Who can use this: Tech admin, Business admin
Available on: Any Gong plan
Ideal for: RevOps
MCP server tools and APIs consume Gong credits when they analyze calls and emails to generate AI insights. Credit consumption is driven by the amount of data analyzed and how often the same data is analyzed. Use the recommendations below to reduce the amount of data included in each request and avoid unnecessary re-analysis.
For information about how Gong credit consumption is calculated, learn more about How Gong credit consumption is calculated for MCP server tools and APIs.
General recommendations
These recommendations apply to all MCP server tools and APIs that consume credits.
Limit the amount of data analyzed
The amount of data included in a request has the largest impact on Gong credit consumption. Date ranges, entity scope, and filters determine how many calls and emails are analyzed.
Use shorter date ranges
The selected date range determines how many calls and emails are included in the analysis. A broad or unspecified date range can cause the MCP tool to analyze hundreds of emails and dozens of calls for a single account.
For many use cases, narrowing the date range significantly reduces consumption without affecting the quality of the answer. We recommend defining a specific period, such as the last 30 days or current quarter, whenever possible.
Most high consumption comes from broad requests against large accounts, not from complex questions.
❌ Broad request
“Give me a full summary of this account and everything discussed with the customer.”
✅ Focused request
“What risks were mentioned for this account in the last 30 days?”
Use the narrowest entity scope
The entity you query determines how much data is analyzed. Account queries analyze conversations across the account, including associated deals and contacts. Deal queries analyze only conversations associated with a specific opportunity, and Contact queries are narrower still.
When the question relates to a specific opportunity, use a Deal rather than an Account. Account queries often analyze significantly more conversations than Deal queries for the same question.
Apply filters
Filters reduce the amount of data included in the analysis. Use workspace, entity, and date filters to limit the request to the conversations that are relevant to the question.
❌ Too much data included
“Show all risks across all deals.”
✅ Restricts the scope
“Show risks for this deal in the last 30 days.”
For customer account queries, use the appropriate workspace so only relevant conversations are analyzed.
Run analysis only when needed
Every MCP request and brief generation consumes Gong credits when it runs. Scheduled jobs and recurring workflows consume credits each time they execute, even when the generated output is never used.
Generate insights when they are needed rather than pre-generating briefs or ASK responses for accounts that may never be reviewed.
Reuse existing smart trackers
Content built on existing smart trackers does not require the same conversations to be analyzed again. The conversations were already analyzed when the smart tracker ran.
For recurring business questions and standardized reporting, use existing smart trackers where possible instead of repeatedly analyzing the same conversations through MCP requests.
Best practices for ask_account and ask_deal
The ask_account and ask_deal tools analyze the selected conversations each time a request is submitted. If multiple requests are made against the same account or deal, the same calls and emails may be analyzed repeatedly.
The most effective way to reduce consumption is to avoid analyzing the same conversations more than once.
Combine related questions
Each request performs its own analysis of the selected conversations. Four separate questions about the same account can result in the same calls and emails being analyzed four times.
When questions are closely related, combine them into a single request so the conversations are analyzed once rather than repeatedly.
❌ Multiple requests
“Who are the stakeholders on this account?”
“What risks were discussed?”
“What priorities did the customer mention?”
“Which competitors were discussed?”
✅ Single request
“Summarize the stakeholders, risks, customer priorities, and competitor mentions for this account in the last 30 days.”
Do not combine unrelated questions solely to reduce consumption. Very broad requests can produce less focused answers than targeted questions.
Avoid stakeholder-by-stakeholder requests
Asking separate questions about individual stakeholders can cause the same conversations to be analyzed repeatedly. This is one of the most common sources of unnecessary consumption.
Instead of asking about each stakeholder separately, ask about all stakeholders in a single request.
❌ Multiple requests
“What concerns did Sarah raise?”
“What concerns did John raise?”
“What concerns did Priya raise?”
✅ Single request
“Summarize the concerns raised by all stakeholders on this account during the last 30 days.”
Design autonomous agents carefully
Autonomous agents can generate high consumption when they submit separate requests for each stakeholder, risk, topic, or account. This can multiply analysis quickly because each request analyzes the selected conversations independently.
Set clear instructions for the agent to batch related questions by entity. For example, the agent should gather the required information for one account or deal in a single request before moving to the next entity.
Best practices for generate_brief
The generate_brief tool analyzes the selected conversations once for each open-ended section in the brief. The number of sections included in a brief has a direct impact on Gong credit consumption.
Reduce the number of open-ended sections
Each open-ended section performs a separate analysis of the selected conversations. A brief with ten open-ended sections analyzes the same conversation set ten times, while a brief with one open-ended section analyzes it once.
The number of sections is often the largest consumption lever when designing briefs. Include only the sections required for the use case.
Create briefs specifically for MCP use
Many organizations use detailed briefs for account reviews, deal reviews, and other workflows inside Gong. These briefs are often designed to provide comprehensive information across multiple topics.
When using generate_brief through MCP, create dedicated briefs that contain only the sections needed for the workflow. This reduces the number of analyses performed and avoids generating information that will not be used.
Use web search sections selectively
Web search sections typically consume more Gong credits than other section types because they incorporate external information in addition to Gong data.
Include web search sections only when external context is required. Avoid using them as a default for every brief.
Review recurring brief generation
Every time a brief runs, the selected conversations are analyzed again and the full per-section consumption applies. This includes briefs generated on a schedule or as part of an automated workflow.
Review recurring brief generation periodically and run briefs only as often as the use case requires.