<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/how-spanner-graph-detects-google-ads-fraud-and-abuse-ix01apff8" -->

---
title: How Spanner Graph detects Google Ads fraud and abuse
description: Google moved its Ads fraud detection platform from offline MapReduce batch jobs to Spanner Graph, cutting multi-hop traversal latency from an hour to...
canonical: https://daily.dev/posts/how-spanner-graph-detects-google-ads-fraud-and-abuse-ix01apff8
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: How Spanner Graph detects Google Ads fraud and abuse | daily.dev
og:description: Google moved its Ads fraud detection platform from offline MapReduce batch jobs to Spanner Graph, cutting multi-hop traversal latency from an hour to...
og:url: https://daily.dev/posts/how-spanner-graph-detects-google-ads-fraud-and-abuse-ix01apff8
og:image: https://api.daily.dev/og/posts/ix01ApFF8.png
og:image:alt: How Spanner Graph detects Google Ads fraud and abuse
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# How Spanner Graph detects Google Ads fraud and abuse

**[Google Cloud](https://daily.dev/sources/gcp)** · 7 min read · 0 upvotes · 0 comments

## Summary

Google moved its Ads fraud detection platform from offline MapReduce batch jobs to Spanner Graph, cutting multi-hop traversal latency from an hour to sub-second or single-digit milliseconds. Because Ads data already lived in Spanner, the team layered a property graph directly on existing tables with no migration, and most graph-based abuse detection use cases transitioned within weeks. The platform now runs real-time ISO GQL queries, combining multi-hop pattern matching, full-text search and vector similarity search, across a live graph of over 500 billion nodes and 1 trillion edges. It detects coordinated fraud rings, click farms, account takeovers and repeat financial offenders. Google blocked or removed more than 8.3 billion policy-violating ads and suspended 24.9 million advertiser accounts in 2025.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://cloud.google.com/blog/products/databases/how-spanner-graph-detects-google-ads-fraud-and-abuse>

## Questions this post answers

### How does Google detect coordinated ad fraud rings in real time instead of batch processing?

Google modernized its ads abuse detection platform onto Spanner Graph, replacing offline Apache Flume MapReduce batch jobs that took hours with multi-hop graph traversals that run in sub-second to single-digit millisecond latency. Because the data already lived in Spanner, no migration was needed; engineers defined a property graph directly on existing relational tables and query it using standard ISO GQL.

_Teams weighing graph databases for fraud detection can track real-world architecture write-ups like this on daily.dev._

### What scale does Google's Spanner Graph fraud detection system operate at?

The platform operates as a unified, Google-wide defense system across a live graph of over 500 billion nodes and 1 trillion edges. It uses Spanner Graph's multi-hop pattern matching combined with full-text and vector similarity search to trace connections between flagged accounts, payment methods, and other entities in real time.

_Developers benchmarking graph databases at scale can follow case studies like this on daily.dev._

### Can you query a graph directly on top of existing relational tables without migrating data, using Spanner?

Yes, Spanner Graph lets you define property graphs directly on top of existing relational tables with zero ETL, since the graph layer co-locates with the underlying distributed data splits. Google's ads team adopted this approach and moved most use cases to Spanner Graph within weeks rather than months, using ISO GQL to query the same tables instead of custom traversal code.

_Developers deciding between a dedicated graph database and layering graph queries on relational data can track approaches like this on daily.dev._

## Similar posts on daily.dev

- [Fraud Prevention with BigQuery Graph](https://daily.dev/posts/fraud-prevention-with-bigquery-graph-ijwfemqda) · Google Cloud · 0 upvotes · 0 comments
- [Introducing Spanner Graph algorithms](https://daily.dev/posts/introducing-spanner-graph-algorithms-bpoqhdeun) · Google Cloud · 0 upvotes · 0 comments
- [Introducing BigQuery Graph](https://daily.dev/posts/introducing-bigquery-graph-tig40l393) · Google Cloud · 0 upvotes · 0 comments
- [Stop Token-Incinerating Your Risk Engine](https://daily.dev/posts/stop-token-incinerating-your-risk-engine-nn6d3oith) · neo4j · 0 upvotes · 0 comments
- [The Fraud Analyst AI Coworker: Detecting Telecom Fraud in Real-Time on One Platform](https://daily.dev/posts/the-fraud-analyst-ai-coworker-detecting-telecom-fraud-in-real-time-on-one-platform-rsnrgvqcc) · databricks · 1 upvotes · 0 comments

---

Tags: [#architecture](https://daily.dev/tags/architecture), [#gcp](https://daily.dev/tags/gcp), [#fraud-detection](https://daily.dev/tags/fraud-detection)

[View this post on daily.dev](https://daily.dev/posts/how-spanner-graph-detects-google-ads-fraud-and-abuse-ix01apff8)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"How Spanner Graph detects Google Ads fraud and abuse","url":"https://daily.dev/posts/how-spanner-graph-detects-google-ads-fraud-and-abuse-ix01apff8","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-spanner-graph-detects-google-ads-fraud-and-abuse-ix01apff8"},"datePublished":"2026-11-20T17:08:52.099Z","dateModified":"2026-11-20T17:09:21.143Z","description":"Google moved its Ads fraud detection platform from offline MapReduce batch jobs to Spanner Graph, cutting multi-hop traversal latency from an hour to...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1c524bba539f81ce8b18b169c88d0446?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1c524bba539f81ce8b18b169c88d0446?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Google Cloud","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Google Cloud","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/gcp","url":"https://daily.dev/sources/gcp"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-spanner-graph-detects-google-ads-fraud-and-abuse-ix01apff8","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"architecture,gcp,fraud-detection","timeRequired":"PT7M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Google Cloud","item":"https://daily.dev/sources/gcp"},{"@type":"ListItem","position":3,"name":"How Spanner Graph detects Google Ads fraud and abuse"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-spanner-graph-detects-google-ads-fraud-and-abuse-ix01apff8#faq","mainEntity":[{"@type":"Question","name":"How does Google detect coordinated ad fraud rings in real time instead of batch processing?","acceptedAnswer":{"@type":"Answer","text":"Google modernized its ads abuse detection platform onto Spanner Graph, replacing offline Apache Flume MapReduce batch jobs that took hours with multi-hop graph traversals that run in sub-second to single-digit millisecond latency. Because the data already lived in Spanner, no migration was needed; engineers defined a property graph directly on existing relational tables and query it using standard ISO GQL. Teams weighing graph databases for fraud detection can track real-world architecture write-ups like this on daily.dev."}},{"@type":"Question","name":"What scale does Google's Spanner Graph fraud detection system operate at?","acceptedAnswer":{"@type":"Answer","text":"The platform operates as a unified, Google-wide defense system across a live graph of over 500 billion nodes and 1 trillion edges. It uses Spanner Graph's multi-hop pattern matching combined with full-text and vector similarity search to trace connections between flagged accounts, payment methods, and other entities in real time. Developers benchmarking graph databases at scale can follow case studies like this on daily.dev."}},{"@type":"Question","name":"Can you query a graph directly on top of existing relational tables without migrating data, using Spanner?","acceptedAnswer":{"@type":"Answer","text":"Yes, Spanner Graph lets you define property graphs directly on top of existing relational tables with zero ETL, since the graph layer co-locates with the underlying distributed data splits. Google's ads team adopted this approach and moved most use cases to Spanner Graph within weeks rather than months, using ISO GQL to query the same tables instead of custom traversal code. Developers deciding between a dedicated graph database and layering graph queries on relational data can track approaches like this on daily.dev."}}]}
```

