Conduktor vs Lenses
ComparisonsConduktor vs Lenses, compared
Key takeaway
Conduktor Console Team Edition is 1,200 US dollars per seat per year, and Lenses DevX Team starts at 4,000 a year for up to 15 users, so the two bills are not comparable per unit. Both need a database: Console requires PostgreSQL 13 or later, and Lenses needs Postgres for HQ plus another for every per-cluster Agent. Conduktor Gateway is a proxy in the data path, adding 2 to 10 milliseconds per message. Kpow by Factor House is licensed per cluster at a published price.
What is Conduktor?
Conduktor is a commercial Kafka management and governance platform, sold as three separately licensed products: Console, a web interface over topics, schemas, connectors, consumer groups and access control; Gateway, a proxy that sits between clients and brokers; and a Schema Registry Proxy on its own Enterprise tier.
Gateway is the architectural fact that decides most evaluations. It enforces encryption, data masking, quota policy and multi-tenancy at the wire, so producers and consumers need no change to their code. Kafka is a binary protocol over TCP, and a component that speaks it can sit in the middle unnoticed.
- Assurance: SOC2 Type II certified in 2023.
- Audit: more than 70 event types, each carrying user identity, IP address, timestamp, topic and partition.
- Self-service: Topic as a Service hands topic creation to product teams inside policy guardrails.
- Footprint: PostgreSQL 13 or later for Console, plus 2 CPU and 3 GB of RAM, and a further 2 CPU and 4 GB for Gateway.
What is Lenses?
Lenses is a commercial Kafka governance and data exploration platform that sits on top of existing clusters, adding a SQL interface, topology and lineage visualisation, and a data catalogue. Architecturally it is the opposite of a proxy: it operates as an ordinary Kafka client and nothing in it sits in the data path, which is also why a KRaft cluster needs no modification. The deployment is a central Lenses HQ node with lightweight Agents running per cluster, and one Agent reaches one Kafka cluster.
SQL Studio is the differentiator, and it is what people who run Lenses talk about first: it replaces a search through millions of raw JSON, Avro or Protobuf payloads with a query, aimed at engineers and analysts who do not know Kafka. SQL Processors go further, giving Kubernetes-native stream processing defined in SQL and built on Kafka Streams. The product line splits into DevX, the UI, governance and SQL surface, and K2K, which is replication. Celonis acquired Lenses in early 2022, and 6.2.6 shipped on 19 August 2026.
What is the official 2026 pricing of Conduktor and Lenses?
Neither bill is one number, and the two are not metered on the same thing. Conduktor prices Console per seat: Team Edition is 1,200 US dollars per seat per year, or 125 US dollars per seat per month billed monthly, and the governance layer arrives at that published price rather than at a negotiation. Gateway Enterprise is licensed per cluster with a three-cluster minimum and is contact-only. Lenses prices DevX by capability, user count and cluster scope: Team starts at 4,000 US dollars a year, covers up to 15 users, and is scoped to a single cluster, with Multi-Kafka Enterprise custom priced past either limit.
Work both ends of the range. A team of five is five Conduktor seats at 6,000 a year against a Lenses Team price starting at 4,000, so the crossover sits between three and four engineers. Fifty seats is 60,000 a year at Conduktor list, and fifty users is well past the Lenses Team cap, so that side becomes a custom contract. The free tiers run the opposite way to the price tags: Console Community is free at 50 users and 3 clusters, while Lenses Community is free for up to 5 users with Basic Auth, no SSO and no RBAC. There is also a second Lenses meter, K2K Enterprise, from 1,000 US dollars a month with 5 clusters included and 200 a month per cluster after, paid on top of DevX.
Where does each one run out?
Conduktor’s costs are architectural, and Gateway carries most of them. It adds 2 to 10 milliseconds per message, which can be prohibitive under a sub-10-millisecond requirement, and it is a component in the data path that needs its own high-availability plan. Gateway Enterprise also carries a three-cluster minimum, so a team that wants wire-level policy on one cluster buys three.
- Index: when Console’s internal index is stale or incomplete it queries the cluster directly, which a user sees as slow page loads.
- Graphs: the monitoring graph averages brief spikes across wide time windows, so a short spike renders as zero until the range is narrowed.
- Azure: Event Hubs is reachable over Kafka protocol compatibility, but Azure’s native Schema Registry is not integrated.
- Tracing: no native distributed tracing, which is what stops it being a complete observability platform.
Lenses’ costs sit in the control plane, which is wider than it first looks: HQ plus its Postgres, and then one Agent and one Agent database for every Kafka cluster. Postgres is the only supported store for HQ, and the Community edition carries the same dependency, standing up separate lenses_hq and lenses_agent databases before either component starts.
- Masking: data policies match on field name across every registered dataset, with no escape even for an admin, so masking cannot vary by who is looking.
- Permissions: 6.2.5 split
UpdateTopicDetailsinto two actions, so custom roles granting the old one had to be re-granted by hand. - Portability: SQL Processors are proprietary and compiled, so anything built on them is a re-implementation if the team leaves.
- Reach: one Agent reaches one Kafka cluster, so a four-cluster estate is five databases before anybody logs in.
Which should you pick?
Pick Conduktor if:
- policy has to hold on every client with no way around it
- encryption, masking or multi-tenancy has to be enforced at the wire
- the estate is many clusters and one Console with one Postgres is the simpler shape
- the team is above three or four seats but well below fifty
Pick Lenses if:
- the people who need answers are not Kafka people
- SQL over topics replaces writing a consumer
- stream processing defined in SQL is part of the plan
- the team is between five and fifteen users on a single cluster
Gateway enforces at the wire, and a client cannot route past something it has to connect through, which is a different guarantee from a rule configured in a UI. On the other side, SQL Studio is the one axis where these two are genuinely not substitutes. Price where you expect to be in a year: a team of eight or ten inverts the arithmetic without a single feature changing hands.
Kpow: no database of its own, licensed per cluster instead of per seat
Conduktor and Lenses both bring a database with them before either team logs in. Console requires PostgreSQL 13 or later as a mandatory dependency, and Lenses needs Postgres for HQ plus a separate Agent and Agent database for every Kafka cluster it reaches, so a four-cluster deployment is HQ plus four Agents and their databases before anybody runs a query. Kpow by Factor House needs none of that: it is self-managed, vendor-agnostic tooling that runs against whatever Kafka you already have, as one stateless JVM container configured through environment variables with no external database. One instance manages up to 12 Kafka clusters, licensed per cluster at a published price, and it sits beside the cluster as an ordinary Kafka client rather than in the data path.
A tool that manages your clusters shouldn’t need a database of its own to do it. Start with Kpow on the cluster you already run and see what’s left to maintain.
Sources
- Apache Kafka protocol guide
- Apache Kafka documentation on authorization
- Apache Kafka documentation on basic operations