Looker will bill Conversational Analytics overage from February 1, 2027

6 hours ago

6 min read

Looker will bill Conversational Analytics overage from February 1, 2027, title over a teal city skyline, Data Studios original cover

Google Cloud has set a date for charging Looker customers for Conversational Analytics. In the Looker release notes dated November 13, 2026, Google states that it will begin billing for usage of Gemini-assisted features that exceeds the instance token allocation, and the Gemini in Looker admin documentation, updated the same day, puts the start of that billing on February 1, 2027. Until January 31, 2027, customers keep promotional access without allocation limits or overage charges, within fair usage limits that the pricing page does not define.

The Looker pricing page, checked by Data Studios on November 15, 2026, lists the rates that apply from February: $3.00 per 1 million input data tokens and $20.00 per 1 million output data tokens, billed at the end of each month to the linked Google Cloud billing account based on actual usage. The included allocation is a monthly pool per instance, built from 1 million input tokens and 20,000 output tokens for each internal named user and 250,000 input tokens and 5,000 output tokens for each embed user, with a floor of 60 million input and 1.2 million output tokens for every production instance.

Looker admins have about eleven weeks to measure current usage, link a billing project and decide which users and agents should keep access once the meter starts.

··········

WHAT GOOGLE ANNOUNCED FOR LOOKER CONVERSATIONAL ANALYTICS BILLING.

The November 13 release note groups the billing notice with other changes that were enabled automatically from November 9, 2026 for instances running Looker 26.22. Google describes the charged scope as Gemini-assisted features, which it defines as agentic capabilities including Conversational Analytics and other generally available Gemini features. The admin documentation also states that customers will not be charged for Gemini-assisted features that are in preview.

The pricing page measures usage in data tokens at the instance level and splits them into two types. Input data tokens cover what is sent to the model, such as the prompt, the conversation history of a session and context like LookML semantic metadata and agent instructions. Output data tokens cover what the model generates, including the natural language answer, the API call or query used to retrieve data, other generated text or code, the reasoning shown in Thinking mode and any visualization the model creates.

Google lists the surfaces that count toward the monthly total: conversational experiences in Explores and dashboards, standalone and embedded agents, the Conversational Analytics endpoints in the Looker API, and Looker agents in Gemini Enterprise. The same rules apply to Looker (Google Cloud core) and Looker (original). The tokens are ring-fenced to Looker and cannot be used on other Google Data Cloud products, so they do not offset charges for features such as the Gemini tools covered in our report on Gemini in BigQuery pricing.

··········

LOOKER DATA TOKEN ALLOCATIONS AND OVERAGE RATES.

The allocation grows with the number of paid users enabled on the instance, and all authenticated users draw from the same pool. Allocations reset every month and unused tokens do not roll over.

........

Item	Input data tokens	Output data tokens
Per internal named user, monthly	1,000,000	20,000
Per embed user, monthly	250,000	5,000
Production instance minimum	60,000,000	1,200,000
Non-production or add-on instance	40,000,000 flat	800,000 flat
Overage price from Feb 1, 2027	$3.00 per 1M	$20.00 per 1M

........

Internal named users are Viewer, Standard and Developer users who log in to Looker directly with data access. Embed users are Signed Embed and Powered by Looker users. Google states that the total inclusion is the per-user allocation multiplied by the number of users of each type, and it calls the production figure a minimum, so a small instance receives at least 60 million input and 1.2 million output tokens even when its user count would produce less. At the overage rates, the tokens included for one internal named user are worth $3.40 a month and those for one embed user $0.85, and the production minimum corresponds to $204.00 of usage a month.

The edition prices on the same page remain quote-only, with Standard, Enterprise and Embed listed under an annual commitment and a call-sales price. The data token overage is billed monthly on actual usage, so it appears as a variable line next to a fixed annual subscription.

··········

HOW THE NEW MODEL DIFFERS FROM THE EARLIER LOOKER LISTING.

When Data Studios checked the same pricing page on November 10, 2026 for our comparison of Looker and Data Studio pricing, the included tokens were listed by edition: 60 million input and 1.2 million output tokens for Standard, 300 million and 6 million for Enterprise, and 1.2 billion and 24 million for Embed. The page also promised at least 90 days of notice before billing and quota enforcement began, without a date.

The current page drops the edition figures and replaces them with the per-user pool, the production minimum and a flat allocation for non-production instances, and it names February 1, 2027 as the start of allocations and pay-as-you-go billing. The overage rates of $3.00 and $20.00 per million tokens are the same as in the earlier listing. For a large Embed customer the change can be significant, because the included pool now depends on how many embed users are enabled rather than on the edition purchased.

··········

[[ADPLUS_300x250_1]]

LOOKER CONVERSATIONAL ANALYTICS COST EXAMPLE FOR TWO INSTANCES.

The two cases below are illustrative. User counts and monthly usage are assumptions, and the calculation applies the published allocations and rates; every figure was computed in Python.

........

Item	Instance A	Instance B
Internal named users	150	30
Embed users	400	0
Per-user pool, input	250M	30M
Per-user pool, output	5M	0.6M
Allocation applied, input / output	250M / 5M	60M / 1.2M (minimum)
Usage in the month, input / output	320M / 7.5M	75M / 1.5M
Overage, input / output	70M / 2.5M	15M / 0.3M
Overage charge	$210.00 + $50.00 = $260.00	$45.00 + $6.00 = $51.00

........

In Instance A, the 400 embed users add 100 million input tokens to the 150 million contributed by internal users. Instance B falls back on the production minimum, because 30 internal users produce only 30 million input tokens. Output tokens cost almost seven times as much per million as input tokens, so agents that write long answers, show their reasoning or generate many charts will move the bill faster than agents that receive long instructions.

users = {"internal": 150, "embed": 400}
pool_in  = max(users["internal"] * 1_000_000 + users["embed"] * 250_000, 60_000_000)
pool_out = max(users["internal"] * 20_000 + users["embed"] * 5_000, 1_200_000)
used_in, used_out = 320_000_000, 7_500_000
charge = max(0, used_in - pool_in) / 1e6 * 3.00 + max(0, used_out - pool_out) / 1e6 * 20.00
print(round(charge, 2))  # 260.0

··········

WHAT LOOKER ADMINS NEED TO SET UP BEFORE FEBRUARY 2027.

On Looker (original), the billing project is designated in the Admin panel under Platform, on the Gemini in Looker page. The admin selects Add project, enters the Google Cloud project number, grants the service account shown in the panel the Viewer IAM role on that project in the Google Cloud console, and then selects Validate and link. The Looker Admin role is required and the project must be linked to an active billing account. Google recommends using the project that should receive the charges.

Gemini in Looker on Looker (original) is available only for Looker-hosted instances, and the Gemini overview lists Looker 25.2 or later as the requirement for enabling it. On Looker (Google Cloud core), the features are enabled in the Google Cloud console by a user with the Looker Admin IAM role, and overage goes to the billing account linked to the instance.

For measurement, users with System Activity access can follow engagement and token usage across the instance through the Conversational Analytics dashboard in the System Activity section of the Admin panel. The pricing page explicitly invites customers to use the promotional period to baseline usage with these dashboards. Access itself is controlled through permissions: users need the gemini_in_looker permission on the relevant models, usually through the Gemini default role, Conversational Analytics roles or a custom role, plus access_data for the underlying data.

··········

LOOKER GEMINI BILLING: WHAT DEPENDS ON YOUR INSTANCE.

Several points are confirmed by Google in its documentation: the billing start date of February 1, 2027, the overage rates, the per-user and minimum allocations, the monthly reset without rollover, the list of surfaces that consume tokens and the exclusion of preview features from charges. The admin documentation describes the charged scope as Conversational Analytics and other generally available Gemini features, a group that has grown with recent releases such as the one described in our report on Looker 26.20.

Other points depend on each contract and instance. The size of the pool follows the mix of internal and embed users actually enabled, non-production instances receive a separate flat allocation, and the edition price is negotiated with Google. Google has not defined the fair usage limits that apply during the promotional period, and the pricing page does not explain whether the production minimum and the per-user total can combine in any way other than as a floor. The documentation also leaves open how generally available assistants outside Conversational Analytics, such as the Visualization Assistant, are measured against the data token pool, since the pricing page lists only conversational and agent surfaces. Teams that give analysts natural language access to warehouse data from spreadsheets should budget Conversational Analytics in Connected Sheets separately, because the Looker tokens cannot be used outside Looker.

Before February 1, 2027, admins can link a billing project, read the Conversational Analytics System Activity dashboard for at least one full month, count enabled internal and embed users, and compare expected usage with the pool each instance will receive.

··········

·····

DATA STUDIOS

·····

datastudios.org