Snowflake Finance Team Speeds Forecasting With CoCo

Snowflake says its sales finance team reduced monthly forecasting and variance analysis from more than a week to a few hours using CoCo, its AI tool, after the team gave it standing context about data, terminology and workflows. The change lets the team publish a midmonth forecast update and lets any analyst rerun the forecast skill to test a scenario. The team previously ran more than 30 recurring analyses each month for sales leaders, including work on forecasts, sales pipelines, commissions and business performance. The forecast was the most involved and could run only once a month because several analysts had to review deals, CRM notes, historical trends and current field behavior.

Snowflake’s analytics team built a GTM Finance profile that CoCo reads before answering. It contains more than 20 certified data sources, definitions for metrics such as bookings, pipeline and quota, recurring monthly and quarterly tasks, standard report formats, and escalation paths for questionable figures. Snowflake says the profile cuts the steps needed to reach an answer from about 3.0 to 1.3 across 44 finance users, based on internal usage from February to May 2026. Each extra step without a profile adds about 0.05 credits, roughly 10 cents; a five-step session costs about three and a half times as much as a one-step session, and profile-assisted sessions cost less at every step count measured. One credit equals US$2.

With the profile in place, analysts built 36 reusable AI workflows, called skills, in two months, up from three starter skills. Three months after launch, 87% of first adopters still used CoCo daily. Examples include a bookings variance analysis that produces month-end results across deal categories in minutes, and a travel expense report that generates leader-level budget, spending and policy insights before drafting emails. Snowflake says travel reporting that took hours across two teams now takes about 15 minutes. The company says analysts can build skills without technical backgrounds or heavy reliance on the analytics team.

Snowflake says governance rests on three controls: the analytics team certifies approved data sources and documents their use; it centrally owns the profile’s data sources and terminology; and skills are shared and inspectable by the teams using them. The company recommends that other teams certify authoritative sources first, record terminology and output formats centrally, then let staff closest to each workflow build skills. Snowflake says its corporate financial planning and analysis team has also applied the approach to long-range planning. The company cites internal CoCo usage data covering 44 finance users from February to May 2026.

Snowflake says its sales finance team reduced monthly forecasting and variance analysis from more than a week to a few hours using CoCo, its AI tool, after the team gave it standing context about data, terminology and workflows. The change lets the team publish a midmonth forecast update and lets any analyst rerun the forecast skill to test a scenario. The team previously ran more than 30 recurring analyses each month for sales leaders, including work on forecasts, sales pipelines, commissions and business performance. The forecast was the most involved and could run only once a month because several analysts had to review deals, CRM notes, historical trends and current field behavior.

Snowflake’s analytics team built a GTM Finance profile that CoCo reads before answering. It contains more than 20 certified data sources, definitions for metrics such as bookings, pipeline and quota, recurring monthly and quarterly tasks, standard report formats, and escalation paths for questionable figures. Snowflake says the profile cuts the steps needed to reach an answer from about 3.0 to 1.3 across 44 finance users, based on internal usage from February to May 2026. Each extra step without a profile adds about 0.05 credits, roughly 10 cents; a five-step session costs about three and a half times as much as a one-step session, and profile-assisted sessions cost less at every step count measured. One credit equals US$2.

With the profile in place, analysts built 36 reusable AI workflows, called skills, in two months, up from three starter skills. Three months after launch, 87% of first adopters still used CoCo daily. Examples include a bookings variance analysis that produces month-end results across deal categories in minutes, and a travel expense report that generates leader-level budget, spending and policy insights before drafting emails. Snowflake says travel reporting that took hours across two teams now takes about 15 minutes. The company says analysts can build skills without technical backgrounds or heavy reliance on the analytics team.

Snowflake says governance rests on three controls: the analytics team certifies approved data sources and documents their use; it centrally owns the profile’s data sources and terminology; and skills are shared and inspectable by the teams using them. The company recommends that other teams certify authoritative sources first, record terminology and output formats centrally, then let staff closest to each workflow build skills. Snowflake says its corporate financial planning and analysis team has also applied the approach to long-range planning. The company cites internal CoCo usage data covering 44 finance users from February to May 2026.