Neo4j review

Neo4j is a graph database platform for storing and analyzing connected data. Instead of treating records as isolated rows, it represents entities and their relationships as a graph. That makes it a fit to consider when the connections between people, products, events, or other data are central to the work.

The product range covers graph databases, analytics, and AI applications, with fully managed and self-managed options listed on the homepage. Neo4j also positions its graphs as a knowledge layer for AI: knowledge graphs can capture relationships, history, and decisions, while GraphRAG retrieves context from connected data for AI systems. Its graph analytics and data science tools run graph algorithms on that data.

The practical question is workload fit. Neo4j is not a general-purpose answer for every database problem; the case is strongest when queries and analysis depend on traversing relationships.

Neo4j’s main strength is the breadth of graph-related work it puts in one platform. The listed capabilities include graph storage, knowledge graphs, graph analytics and data science, and GraphRAG. Teams can choose a fully managed service or a self-managed product, rather than being limited to one deployment model.

There are also user reports of graphs at meaningful scale. In one Reddit thread, a commenter described a project with 6 million nodes and 60 million relationships on Azure, used for reads, writes, and clustering algorithms. Another commenter reported a startup application with 200,000 users and 200 million nodes; they also said performance and stability issues improved after data modeling changes. These are individual accounts, not independent performance evidence, but they suggest results can depend heavily on the workload and graph design.

The free AuraDB tier is useful for learning and exploring: the pricing page lists it at $0 per month, with no payment method required and access to all graph tools. That gives teams a low-cost way to assess whether graph modeling suits their problem before paying for production capacity.

Production pricing is not presented as a fixed monthly or annual figure. The pricing page lists AuraDB Professional at $0.09 per GB per hour, with a 1GB minimum instance, and Business Critical at $0.20 per GB per hour, with a 2GB minimum. Your bill therefore depends on provisioned capacity and runtime; the pricing page does not give a monthly estimate. Professional lists up to 128GB of memory and 2TB of storage. Business Critical lists up to 2TB of memory on AWS and Google Cloud, 512GB on Azure, and 5TB of storage per instance.

The higher tier also lists a 99.95% uptime SLA and 30-day backup retention; Professional lists daily backups with 7-day retention. Dedicated VPC infrastructure is available through Virtual Dedicated Cloud, but pricing is custom. These are material differences to price out before choosing a tier.

Fit is another risk. Reddit commenters raise concerns about performance, stability, scaling, and modeling; others describe successful large graphs. One commenter also said they sync with Elasticsearch for faster, distributed search. Treat those reports as workload-specific, not a settled verdict on Neo4j.

Neo4j lists hourly capacity rates for its paid AuraDB plans, not monthly or annual prices. AuraDB Free costs $0, requires no payment method and includes access to all graph tools. Professional starts at $0.09/GB/hour, with a 1GB minimum instance. Business Critical starts at $0.20/GB/hour and requires at least 2GB. That is a higher rate and a larger minimum, so costs rise with both capacity and time running; the pricing page does not publish a standard monthly total.

Professional includes daily backups with seven-day retention. Business Critical adds a 99.95% uptime SLA and 30-day backup retention. Both plans list cloud availability on AWS, Azure and Google Cloud, though Business Critical’s memory limits vary by provider.

Virtual Dedicated Cloud has custom pricing; Neo4j asks buyers to contact sales. The pricing page lists no add-ons. Its hourly billing is the main gotcha: estimate your expected capacity and runtime before comparing it with a flat monthly database subscription. These prices were checked on 2026-11-07.

Switching makes the most sense when your application depends on connected data, graph analytics or graph-based context for AI. Neo4j offers fully managed and self-managed products, so a move between those options may preserve the platform choice while changing who runs the infrastructure. The published material does not describe migration tooling, supported import or export formats, or the effort involved; check those details against your current database and data model before planning a cutover.

Leaving may require more than moving records if application logic relies on graph relationships, analytics or GraphRAG. The available material does not document export behavior or portability limits, so we can’t say how much work a specific migration would take. It also contains no Reddit posts or comments, so we can’t responsibly report what users say makes them leave. Before committing, confirm how you can retrieve your data and estimate the cost of running the required capacity by the hour. For dedicated infrastructure, note that public pricing is not available.