Request an audit

See if they name you.

Spilld measures how often AI assistants recommend your clients' repair shops, which shops they name instead, and the exact answer behind every finding.

Inside the audit

The answer behind every finding.

See when AI names a competitor, when it overlooks your client, and the answer that produced the result, ready to put in front of a prospect.

Illustrative · web search on

“Who’s a good independent shop for brakes in Millbrook?”

For brake work in Millbrook, drivers often mention name withheld on Route 9 and name withheld near the town square. Both are independent and list brake inspections on their websites. Maple Street Auto also comes up, but I couldn’t find enough recent information to say how it compares.

Rival named

Counted as Mentioned: named, but not recommended.

Illustrative: the format of a stored answer. The shop is invented; other names withheld. A real report names them.

Why agencies call

One measurement. Three ways to use it.

Win the account

Open the pitch with the exact answers AI gives about a prospect's shop, and the rivals it names instead.

Set the baseline

A dated snapshot of how often AI assistants recommend your client today, with the stored answer behind every finding.

Name the rivals

See which shops AI names instead of your client, and how often.

How it works

Asked like a customer. Counted like an auditor.

  1. We ask what drivers ask

    Up to 97 questions, written the way drivers ask them (“best brake shop near me,” “my car's overheating, who do I call”), put to 4 major AI assistants: ChatGPT, Claude, Gemini and Perplexity. Every question is asked with web search off and on. ChatGPT, Claude and Gemini answer from memory and again after searching the web; Perplexity always searches, so it answers with search on only. Every question is asked 3 times in each mode. Hundreds of stored answers per audit.

  2. Every answer gets a verdict

    Each answer is marked Recommended, Mentioned or Not found. A shop counts as Recommended only when the answer endorses it by name. We keep every answer word for word. When an assistant's repeat answers split with no majority, the result is marked No majority. It stays in the count, but it is not counted as recommended.

  3. A gate that can say no

    Every audit passes a quality gate before delivery. If the gate can't stand behind the numbers, the audit is held instead of delivered.

The full pipeline, step by step →

AI visibility, built around how drivers choose a repair shop.

Spilld measures whether AI recommends your shop when drivers look for automotive service, from “Who do you recommend for brake repair?” to “Who is the most honest mechanic?”

Built specifically for auto repair shops, by a founder who has spent years around them and their data.

  • Service-specific questions

    See whether your shop appears for the repair and maintenance services you offer.

  • Local competitor comparisons

    See which nearby shops AI names for the same questions.

  • Local shops and national chains, shown separately

    Understand who appears in your market.

See what AI tells your clients' customers.

Pricing on request