietly restricting Search reach. Check that the schedule fits the hours demand exists and someone can answer. Performance Max audience signals don't behave like Search audience targeting. So identify the campaign type before prescribing a control.
Action: correct the actual footprint and the segment that is failing. Recheck: location and segment reports match the business's own map of where work comes from.
Check 6: Do budgets and bids fit the goal?
Symptom: budget gone by lunchtime, or a status that reads "Limited by budget". Separate a test budget that is actually too small from the four things that imitate it:
Spend fragmented across campaigns and networks.
Unsuitable demand eating the allowance.
Targets so restrictive the strategy cannot bid.
Tracking that gives the bidding nothing true to learn from.
"Limited by budget" is a delivery status. It is not proof that raising the budget would be profitable.
Focus a constrained test where the commercial value and the demand justify it. That is the subject of our guide to the Google Ads minimum budget. You can model the assumptions in our Google Ads budget estimator.
On bidding, review the outcome being optimised, whether the target is realistic, the lag and the recent changes. Our rule of thumb is 15 to 30 conversions a month before trusting a target. Our guide to Smart Bidding strategies explains this. Google itself sets no minimum for Target CPA. Nor does every new account have to begin on Manual CPC.
Don't change the bidding, the targeting and the page in the same week while the cause is uncertain. You won't learn which one mattered. But deterministic failures are fixed at once. Controlled testing is reserved for real uncertainty. Choosing the strategy itself belongs to the same Smart Bidding guide.
Recommendations deserve their own review. Google's auto applied recommendations page says auto applying recommendations won't increase your budget. But the changes it can apply include targets, bidding and reach. So it can still change how the budget is spent.
Our audit sample is 180 Google Ads account audits from August 2021 to September 2026. We asked owners about auto apply at the audit. In about 1 in 4 accounts, the owner didn't know a category was switched on. Our guide to Google Ads settings that waste budget covers which categories to switch off. It also covers the audience, network and schedule controls from check 5.
Google's optimisation score page says the score runs from 0 to 100%. It changes when recommendations are applied or dismissed. So it's a measure of how the account relates to Google's recommendation list. It isn't simply an adoption score, nor a measure of profit.
Exact change and affected campaign: [what would change and where]
Business objective it supports: [the goal it serves, or does not]
Evidence and expected effect: [what Google predicts and what the account's data says]
Effect on reach, targets or spend: [what moves inside the budget]
Test or approval and accountable owner: [experiment, approve or dismiss, and who decides]
Review date and rollback: [when it is judged and how it is reversed]
Action: review each recommendation against the business goal, its expected effect, the evidence and whether it's reversible. Read the auto apply subscriptions. Then check Change history for what was applied.
Recheck: spend concentrates where the value is. The applied changes are the ones someone approved. And the strategy's targets match an outcome the business recognises.
Check 7: Does the ad promise fit the offer?
Symptom: people click and leave, or the wrong people call. Read the ad for a clear service, a real point of difference and an accurate next step. Check the price basis, service area, availability and proof against what the business can actually substantiate.
The ACCC's guidance on false or misleading claims covers availability, pricing and service claims. It applies the overall impression test to them. Take an urgency line, a guarantee, a review count or a response promise that can't be backed up. That's a legal problem before it's a marketing one.
A planned service electrician advertises immediate 24 hour emergency help
The ad describes planned electrical work and the actual booking availability
The electrician is an example. Correcting the promise isn't evidence of a conversion lift. It's the removal of a mismatch. A low clickthrough rate does not by itself prove the copy is generic. And a high one can mean the ad is attracting people the business cannot help.
Action: rewrite the promise to the service, the area and the availability that exist. Recheck: the enquiries that arrive ask for what the business sells. Our guide on how to write Google Ads owns the craft.
Check 8: Does the enquiry journey work?
Symptom: clicks, spend, a page that looks fine, no enquiries. Inspect the destination the click actually reaches. That includes the sitelink destinations, every redirect and any handoff to a booking or payment domain. Do it on the devices that carry the traffic.
Then work the form with a keyboard and a finger. Tap the number to confirm it dials the right line in the hours the ads run.
The success message proves less than it looks. In about 1 in 5 accounts we audited, a labelled test enquiry never reached anyone who'd respond. The success message showed anyway.
A homepage can fit some intent. So a dedicated page is a decision rather than the default cure.
Action: repair the reproducible fault first. Then match the page to the promise and the decision the visitor is making. Recheck: a labelled test enquiry from the ad's actual destination reaches a person.
Clicks may keep arriving without leads after that. If they do, our guide to Google Ads clicks but no leads runs the deeper investigation. It works from reporting through traffic, page and follow up.
Check 9: Are useful calls visible?
Symptom: a phone that rings and a report that says nothing happened. In Elev8d's service accounts, calls are commonly 30% to 50% of leads, more for emergency trades. That's our observation, not a market statistic. It covers the 45 service accounts we managed from May 2024 to September 2026, all with call tracking.
Our guide to Google Ads call tracking explains how to count calls. A report blind to calls can make a working campaign look broken.
Four things get called a call:
a tap on a phone link
a connected call
an answered call
a useful enquiry
Google's call reporting setup page says call reporting uses Google forwarding numbers. They measure call assets, location assets and call ads. Turning call reporting off stops calls from ads counting as conversions. That affects Smart Bidding.
Google's about call reporting page lists Australia among the countries with forwarding numbers. It also says the advertiser can choose that call length. So a duration is a configured proxy, not a universal qualification test. A 40 second call can be a booked job and a six minute call can be a supplier. Do not present every missed call as an identified qualified lead either.
Two terminology points matter here. They are call assets, not call extensions. And nobody should be told to create new call only ads. Google's call ads notice says new call only ad creation ended in February 2026. Existing call only ads stop receiving impressions in February 2027.
Action: measure calls from ads and website calls on their own lines. Calibrate any duration threshold from the business's own call log. Roster someone for the missed calls. Recheck: the call count in the report reconciles with the phone system for the same week.
The forwarding numbers and website call tracking belong to the call tracking guide. So does the move away from call only ads.
Check 10: What happens after the enquiry?
Symptom: enquiries arrive and the business still calls the campaign a failure. Trace each one from receipt to outcome.
Look for:
spam queues
broken CRM integrations
unassigned leads
missed calls
appointments that are weeks away
follow up that happens on Thursday for a Monday enquiry
A working form and an accurate tag prove nothing about the sales process behind them.
In about 2 in 5 cases we've diagnosed, the first confirmed fault sat in measurement or lead handling. It was not in keywords, bids or the page.
Record one consistent qualified lead definition and the reasons for rejection, without putting personal data in the record.
The evidence calls for two separate fixes: targeting for the five and handling for the six. It doesn't establish that the page failed or that all six would have bought.
If the enquiries are genuine and the handling works, the last question is commercial. Compare the cost of acquiring a customer with the contribution that customer leaves after fulfilment. Check the team can actually do the work.
A cheap enquiry is not automatically valuable business. A campaign that sells work at a loss is working perfectly. So is one that sells more than the crew can deliver. Both still need an offer, operations or spend decision rather than a campaign fix.
The return calculation and the lifetime value model belong to our guide to Google Ads ROI. Recheck: every enquiry in the period has an owner, a response time and an outcome.
Understand how mistakes combine
An account that "does not work" may have no single villain, just several small faults that feed each other.
That's an example mechanism, not proof of any account's cause. The last step is why the stack compounds. Once a button tap counts as a lead, automated bidding goes looking for more people who tap buttons.
Not every broad keyword or homepage builds that stack. Confirm each layer with its own check before repairing it. Keep the proven failures separate from the plausible explanations in the write up. The second kind is where budgets go to be wasted twice.
Our free audit is a real look at your account by the people who would run it. You get a short, prioritised list of what to fix first, whether you work with us or not.
Choose what to fix, test or pause
Priority follows evidence and harm, not a fixed order.
Decision | Examples | Recheck |
|---|---|---|
Fix or limit exposure now | Unavailable destination, failed submissions, unsupported promise, unauthorised change, unsuitable spend | Repair and test the affected route the same day |
Investigate the observed constraint | Low delivery, delayed outcomes, poor quality, fragmented spend | Verify the cause and the expected commercial effect before expanding |
Test an uncertain improvement | An ad, page, bidding or reach hypothesis | Comparable traffic, a useful outcome, an affordable ceiling, a review point and a rollback |
Pausing the affected spend is the right call when the route is broken. The same goes when the offer cannot be fulfilled or the agreed loss limit is reached. The money already spent is not a reason to keep spending. The account's history does not vanish when a campaign is paused.
Waiting is the right call when the tracking works and the outcomes are still maturing. That only holds inside a plan the business can afford.
Allow 4 to 8 weeks for a first read and about 90 days to judge cost per lead. Allow longer when the sales cycle is long. Our guide to how long Google Ads takes to work explains the timing. A confirmed break does not wait for either review point.
Working through these checks is an initial screening, not a resolution. Write the unresolved checks down with the date they will be revisited.
Get the right help for the remaining problem
Three outcomes follow the screening. If the checks found a fault the business can repair, repair it and recheck. Our free scorecard is a guided self assessment rather than an account audit. It can show which part of the setup is weakest before you start.
If the checks found a constraint whose cause is still unclear after gathering evidence, that's a specialist diagnosis. It starts with scoped read access granted through the account's own user management. It never starts with a login handed over.
Our ongoing PPC management begins with an audit. Access is set up in the business's own name. Then a plan is approved before any money is spent.
If the diagnosis is done and the business wants the repairs built and watched, that is managed implementation.
No account needs a rebuild because a guide said so. It needs the repairs the evidence names, and someone who reads the reports the following week.
What we recommend at Elev8d
We start with the record, not the settings. Before anyone touches a bid, we write the change timeline and send a labelled test enquiry. We also make a test call to the number in the ads. If the enquiry never reaches a person, no keyword or bid change will help.
A confirmed break gets fixed first, the day it is found. After that, one change goes in at a time on the route the symptom points to. Each gets a review date and a way to roll it back.
Two situations change that order. A campaign in its first weeks keeps its structure until the first read, unless something is broken.
When every check passes but the economics still fail, we stop tuning. No amount of keyword work fixes an offer the market won't pay for. Nor does it fix a crew with no room for more jobs. That's a conversation about the offer, capacity or budget. Our guide to whether Google Ads is worth it for a small business has the maths.
Next steps
Name the symptom and date it
Decide which of the five complaints it is and the date the data changed. Fill in the change timeline before touching a setting.
Run the route, not the whole list
Follow the checks the router sends you to, gathering the evidence each one names. Record what was confirmed and what was only plausible.
Sort and date every finding
Put each finding in one row of the priority table. Then give it an owner and a review date.
