Your Google Ads did not stop working for one reason. They stopped working in one of four distinct ways, and the fix for each one is close to the opposite of the fix for the others. Raising your budget solves the third problem and makes the second one worse. Rebuilding the campaign feels productive and is almost always the most expensive thing you can do.
The fastest way to tell them apart takes about ten minutes and one number you already have.
Start here: which of the four failures do you actually have?
Open your Google Ads account, set the date range to the last 30 days, and compare it to the previous 30 days. Write down two percentages: how much clicks fell, and how much conversions fell. Then read the row that matches.
| What you see | What actually broke | First move |
|---|---|---|
| Impressions and clicks near zero | Serving problem: billing, policy, verification or budget | Check account status and billing before anything else |
| Clicks flat, leads collapsed | Conversion tracking or the landing page | Place a test call and a test form fill yourself |
| Clicks down, leads down by a similar percentage | The auction repriced, or demand genuinely dipped | Check auction insights and CPC trend, not settings |
| Clicks down, leads down much further | The traffic mix changed underneath you | Pull the search terms report and the network breakdown |
That last distinction is the one nearly everybody misses. A Reddit analyst working through a struggling account laid it out cleanly: clicks fell 27 percent while conversions fell 43 percent, and as they put it, "If August were only the seasonal thing people are describing, fewer buyers looking but the same kind of traffic, those two fall together. They didn't." Seasonality moves clicks and leads roughly in step. When leads fall faster, the traffic that arrived was different traffic.
Tip
There is a shortcut for reading conversion rate off a report that only shows CPC and CPA. Divide CPC by CPA and you get the conversion rate for every row. It works because CPA is just CPC divided by conversion rate. This lets you spot the real break in a monthly summary table without exporting anything.
Failure 1: the ads genuinely stopped serving
If impressions went to near zero, nothing about your keywords, copy or landing page matters yet. Something switched the account off.
Google's own troubleshooting guide lists account suspension, billing holds, unconfigured payment methods and security reviews first, and for good reason. A security review can produce zero impressions or a temporary daily spending limit with no dashboard notification at all. Google is explicit that these automated reviews cannot be expedited and that no specific alert appears when one clears.
For contractors there is a sneakier version of this. Local Services Ads and standard Google Ads are two different products with two different eligibility systems. LSAs charge per lead rather than per click and require a passing background check plus current license and insurance on file. When your insurance certificate expires or a background check lapses, your LSAs pause while your standard search campaigns keep running normally. Owners see the phone go quiet, check their search campaigns, find them healthy, and conclude Google broke something.
Run this list in order before touching a campaign setting:
- Account status and billing. Expired card, failed payment, or a hold.
- Policy manager. Disapproved ads or disapproved assets in an otherwise enabled campaign.
- Campaign end dates. A campaign with an end date that quietly arrived.
- LSA verification, separately. Background check, license expiry, insurance expiry.
- Ad schedule. Somebody restricted hours and forgot.
- Daily budget versus account spending limit. These are two different caps.
Failure 2: clicks are normal but the leads vanished
This is the most common version of "my Google Ads stopped working," and the diagnosis usually takes two minutes: call your own tracking number from a phone that has never called you, and submit your own form. Most owners who do this find the problem on the first try.
It matters far more than it looks, because broken tracking does not merely hide leads from your report. It destroys real leads, on a delay.
Pete Bowen, a Google Ads consultant, documented a case worth understanding in full because the same sequence plays out constantly in the trades. A campaign producing about 25 leads a day dropped to one or two a day and did not recover. Target CPA was $50, actual cost per lead about $43. The change history showed nothing: no budget change, no bid strategy change, no targeting change. The client insisted the landing page was untouched.
Pressed harder, they remembered one small thing. A few days before the drop, a department manager who disliked receiving calls through a shared call centre had IT replace the call tracking number on the landing page with her department's direct line. That single edit broke call conversion tracking. Form fills and the click-to-call button still worked, so it looked fine. And because conversions dropped about four days after the number change, the client had already ruled it out as the cause.
Why the algorithm makes it worse instead of better
Here is the part that turns a reporting gap into a revenue emergency. A conversion-based bid strategy asks two questions every time your ad could show: should we show this person an ad, and how much should we bid. It answers both from its estimate of how likely that person is to convert, and conversion tracking is the only feedback telling it whether it guessed right.
Bowen describes bidding algorithms as running in two modes. Exploit mode is when it knows what works and spends your budget on it, which is what a steady flow of leads looks like. Explore mode is when it does not know, so it hunts, and it bids lower per click because it cannot predict the outcome.
When the tracking broke, the algorithm read a wall of failures, concluded that what it had been doing no longer worked, and dropped out of exploit into explore. Lower bids meant ads showed less often and lower on the page, so fewer people saw the form. And since generally only the top ad position carries a click-to-call button, even the calls still being tracked dried up. That is why all the conversions fell, not just the ones the broken number was supposed to record. The lag is not a bug either: it takes the algorithm a while to learn something works, and the same while to learn it stopped.
Watch out
In a home services account the phone is the conversion, and the phone number is the single most fragile piece of your tracking. Any of these will break it without breaking your website: a new site launch, a designer "cleaning up" the header, an office manager who prefers the real number, a Google Business Profile edit, or a non-www to www redirect that strips the tracking parameter. Check GA4: if Google CPC traffic fell while Direct or Unassigned traffic rose by a similar amount, your tracking is leaking, not your demand.
If your ads are working and your tracking is not, you are paying full price for leads you cannot see and teaching Google to stop finding them. We build the conversion page, the qualifying form and the lead-to-sale tracking as one connected system, so a broken number shows up in a report the same day instead of six weeks later.
Failure 3: clicks and leads fell together, so the auction repriced
If both numbers fell by roughly the same percentage, your account probably did not break. Your market got more expensive, or fewer people searched.
The 2026 benchmark data makes the first part uncomfortable to read. Cross-industry average search CPC hit $2.96 in the first quarter of 2026, up 12 percent from $2.64 a year earlier. Cost per lead moved from $62.50 to $70.11 over the same period, also 12 percent.
Local service categories took it worse than average:
| Category | 2026 avg search CPC | Year over year change |
|---|---|---|
| Consumer services | $6.40 | +18% |
| Home goods | $2.94 | +7% |
| Industrial services | $2.56 | +5% |
| All industries | $2.96 | +12% |
Consumer services, where most home service advertisers land, posted the steepest CPC inflation of the fifteen categories tracked. WordStream's 2026 numbers put home and home improvement specifically, meaning HVAC, plumbing, roofing and electrical, at $8.33 per click, against roughly $7.85 in LocaliQ's 2025 home services analysis.
Read that against your own account before you fire anybody. If your cost per lead rose less than 18 percent this year, your campaign management got better while your market got more expensive. A flat budget simply buys meaningfully fewer clicks than it did twelve months ago, which shows up as ads that "stopped working."
Typical 2026 cost per lead lands around $45 to $85 for HVAC and $40 to $75 for plumbing, while roofing frequently runs past $200 because job sizes are large and storm-chasing crews flood the auction seasonally. If your number sits inside that band, the ads are behaving. Whether they are profitable is a different question that comes down to your own math, not the benchmark. Our breakdown of what a good cost per lead actually is for contractors works it out from job gross profit and close rate instead.
Failure 4: leads fell faster than clicks, so the traffic changed
When conversions fall significantly faster than clicks, a similar volume of people arrived and a worse-fitting group of them did. Four usual suspects, in order of how often they turn out to be the cause:
Broad match creep. Match types loosened, or close variants widened on their own, and your budget started buying queries you would never bid on deliberately. This is the first report to pull, always. One agency responding to a struggling account put it bluntly: open the search terms report and burn the junk with negatives immediately.
A new website or landing page. Clicks stay stable, conversion rate craters. As one practitioner noted on a thread about exactly this, small trust and friction changes wreck conversion rate fast on high-ticket work even when traffic looks identical: slower mobile load, a longer form, review badges and before-and-after photos that quietly disappeared in the redesign.
Performance Max absorbing your search budget. Multiple home services practitioners describe PMax as causing more confusion than value in this vertical unless you already have strong conversion data feeding it, because it blends cheap Display and video impressions into what was a clean search campaign.
Non-human clicks. Worth understanding, worth not panicking about. A bot-protection vendor publishing its own detection data, restricted to clicks it confirmed objectively rather than merely flagged as suspicious, reported roughly 13 percent bot traffic on Google Search and 27 percent on Google Display. Treat a vendor's own numbers with appropriate skepticism, but the directional point matches what trade advertisers report. On the same home services thread, a practitioner described click attacks arriving through residential proxies, which look like real users on real ISP connections and slip past Google's invalid click filters. Their note on which keywords get hit is specific: emergency plumbing, HVAC and locksmith terms take the worst of it, because the CPCs are highest and the attacker does not need to convert anything, only to drain your daily budget so their own ad runs unopposed the rest of the day. A cabinet painting contractor on that thread described the symptom exactly: heavy clicks between 7am and 8:30am every morning, then nothing, and no leads from any of it.
The practical response is unglamorous. Exclude the Display Network and Search Partners from search campaigns, review the IP exclusion list, and check your hourly breakdown for a spend pattern that does not match when your customers actually call.
The three things not to do
Owners in a panic reliably reach for the three moves that make recovery slower.
Do not rebuild the campaign. A new campaign discards every conversion the old one accumulated and re-enters the learning phase, which means roughly two weeks of near-zero leads before it is even as good as the thing you deleted. As one commenter told an account manager considering exactly this: blow it up and start fresh tomorrow and you will get zero leads for two weeks, and that is what actually gets you fired. Another added the cleaner rule: do not rebuild something whose conversion rate doubled.
Do not lower your target CPA to force costs down. It feels like tightening the screws. What it does is instruct Google to skip auctions it cannot win at your target, so delivery quietly dries up and ads stop showing rather than getting cheaper. In the case study above, the fix included deliberately raising target CPA from $50 to $100 to shorten the relearning period, then walking it back down once leads were flowing.
Do not just add budget. More budget on broken tracking buys more untracked clicks and teaches the algorithm the same wrong lesson faster. Budget is the answer to failure 3 and nothing else.
The recovery protocol
Once you have found and fixed the actual break, run this sequence.
- Fix the input first. Restore call tracking, republish the landing page, clear the disapproval, renew the insurance certificate. Nothing downstream works until the signal is real again.
- Verify it end to end yourself. Place a real call from an unknown number and submit a real form. Confirm both land in Google Ads, not just in your inbox, and wait the 24 to 48 hours conversions take to appear.
- Bridge with a volume-based bid strategy. If the account went days with no recorded conversions, smart bidding has no ground to stand on. Switch to Maximize Clicks with a CPC cap for seven to ten days, then return to conversion bidding once roughly 10 to 15 real leads are logged.
- Loosen the target, then tighten slowly. Give the algorithm room to relearn, then bring the target back in small steps rather than one jump.
- Clean the search terms report weekly during recovery. This is when junk queries do the most damage, because the algorithm is exploring and will happily explore into the wrong audience.
- Then look at the offer and the page. Once delivery is stable, conversion rate is the lever you actually control. CPC is set by an auction you do not run.
Expect seven to fourteen days. The documented case above went from one or two leads a day to eleven, nine and ten on the three days following the fix, but that account had a long history for the algorithm to fall back on.
By the numbers
Home and home improvement search advertising now averages $8.33 per click, and consumer services CPC rose 18 percent year over year, the steepest increase of any category tracked in 2026. At those prices a two-week tracking outage is not a reporting inconvenience. On a $2,000 monthly budget it is roughly $900 of spend teaching Google the wrong lesson about who your customers are.
The possibility nobody wants to hear
Sometimes the ads never stopped working. The follow-up did.
An agency writing about home services put it plainly: owners who lose money on Google Ads usually do not have an ad problem, they have a follow-up problem. Leads arrive, nobody answers fast enough, the homeowner books the next contractor on the list, and the owner concludes Google Ads does not work. A practitioner managing search for home services accounts summarized the same idea from the campaign side: most accounts do not fail because of keywords, they fail because calls are not tracked properly, leads are messy, or landing pages dilute urgency.
There is a specific test. If your close rate on inbound leads sits below 40 percent, audit what happens after contact before you touch your ad copy. Pull last month's leads and check how long each one waited for a first response. In a trade where the customer has water on the floor, an hour is not a delay, it is a loss.
This also explains a pattern that confuses owners: ads that "worked fine last year" on the same budget and the same settings. Nothing in the account changed. The dispatcher who used to answer on the second ring left, or volume grew past what one person can catch, and the leads still arrive and still go cold. Our guide to tracking where your leads actually come from covers the manual version of this audit.
If that is the shape of your problem, more ad spend is the most expensive possible fix. The lead generation system we build is designed around the handoff rather than the click: a qualifying form that arrives with the answers already attached, so the first call back is a quote conversation instead of a discovery call.
The ten-minute checklist
Work top to bottom. Stop at the first thing that is wrong, fix it, and wait 48 hours before continuing.
- Account status: any billing hold, suspension or security review?
- Policy manager: any disapproved ads or assets?
- Campaign end dates and ad schedule: anything expired or restricted?
- LSA eligibility separately: background check, license, insurance all current?
- Place a test call from an unknown number. Did it record as a conversion?
- Submit a test form. Did it record as a conversion?
- GA4: did Google CPC traffic fall while Direct or Unassigned rose?
- Change history: any edits in the ten days before the drop, including website edits?
- Search terms report: what percentage of last month's spend went to queries you would not have chosen?
- Network breakdown: is Display or Search Partners inside a campaign you thought was search only?
- Hourly report: does spend cluster at hours your customers do not call?
- Auction insights: did a new competitor show up at scale?
Every one of them is free and takes a few minutes, and none require asking your agency for anything. That is the point. The most expensive Google Ads problems in the trades are not sophisticated, they are invisible, and they stay invisible precisely because the account keeps reporting clicks the entire time it is quietly failing.