A documented Local Services Ads failure offers a warning for every small business that depends too heavily on a platform it cannot inspect, control or compel to explain itself.*
I am not angry that an advertising campaign had a bad stretch.
Advertising fluctuates. Competitors change their bids. Customers behave differently from one week to the next. Algorithms adjust. Reports lag. Shit happens.
What made this incident a line in the sand was not simply that a local home-services company suddenly received fewer leads. It was that a healthy, funded Google Local Services Ads account suffered a documented delivery collapse—including a day with zero impressions—and repeated requests for a technical explanation produced generic optimization advice, present-day ranking observations and repeated declarations that the case had been resolved.
It had not been resolved. The calls eventually began to recover. The historical failure was never explained.
That distinction matters because this company had become deeply dependent on Google for new business. Its Local Services Ads budget had grown to $2,030 per week, equivalent to a monthly maximum of approximately $8,816. In a small, competitive market, that was not experimental spending. It was part of the company’s operating bloodstream.
Had the collapse continued, the consequences could have become existential.
The break in the data
Before drawing conclusions, I went back through the primary evidence: Google reports, account screenshots, lead records, billing records and support correspondence. I separated completed reports from intraday screenshots and kept Google’s charged leads separate from the company’s actual incoming calls and messages.
The equal-period comparison was stark:
| Period | Impressions | Charged leads | Phone leads | Message leads | Spend |
|---|---|---|---|---|---|
| Aug. 21–28, 2026 | 2,651 | 29 | 21 | 8 | $2,598.04 |
| Aug. 29–Sept. 5, 2026 | 1,151 | 10 | 6 | 4 | $805.90 |
Across those two eight-day periods:
- Impressions fell 56.6%.
- Charged leads fell 65.5%.
- Phone leads fell 71.4%.
- Spend fell 69.0%.
The account did not merely spend the same amount and convert poorly. Google delivered far fewer impressions and spent far less of the available budget.
The clearest single data point came on August 31: zero impressions, zero charged leads and zero spend.
That does not, by itself, prove a Google system error. A zero-impression day can have more than one possible cause. But it is a specific delivery event that should be susceptible to a specific account-level review: Was the ad eligible? Was it entered into auctions? Was there a hold, pause, review, migration event, targeting problem, bidding limitation or system interruption?
Those were the questions we asked.
High placement when shown, but far fewer opportunities to show
Google support ultimately focused on ranking. The company was told that its advertisement was currently appearing at “rank 10” and that lower placement reduced lead volume. Google repeatedly recommended general optimization work: improve the profile, add photos, collect reviews and review performance guidance.
Competition and rank are legitimate considerations. Google’s own Local Services Ads ranking documentation says placement is determined by an auction that considers bid, profile quality, responsiveness, reviews and the context of a customer’s search.
But a current ranking observation is not a historical diagnosis.
During the healthy August 21–28 period, the account’s top impression rate was 93.82%. During the sharply weaker August 29–September 5 period, it was still 92.55%. That metric does not tell us why the account entered far fewer auctions or received far fewer impressions. It tells us that when the ad did appear, it was usually shown above organic results.
The account’s absolute-top rate did decline, from 28.26% to 23.98%, but that modest change does not independently explain a 56.6% collapse in impressions or a day with no impressions at all.
Google may have internal evidence showing that auction rank caused the delivery loss. If so, that evidence was not supplied to us. No historical rank, auction, eligibility, query-volume or impression-share analysis for August 31 was provided.
“Healthy account” followed by “but…”
One Google support response confirmed that the account was in excellent administrative standing, its specialized verifications were complete and it had no active policy violations.
Then came the “but”: lead volume was supposedly limited by low ranking relative to competitors.
That answer addressed the account’s current condition in broad terms. It did not explain the documented historical break. When we pressed for the date-specific evidence, the assigned representative later said Google did not have an option to identify lead and impression performance by day because ranking changes over time.
That admission should have narrowed the conclusion. If support could not reconstruct the daily historical conditions, the accurate response would have been: “We cannot determine the cause from the information available to us.”
Instead, the representative continued citing the ad’s current rank as the resolution, repeatedly closed or attempted to close the case and did not honor repeated requests for escalation to a more experienced technical reviewer. One response instructed us not to reply because a resolution had been provided.
The issue was not that Google refused to guarantee a certain number of leads. We never requested a guarantee. The issue was that present-day information and generic optimization guidance were repeatedly substituted for a historical technical explanation.
Whether Google support was unable to investigate or unwilling to escalate, the practical result for the advertiser was the same: no answer.
Two major Google changes occurred near the incident
After the original case had been repeatedly closed, Google’s automated Ads support intake surfaced two technical areas that deserved specialist review.
First, Google officially began moving selected U.S. home-service advertisers—including the category involved in this case—from the legacy Local Services Ads system into Performance Max campaigns with pay-per-lead goals in August 2026. Google says advertisers are migrated in waves and may need up to two weeks for the process and performance to stabilize. Google also warns that legacy performance reports will not transfer into the new system. Google’s migration documentation
Second, Google made a global change to target-based bidding beginning August 17 and completed the rollout on August 27. Google’s documentation says qualifying campaigns that were limited by budget and used affected target-based strategies could experience performance volatility as bidding began optimizing more closely to their stated targets. Google’s bidding-update documentation
The timing is notable: the verified decline began immediately after that bidding rollout completed, with the zero-impression day occurring on August 31.
Timing, however, is not causation.
We do not yet know whether this account had entered the migration process, whether its configuration qualified for the bidding change, or whether either event affected delivery. The account continued to expose the legacy Local Services dashboard later in September, which weighs against assuming that a full migration had already occurred. Google’s automated intake also identified two active Local Services-related campaign entities, but it did not explain whether both were serving, whether one was a system representation of the other, or whether either changed during the incident.
Those are precisely the questions a technical specialist should be able to confirm or reject from the account’s change history.
They are not proof of a cause. They are concrete reasons the repeated “rank 10” answer was incomplete.
Recovery does not erase the failure
The company’s impressions and calls eventually improved. That is good news operationally, but it does not retroactively explain what happened.
The recovery also introduced another complication: the lead mix changed. In a partial September 20–26 report, Google showed 13 charged leads—five phone leads and eight message leads. Message leads represented 61.5% of the charged leads and 55.2% of the spend in that snapshot.
That matters because a “lead” is not a uniform business outcome. A phone call from a homeowner with an urgent service need is not economically identical to a message requesting quotes from several businesses. Google provides one combined budget and bidding strategy; messages can be turned on or off, but the advertiser cannot instruct the system to devote a particular share of the budget to phone calls.
The company therefore had to begin tracking a second question: Were the increased message leads producing estimates, approved work and revenue—or merely consuming budget?
Without that connection to booked jobs, “lead volume recovered” would be an incomplete and potentially misleading conclusion.
The larger danger: platform dependence without platform accountability
This is where the story stops being about one home-services company.
Small businesses are routinely encouraged to build their reputations and customer-acquisition systems inside platforms they do not control. They accumulate reviews, complete verification processes, upload photographs, answer messages quickly, expand budgets and follow optimization recommendations. Over time, the platform becomes less like an advertising channel and more like infrastructure.
But it is infrastructure without the accountability businesses normally expect from infrastructure.
The advertiser cannot inspect the auction. It cannot audit the algorithm. It cannot see the historical internal conditions that determined eligibility and delivery. It cannot select a support engineer or compel escalation. When first-tier support repeats a generic answer and marks a case resolved, the small business has little practical leverage.
That asymmetry becomes dangerous when one company supplies most of the leads keeping another company alive.
Google did not force this business to become dependent on Google. That is an important part of the lesson. A platform can be highly productive and still be too risky to serve as a company’s single point of failure.
What small businesses should do before something breaks
This incident produced several practical lessons.
Diversify before you are desperate. Build referral systems, customer-retention programs, commercial relationships, organic search visibility, email and direct outreach while paid leads are still flowing. Diversification is slow; it cannot be switched on the morning a platform stops delivering.
Preserve your own evidence. Save regular screenshots or exports of impressions, lead types, spend, bidding settings and account status. Platform reports can backfill, interfaces can change and historical reports may disappear during migrations.
Maintain change control. Record who changed the budget, messaging, photos, service areas, schedules and bidding settings—and when. Multiple reactive changes destroy the ability to identify cause and effect.
Do not confuse platform metrics with business outcomes. A call-button click is not necessarily a completed call. A charged lead is not necessarily a booked job. A message is not necessarily a qualified opportunity. Connect every lead source to estimates, booked work, completed work and revenue.
Set internal warning thresholds. A sudden fall in impressions, spend or calls should trigger documentation and review before days become weeks.
Ask support for evidence, not reassurance. Request dates, statuses, change history, campaign identifiers and the data supporting any diagnosis. If support cannot provide that information, ask for that limitation in writing.
Where the matter stands
On September 28, after the original case had been repeatedly treated as resolved, a formal complaint was submitted through Google Ads Customer Service with six supporting attachments. Google opened a separate complaint case under its gTech Customer Experience workflow.
The complaint requests an independent review of the prior case handling, confirmation of the account’s migration and bidding history, and either a supported historical explanation or an honest written statement that Google cannot determine the cause.
As of this writing, that review is pending.
I am not asking Google to promise that advertising will never fluctuate. I am asking something more basic: when a business-critical advertising system abruptly stops delivering, and when the advertiser provides detailed evidence and asks precise questions, the case should not be called resolved until those questions are answered—or until Google plainly admits that it cannot answer them.
Calls may recover. The underlying accountability problem does not.
Reporting note: The company has been anonymized to protect the client. All figures come from preserved Google account reports, lead records and support correspondence. Intraday screenshots were not treated as completed daily totals, and Google’s charged leads were kept separate from actual customer contacts. No unverified theory is presented as the cause of the incident. Google’s review of the formal complaint remains pending as of September 28, 2026; this article will be updated if Google provides additional findings.