For years, the daily ritual of a Local Services Ads (LSA) advertiser was oddly familiar: open the dashboard, scroll the recent leads, and flag the ones that did not belong — the wrong-trade calls, the out-of-area inquiries, the obvious spam. Each flag was a manual dispute, reviewed by Google, and, if approved, credited back to your account. It was tedious, but it was a lever you could pull.
That lever is gone. Around July–August 2024, Google retired the manual dispute workflow and replaced it with a machine-learning system that assesses lead validity automatically. If you have not adjusted how you think about lead credit since then, your budget is almost certainly being managed against an outdated mental model. This article explains what actually changed, what the new system does and does not credit, and how to operate profitably inside it.
What replaced the dispute button
The new system has two parts. First, an automated model evaluates each charged lead for validity behind the scenes. Based on third-party accounts of the rollout, that assessment typically lands within roughly 72 hours, and any credit that results is applied to your account within about 30 days. You are no longer filing individual claims and waiting for a human reviewer; the evaluation happens whether or not you do anything.
Second, Google added a "Rate this lead" survey. Instead of a binary dispute, you are asked to give structured feedback on lead quality. That feedback is not a dispute in the old sense — it is signal. It helps Google's model understand what a good lead looks like for your business, and over time it shapes both crediting decisions and, indirectly, the leads you are matched with.
The practical shift is from adversarial to informational. You are not arguing a case anymore. You are training a system.
What the model will and will not credit
Not every bad-feeling lead is a creditable lead, and this is where many advertisers lose money to false expectations. In broad terms, the categories that tend to qualify for credit are the ones that represent a genuine mismatch with what you signed up to receive.
| Often creditable | Generally not creditable |
|---|---|
| Job type outside your listed services | A qualified lead who simply did not book |
| Geographic mismatch outside your service area | Price objections or "just shopping" callers |
| Spam and clearly non-genuine contacts | Leads you failed to answer or follow up on |
Two verticals sit outside the standard crediting flow entirely: healthcare and tax-related services operate under different rules. If you are in one of those categories, do not assume the general guidance here applies to your account.
The headline reality is modest. Independent estimates put recoverable spend — the share of your LSA charges that legitimately qualify for credit — at roughly 6–7% for a well-run account. That is meaningful money over a year, but it is not a rescue plan for a campaign with a lead-handling problem. Credit recovery trims waste at the edges; it does not fix the middle.
Why "recoverable" is smaller than "annoying"
The gap between how many leads feel bad and how many are actually creditable is the core trap of the new era. Consider that a large share of raw LSA leads — third-party estimates put it near 45% — never turn into a booked job. But "unbookable" and "creditable" are different sets. A caller who was a perfect fit but chose a competitor is unbookable and not creditable. A caller who wanted a service you do not offer is both.
Chasing credit on the whole 45% is a losing game that wastes your time and pollutes the model with noise. The disciplined approach is to feed accurate signal on the genuinely invalid leads and pour your real energy into converting the bookable ones.
How to operate inside the ML system
Three habits separate advertisers who thrive under auto-credit from those who quietly bleed budget:
- Respond to every lead survey with honest, consistent signal. The model learns from your feedback. Sloppy or emotional ratings degrade its accuracy for your account.
- Tighten your service and geo settings. Many "mismatch" leads are self-inflicted — a service area drawn too wide or job types left too broad. Fixing the inputs prevents bad leads instead of crediting them after the fact.
- Track credits as a trailing metric, not a real-time one. Because assessment takes days and credits post within about a month, your daily spend view will always look worse than your true net cost. Reconcile monthly.
There is also a speed dimension. Because responsiveness and speed-to-lead are widely understood to influence LSA performance, the fastest fix for "bad" leads is often not a credit request at all — it is answering more of the good ones before a competitor does.
The strategic takeaway
The end of manual disputes is not a downgrade; it is a change in where your leverage lives. Under the old system, the advantage went to whoever was most diligent about clicking the dispute button. Under auto-credit, the advantage goes to whoever gives the cleanest signal, keeps their targeting tight, and converts the leads worth converting. The work moved upstream, from arguing about charges to preventing bad matches and closing good ones.
That is harder to do by hand, because it requires watching every lead, rating it consistently, and continuously tuning targeting — not once a month, but continuously.
Frequently asked questions
Can I still manually dispute LSA leads?
No. Around July to August 2024 Google retired the manual dispute workflow and replaced it with a machine-learning system that assesses each charged lead automatically, typically within about 72 hours, with any resulting credit applied within roughly 30 days.
Which LSA leads does the auto-credit system actually credit?
Generally genuine mismatches: a job type outside your listed services, a location outside your service area, and clear spam. It generally does not credit a qualified lead who simply did not book, price shoppers, or leads you failed to answer. Healthcare and tax verticals sit outside the standard flow.
How much LSA spend can I expect to recover?
Independent estimates put recoverable spend around 6 to 7 percent for a well-run account. That is meaningful over a year but not a rescue plan; credit recovery trims waste at the edges, while tighter targeting and faster response fix the middle.