Why Google Ads is not working: the causes we check first

Google Ads rarely fails for one dramatic reason. It fails quietly, through a handful of small settings that each drain a share of the budget and, together, make the channel look hopeless. The account keeps spending, the clicks keep coming, and the phone stays quiet.

This article walks through the causes we check first when an owner asks why Google Ads is not working: tracking, search terms, landing pages, budget, bidding, campaign types, targeting and the offer itself. For each one you get the symptom you would see in the account, a way to check it yourself, and what the fix usually looks like.

You do not need to be a specialist to run these checks. Most of them take fifteen minutes with access to the account and to your analytics. The point is to find out whether the channel is broken or the setup is, because those two problems have very different answers.

Conversion tracking: missing, broken or counting twice

Every other diagnosis depends on this one. If the account does not know which clicks turned into a lead or a sale, automated bidding optimizes toward noise, reports show numbers nobody trusts, and you end up judging the channel by gut feeling.

Symptoms in the account. Conversions sit at zero for weeks while your inbox shows inquiries that mention Google. Or the opposite: conversions look wonderful, but sales cannot find the customers behind them. A third pattern is a conversion rate higher than your site could plausibly deliver, which usually means a page view or a button click has been labeled as a conversion.

How to check.

• Open the conversion actions list and read what each action actually measures. A page view or a button click counted as a primary action is a red flag.
• Submit a test inquiry through your own form and confirm it shows up in the account within a day.
• Compare the monthly count in Google Ads with the leads your CRM attributes to paid search. A gap of a few percent is normal; a gap of half is not.
• Look for the same event imported twice, once from the website tag and once from GA4.

The fix. Keep one primary conversion per real business outcome (a qualified form, a call longer than a minute, a purchase), demote everything else to secondary, and give the account at least a month of clean data before you judge results or change the bidding strategy.

Search terms and match types: where the budget leaks

Broad match without a negative keyword list is one of the most frequent reasons a small account burns through its budget on searches that have nothing to do with the business. The platform interprets a broad match keyword generously, so a phrase like “commercial roofing company” can trigger ads for roofing jobs, roofing videos and another company's brand name.

Symptoms in the account. Plenty of clicks, a weak click-through rate on the terms you care about, and a search terms report full of queries you would never pay for: job seekers, “free”, “DIY”, unrelated cities, products you do not sell.

How to check. Open the search terms report for the last thirty days, sort by cost, and read the first fifty rows. Mark each one as “would pay for it” or “would not”. If more than a quarter of the spend sits in the second group, match types are the problem, not the channel.

The fix.

• Move the terms that convert into exact and phrase match, and keep broad match only in campaigns with enough conversion data for the algorithm to learn from.
• Build a shared negative list from the search terms report and add to it every week; a mature account keeps collecting negatives for months.
• Split by intent: long-tail keywords with buying words (“near me”, “quote”, “pricing”) deserve their own ad group and their own page.

The landing page is part of the campaign

An ad can only earn the click. What happens in the next eight seconds is decided by the landing page, and this is where many accounts lose the money they spent well in the auction.

Symptoms. Click-through rate is healthy, cost per click is reasonable, and almost nobody converts. Bounce rate on the paid page is far above the site average. Mobile conversion rate is a fraction of desktop.

How to check.

• Click your own ad on a phone over cellular data and time the load. Anything past three seconds costs you a share of visitors before they read a word.
• Compare the headline of the ad with the headline of the page. If the ad promises “emergency HVAC repair” and the page opens with the company story, the visitor has to work to find the match.
• Count the steps to the goal: a form with twelve fields, a phone number hidden in the footer, or a demo button that leads to a calendar for next month all push conversions down.
• Check whether the page is the home page. Sending paid traffic to the home page is a common shortcut and a common cause of weak results.

The fix. One page per intent, the ad's promise repeated in the first headline, one obvious action, proof that a stranger would believe (real photos, the process, what happens after the form), and a form short enough to finish on a phone. Speed fixes are usually image compression, fewer scripts and faster hosting, in that order.

A budget too thin for the auction, or spread too wide

Google Ads is an auction, and an auction has a price of entry. In many categories a daily budget that covers only a handful of clicks cannot buy enough data for anyone, human or algorithm, to learn from. The account looks alive, but it is guessing.

Symptoms. Campaigns show “Limited by budget” for most of the month. Search impression share sits below a fifth, and the share lost to budget is larger than the share lost to rank. Twelve campaigns share a budget that would be modest for two.

How to check. Take your monthly budget, divide it by the average cost per click in the account, and you get the number of clicks you can afford. Then apply an assumed conversion rate, say one visitor in twenty. If the math gives you fewer than fifteen to twenty conversions a month per campaign, the setup is starving.

The fix.

• Consolidate. Fewer campaigns with more data each beat many campaigns with a trickle. Start with the one product or service that has the shortest sales cycle and the clearest margin.
• Narrow instead of shrinking: fewer keywords, a tighter geography, business hours only, and the budget concentrated where the closed deals come from.
• If the affordable click count is still too low after narrowing, the honest answer may be a different channel for now, and a return to search when the budget grows.

Bidding strategy switched too early

Automated bidding strategies such as target CPA or target ROAS work well once they have something to learn from. Switched on in a new account with a dozen conversions, they either stop spending or spend on the wrong clicks, and the owner reads that as proof of why Google Ads is not working for the business.

Symptoms. Spend collapses the week after the strategy change. Cost per click doubles with no change in competition. The cost per acquisition target sits far below anything the account has ever achieved, so the system bids cautiously and the campaign goes quiet.

How to check. Open the change history and line up the strategy change against the performance chart. Then count conversions in the thirty days before the switch. Fewer than thirty is thin; fewer than fifteen is a guess.

The fix. Go back to a simpler strategy (maximize clicks with a bid cap, or maximize conversions without a target) until the account collects a steady stream of real conversions. When you do set a target, start at the level the account actually delivers and tighten it in small steps every two weeks. Every change resets part of the learning, so batch your changes instead of tweaking daily.

Performance Max, geography and the ad schedule

Three settings that look harmless on the surface quietly redirect the budget away from the searches that were working.

Performance Max cannibalizing search. A Performance Max campaign will happily serve on your brand name and on your exact keywords, because those are the easiest conversions to claim. Symptom: search campaign volume drops the month PMax launches while total conversions barely move. Check the search terms insight in PMax and the brand share of its conversions. Fix: add brand exclusions, keep your converting exact keywords in search, and judge PMax on the conversions the account did not have before it launched.

Geography. The default location setting includes people who showed interest in your area, not only people in it. Symptom: clicks from cities you cannot serve, or from another country. Check the locations report by user location. Fix: switch to presence targeting, exclude regions you cannot serve, and set a radius around the locations that matter for a local business.

Schedule. Ads that run at 3 a.m. for a business that answers calls at 9 a.m. pay for leads nobody follows up on. Check the hour of day report against your own response records. Fix: a schedule that mirrors when you can actually respond, with bid adjustments rather than a hard cutoff, so that after-hours forms still get answered the next morning.

The offer, the pricing page and the calendar

Some accounts are set up correctly and still disappoint, because the problem sits outside the account. Paid search brings people who are ready to compare, and it shows you quickly whether the offer holds up under comparison.

Symptoms. Visitors reach the pricing or quote page and leave. Leads arrive but half of them go cold after the first call. Rivals in the results show a concrete promise (same-day service, a fixed package, financing) and your ad shows a company name.

How to check. Search your own main keywords in a private browser window and read the ads around yours as a customer would. Then ask sales what the last ten paid leads objected to. Look at the path report in analytics for the pricing page: entries, exits and where people go next.

The fix. Usually not in Google Ads at all: a clearer promise in the headline, a pricing page that explains what changes the number instead of hiding it, a quote process that responds within the hour, and a follow-up sequence for leads that are not ready yet. Remarketing to visitors of the pricing page is an inexpensive way to stay in front of people who compared and left.

The calendar. Expecting a verdict in days is its own cause of failure. A new search campaign needs a few weeks of data before the search terms, the bids and the pages settle. Judge it on a 90-day window, with a checkpoint every two weeks to catch leaks, not to rewrite the plan.

What a specialist does in the first two weeks of a takeover

When an experienced specialist takes over an account that has been written off, the first two weeks look less like optimization and more like an audit. The sequence matters, because adjusting bids before tracking is fixed only makes the numbers lie faster.

1. Days one to three: tracking. Rebuild conversion actions around real outcomes, remove double counting, connect the CRM if there is one, and record a baseline nobody will argue with later.
2. Days four to seven: the money map. Search terms for the last 90 days, negative lists, match types, geography, schedule and device splits, with a written list of where spend went that should not have.
3. Week two: structure and pages. Consolidate campaigns, separate brand from non-brand, fix the ad-to-page match on the two or three pages that receive most of the traffic, and pause everything the audit could not justify.
4. End of week two: the plan. A one page note for the owner: what was broken, what was changed, what the account should be judged on, and the date of the next review.

That last document matters more than the changes, because it turns “why Google Ads is not working” into a short list of things that were fixed and a smaller list that still needs a decision from you. If you would rather not run this audit yourself, Google Ads management inside an outsourced marketing department begins with exactly this two-week sequence, with a senior project manager as your single point of contact and a monthly report that puts spend next to results. The marketing KPIs by channel guide covers which numbers belong in that report.

FAQ