Home

Features

Pricing

Case Studies

Managed Ad Service

Blog

News

About Us

Contact

Privacy Policy

Terms of Service

Legal Notice

Blog posts

In-House Ad Operations: Team Setup and Risk Prevention (2026)

7 Rules for Successful In-House Ad Operations in the AI Era

"Should we keep outsourcing ad operations to an ad agency or bring them in-house?" This is no longer simply a cost question. It is a strategic decision that shapes your business's competitiveness.

As markets change faster and restrictions on third-party cookies advance in stages, a rapidly growing number of companies are bringing ad operations in-house to build internal expertise and take control of their data.

Drawing on the ad operations landscape in 2026, this article explains practical strategies for making the transition successful and establishing a sustainable operating structure.

The strategic value of bringing ad operations in-house

The benefits of reducing fees are clear, but they are only the beginning. The real value lies in stronger organizational capabilities and deeper use of data.

The impact of reducing fees

Ad agency fees are charged as a percentage of ad spend, so the larger your ad budget, the greater the potential savings.

Monthly ad spend

Potential savings from bringing ad operations in-house

¥1 million

Significant room for savings even after staffing and tool costs

¥5 million

Equivalent to the cost of several dedicated full-time staff members

¥10 million

A budget sufficient to build an advanced team structure

3 strategic benefits beyond cost

Strategic value

Specific benefits

Why outsourcing to an ad agency cannot provide it

Make expertise a lasting asset

Retain lessons from every success and failure within the organization, strengthening long-term marketing capabilities

Expertise is lost when the contract ends

Much faster execution

You can implement a tactic in the afternoon that was decided in the morning meeting

Communication delays and gaps occur

Deeper business understanding

Ads are managed by people who understand the product, customers, and market most deeply

Ad agencies find it difficult to understand every nuance of the business

Preventing the biggest risk, dependence on one person, from day one

Dependence on one person occurs when ad operations knowledge is concentrated in a single team member. If that person leaves, ad operations stop, damaging the business.

Warning signs of dependence on one person

Warning sign

Risk level

Specific damage

"Only that person knows what's in the account"

High

Operations stop and performance drops sharply when they leave

Operating rules are not documented

High

Handover is impossible, and rebuilding takes months

No regular reviews

Medium

Knowledge stays with individuals, and organizational capabilities do not improve

Only one person accesses the management interface

High

Concentrated access permissions = a single point of failure

3 protocols to prevent dependence on one person

  • Thorough documentation: Document campaign naming conventions, budget management rules, report formats, and operating procedures in a manual.

  • A team of multiple people: Have at least 2 people manage operations and build a structure where they can cover for each other.

  • Regular knowledge-sharing meetings: Share successes and failures weekly/monthly to turn individual experience into organizational knowledge.

A hybrid model for a phased transition

Immediately ending your ad agency contract and rushing to move everything in-house is a common path to failure. A phased approach is essential for developing skills reliably while limiting risk.

4 stages of the hybrid model

Hybrid model

External role

Internal role

Best stage

External strategy + internal execution

Strategy development and advanced analysis (strategic advisor)

Day-to-day execution

Early stages of bringing ad operations in-house

External creative + internal operations

Production of banners, videos, and other assets

Ad operations and performance analysis

When there is no in-house designer

In-house operations with consulting

Regular advice and account audits

All operational work

Once the team can operate independently

Fully in-house

None (occasional consultation only when needed)

Everything

Once sufficient experience and structure are in place

Suggested transition timeline

  1. Phase 1: Preparation - Define objectives and KPIs, assemble the team, select tools, and plan joint operations with the ad agency.

  2. Phase 2: Joint operations - Learn through hands-on work alongside the ad agency. Gradually expand the scope of internal operations.

  3. Phase 3: Transfer of control - Your company takes the lead, and the ad agency serves as an advisor.

  4. Phase 4: Full transition - Operate entirely in-house, using external consultants only when needed.

Why cross-functional team design matters

A single person handling ad operations is not enough for success. Ad operations require multiple skill sets, and an appropriate division of responsibilities determines results.

3 required skill sets

Skill set

Specific capabilities

Role

Strategy and analysis

Customer analysis, competitive research, KPI design, and data-driven decision-making

Develop strategies that directly contribute to the business

Creative and content

Copywriting, design (Photoshop, etc.), and video editing

Create ad copy, banners, and landing pages

Technical

HTML/CSS basics, SEO, CMS management, and tag implementation

Build and maintain measurement infrastructure

An ideal team structure

  • Project leader (1 person) - Manages progress across initiatives and is responsible for meeting KPIs.

  • Marketer/ad operations specialist (1–2 people) - Handles research, strategy design, ad operations, and data analysis.

  • Creative specialist (1 person or a mix of internal and outsourced work) - Produces ad banners, landing pages, videos, and articles.

  • Engineer (assigned as needed) - Builds the site, implements tags, and supports technical issues.

Related reading

Core ad operations strategies and performance improvements: Practical methods by budget size

Practical approaches covering core ad operations strategies and optimization methods suited to different budget sizes.

Clarify objectives and KPIs

Unclear objectives are the biggest cause of failure. Clearly define why you are bringing ad operations in-house and establish quantitative KPIs to measure results.

Set KPIs for each objective

Objective of bringing ad operations in-house

Specific metric

Measurement method

Reduce costs

Percentage reduction in total ad operations costs (fees + staffing costs + tool costs)

Monthly cost comparisons

Build expertise

Improvement in internal skill levels (broader operational scope and share of work handled independently)

Quarterly reviews

Increase speed

Shorter lead time from planning to implementation

Measure time spent on each initiative

Deepen data use

Broader use of first-party data

Types of data integrations and number of use cases

Specific KPI examples

  • Ad performance: CPA, ROAS, CVR, and CTR, compared with the baseline before the transition.

  • Operational efficiency: Report creation time and implementation lead time, with quantitative time reduction targets.

  • Organizational capabilities: Number of ad platforms the team can manage and share of work handled independently, with targets for gradual expansion.

  • Costs: Monthly ROI evaluation of total operating costs (staffing costs + tool costs vs. previous fees).

Responding to tighter restrictions on third-party cookies

Google Chrome's phaseout of third-party cookies has been postponed, but the shift toward tighter restrictions is irreversible. First-party data, which companies collect directly from customers, is becoming key to competitive advantage.

Why in-house operations have an advantage in using first-party data

In-house advantage

Specific benefits

Why outsourcing to an ad agency makes this difficult

Security

Manage and use customer data securely in-house without sharing it externally

Risks of sharing data with third parties

Fast data integration

Work directly with engineers and data teams. Integrate immediately with CRM/SFA systems.

Delays from coordinating integration through an ad agency

Advanced customer outreach

Similar audiences based on high-LTV customers and reactivation of dormant customers

Restricted access to internal data

Practical ways to use first-party data

  • Lookalike targeting based on high-LTV customers - Create a lookalike audience using a list of your highest-LTV customers from your CRM (purchase history and LTV scores).

  • Reactivate dormant customers - Use marketing automation (MA) data, including last contact dates and behavior history, to retarget customers who have taken no action for a set period.

  • Integrate offline conversions - Import SFA data on sales opportunities and closed deals into Google Ads/Meta Ads to improve AI learning.

  • Use cart events - Use your core ecommerce system for dynamic retargeting of users who abandoned their carts.

How AI technology in 2026 solves in-house operations challenges

Rapid advances in generative AI, including ChatGPT, Claude, and Google Gemini, have largely addressed barriers that previously hindered bringing ad operations in-house: shortages of specialized skills, dependence on individual staff members, and limited production resources.

3 challenges AI addresses

Traditional challenge

AI solution

Example

Lack of specialized skills

AI automates complex data analysis and suggests improvements

Quickly identify root causes of performance changes

Risk of dependence on one person

AI objectively analyzes performance and recommends optimal budget allocation

Eliminate inconsistency in decisions

Limited production resources

AI assists with generating ad images and copy

Accelerate creative testing cycles

Implementing AI at each layer

  1. Bid optimization - AI optimizes budget allocation through Google Smart Bidding and Meta Advantage+, while people design target KPIs and guardrails.

  2. Reporting - Google Analytics Intelligence and Looker Studio's AI features generate reports automatically, while people interpret the data and develop strategies.

  3. Creative - ChatGPT, Claude, and Midjourney generate ad copy and banners at scale, while people ensure brand fit and final quality.

  4. Analysis - AI offers performance analysis and budget optimization recommendations, while people extract insights and provide business context.

Related reading

How to use Looker Studio: Making the most of a free data visualization tool

A detailed explanation of how to create effective reports and build dashboards with Looker Studio.

AI agents that automate ad operations, such as Cascade, can provide even more advanced support for bringing ad operations in-house.

Practical checklist

Use this checklist for each stage of bringing ad operations in-house to track progress and establish a reliable operating structure.

When considering the transition

  • Clearly defined the objectives of bringing ad operations in-house: cost reduction, expertise development, speed, or data use.

  • Set specific KPIs, such as CPA, ROAS, and operating costs.

  • Compared current ad agency fees with expected in-house costs (staffing + tools).

  • Shared the objectives with all stakeholders, from executives to frontline staff.

When building the operating structure

  • Secured a team of at least 2 people to prevent dependence on one person.

  • Documented operating procedures, naming conventions, and report formats.

  • Decided how to cover the required skill sets: strategy, creative, and technical.

  • Developed a hybrid model with a plan for joint operations with the ad agency.

During the transition

  • Developing practical skills while working alongside the ad agency.

  • Holding weekly/monthly knowledge-sharing meetings.

  • Confirmed that the company owns its ad accounts.

  • Gradually expanding the scope of operations.

After operations are established

  • Expanding first-party data use through CRM/SFA integration.

  • Using AI tools to make analysis, bidding, and creative production more efficient.

  • Regularly evaluating KPI improvements compared with the period before the transition.

  • Maintaining protocols to prevent dependence on one person: documentation, team structure, and knowledge-sharing meetings.

For ways to reduce the workload of in-house operations, What is ad operations automation? Tasks, platform-specific features, and how to get started outlines which tasks can be automated and which require human judgment.

For how to choose tools that support in-house operations, see Ad operations tools: Types, comparisons, and selection.

For an option that lets you run in-house operations without manual work, also see What are AI agents for ad operations? Capabilities, differences from traditional tools, and implementation steps.

For how to distinguish ad spend from operating costs, a prerequisite for bringing ad operations in-house, see Google Ads costs: Pricing models, minimum ad spend, and setting a budget.

FAQ

What is the minimum team size for in-house ad operations?

At least 2 people are recommended. With only 1 person, the risk of dependence on that individual is high, and operations may stop during vacations or after their departure. Ideally, a team of 3 covers strategy, operations, and creative, but people covering multiple roles can also work well.

Should we terminate our ad agency contract immediately?

An immediate, full transition is not recommended. It is safer to start with a hybrid model, work alongside the ad agency, and gradually build skills before moving fully in-house.

At what level of ad spend do in-house costs become lower than ad agency fees?

The larger your ad budget, the easier it is to achieve cost advantages even after accounting for staffing and tool costs. However, consider the value of building expertise and making decisions faster, not just the monetary amount.

Can we bring ad operations in-house using only AI tools?

AI is a powerful support tool, but strategy development, brand decisions, and understanding business context remain human responsibilities. Dividing roles appropriately between AI and people is key to successful in-house operations in 2026.

What preparation is needed to use first-party data?

You need to build a system that connects existing databases, such as CRM, MA, SFA, and core ecommerce systems, with ad platforms. Establishing a structure for collaboration with engineers and data teams in advance is important.

Summary

Bringing ad operations in-house is the most reliable path to turning marketing from an "expense" into a "strategic asset."

In 2026, tighter third-party cookie restrictions and rapid advances in AI technology make bringing ad operations in-house even more valuable. These 7 points are particularly important:

  1. Recognize the strategic value - Cost reduction is only the beginning.

  2. Prevent dependence on one person - Address the biggest risk from day one.

  3. Transition in stages - Build skills reliably through a hybrid model.

  4. Build a cross-functional team - Ensure coverage of multiple skill sets.

  5. Set clear objectives and KPIs - Measure results quantitatively.

  6. Use first-party data - A source of competitive advantage.

  7. Use AI technology - A way to solve longstanding challenges.

Build a sustainable in-house structure to develop your company's ad operations capabilities as a strategic asset and establish a long-term competitive advantage.

Cascade product guide

Download the guide

Table of contents

Cascade sets
your marketing free

Cross-channel measurement, daily optimization, and reporting.
Cascade connects the pieces of ad management that used to live in silos.

Blog posts

In-House Ad Operations: Team Setup and Risk Prevention (2026)

Table of contents

7 Rules for Successful In-House Ad Operations in the AI Era
7 Rules for Successful In-House Ad Operations in the AI Era

"Should we keep outsourcing ad operations to an ad agency or bring them in-house?" This is no longer simply a cost question. It is a strategic decision that shapes your business's competitiveness.

As markets change faster and restrictions on third-party cookies advance in stages, a rapidly growing number of companies are bringing ad operations in-house to build internal expertise and take control of their data.

Drawing on the ad operations landscape in 2026, this article explains practical strategies for making the transition successful and establishing a sustainable operating structure.

The strategic value of bringing ad operations in-house

The benefits of reducing fees are clear, but they are only the beginning. The real value lies in stronger organizational capabilities and deeper use of data.

The impact of reducing fees

Ad agency fees are charged as a percentage of ad spend, so the larger your ad budget, the greater the potential savings.

Monthly ad spend

Potential savings from bringing ad operations in-house

¥1 million

Significant room for savings even after staffing and tool costs

¥5 million

Equivalent to the cost of several dedicated full-time staff members

¥10 million

A budget sufficient to build an advanced team structure

3 strategic benefits beyond cost

Strategic value

Specific benefits

Why outsourcing to an ad agency cannot provide it

Make expertise a lasting asset

Retain lessons from every success and failure within the organization, strengthening long-term marketing capabilities

Expertise is lost when the contract ends

Much faster execution

You can implement a tactic in the afternoon that was decided in the morning meeting

Communication delays and gaps occur

Deeper business understanding

Ads are managed by people who understand the product, customers, and market most deeply

Ad agencies find it difficult to understand every nuance of the business

Preventing the biggest risk, dependence on one person, from day one

Dependence on one person occurs when ad operations knowledge is concentrated in a single team member. If that person leaves, ad operations stop, damaging the business.

Warning signs of dependence on one person

Warning sign

Risk level

Specific damage

"Only that person knows what's in the account"

High

Operations stop and performance drops sharply when they leave

Operating rules are not documented

High

Handover is impossible, and rebuilding takes months

No regular reviews

Medium

Knowledge stays with individuals, and organizational capabilities do not improve

Only one person accesses the management interface

High

Concentrated access permissions = a single point of failure

3 protocols to prevent dependence on one person

  • Thorough documentation: Document campaign naming conventions, budget management rules, report formats, and operating procedures in a manual.

  • A team of multiple people: Have at least 2 people manage operations and build a structure where they can cover for each other.

  • Regular knowledge-sharing meetings: Share successes and failures weekly/monthly to turn individual experience into organizational knowledge.

A hybrid model for a phased transition

Immediately ending your ad agency contract and rushing to move everything in-house is a common path to failure. A phased approach is essential for developing skills reliably while limiting risk.

4 stages of the hybrid model

Hybrid model

External role

Internal role

Best stage

External strategy + internal execution

Strategy development and advanced analysis (strategic advisor)

Day-to-day execution

Early stages of bringing ad operations in-house

External creative + internal operations

Production of banners, videos, and other assets

Ad operations and performance analysis

When there is no in-house designer

In-house operations with consulting

Regular advice and account audits

All operational work

Once the team can operate independently

Fully in-house

None (occasional consultation only when needed)

Everything

Once sufficient experience and structure are in place

Suggested transition timeline

  1. Phase 1: Preparation - Define objectives and KPIs, assemble the team, select tools, and plan joint operations with the ad agency.

  2. Phase 2: Joint operations - Learn through hands-on work alongside the ad agency. Gradually expand the scope of internal operations.

  3. Phase 3: Transfer of control - Your company takes the lead, and the ad agency serves as an advisor.

  4. Phase 4: Full transition - Operate entirely in-house, using external consultants only when needed.

Why cross-functional team design matters

A single person handling ad operations is not enough for success. Ad operations require multiple skill sets, and an appropriate division of responsibilities determines results.

3 required skill sets

Skill set

Specific capabilities

Role

Strategy and analysis

Customer analysis, competitive research, KPI design, and data-driven decision-making

Develop strategies that directly contribute to the business

Creative and content

Copywriting, design (Photoshop, etc.), and video editing

Create ad copy, banners, and landing pages

Technical

HTML/CSS basics, SEO, CMS management, and tag implementation

Build and maintain measurement infrastructure

An ideal team structure

  • Project leader (1 person) - Manages progress across initiatives and is responsible for meeting KPIs.

  • Marketer/ad operations specialist (1–2 people) - Handles research, strategy design, ad operations, and data analysis.

  • Creative specialist (1 person or a mix of internal and outsourced work) - Produces ad banners, landing pages, videos, and articles.

  • Engineer (assigned as needed) - Builds the site, implements tags, and supports technical issues.

Related reading

Core ad operations strategies and performance improvements: Practical methods by budget size

Practical approaches covering core ad operations strategies and optimization methods suited to different budget sizes.

Clarify objectives and KPIs

Unclear objectives are the biggest cause of failure. Clearly define why you are bringing ad operations in-house and establish quantitative KPIs to measure results.

Set KPIs for each objective

Objective of bringing ad operations in-house

Specific metric

Measurement method

Reduce costs

Percentage reduction in total ad operations costs (fees + staffing costs + tool costs)

Monthly cost comparisons

Build expertise

Improvement in internal skill levels (broader operational scope and share of work handled independently)

Quarterly reviews

Increase speed

Shorter lead time from planning to implementation

Measure time spent on each initiative

Deepen data use

Broader use of first-party data

Types of data integrations and number of use cases

Specific KPI examples

  • Ad performance: CPA, ROAS, CVR, and CTR, compared with the baseline before the transition.

  • Operational efficiency: Report creation time and implementation lead time, with quantitative time reduction targets.

  • Organizational capabilities: Number of ad platforms the team can manage and share of work handled independently, with targets for gradual expansion.

  • Costs: Monthly ROI evaluation of total operating costs (staffing costs + tool costs vs. previous fees).

Responding to tighter restrictions on third-party cookies

Google Chrome's phaseout of third-party cookies has been postponed, but the shift toward tighter restrictions is irreversible. First-party data, which companies collect directly from customers, is becoming key to competitive advantage.

Why in-house operations have an advantage in using first-party data

In-house advantage

Specific benefits

Why outsourcing to an ad agency makes this difficult

Security

Manage and use customer data securely in-house without sharing it externally

Risks of sharing data with third parties

Fast data integration

Work directly with engineers and data teams. Integrate immediately with CRM/SFA systems.

Delays from coordinating integration through an ad agency

Advanced customer outreach

Similar audiences based on high-LTV customers and reactivation of dormant customers

Restricted access to internal data

Practical ways to use first-party data

  • Lookalike targeting based on high-LTV customers - Create a lookalike audience using a list of your highest-LTV customers from your CRM (purchase history and LTV scores).

  • Reactivate dormant customers - Use marketing automation (MA) data, including last contact dates and behavior history, to retarget customers who have taken no action for a set period.

  • Integrate offline conversions - Import SFA data on sales opportunities and closed deals into Google Ads/Meta Ads to improve AI learning.

  • Use cart events - Use your core ecommerce system for dynamic retargeting of users who abandoned their carts.

How AI technology in 2026 solves in-house operations challenges

Rapid advances in generative AI, including ChatGPT, Claude, and Google Gemini, have largely addressed barriers that previously hindered bringing ad operations in-house: shortages of specialized skills, dependence on individual staff members, and limited production resources.

3 challenges AI addresses

Traditional challenge

AI solution

Example

Lack of specialized skills

AI automates complex data analysis and suggests improvements

Quickly identify root causes of performance changes

Risk of dependence on one person

AI objectively analyzes performance and recommends optimal budget allocation

Eliminate inconsistency in decisions

Limited production resources

AI assists with generating ad images and copy

Accelerate creative testing cycles

Implementing AI at each layer

  1. Bid optimization - AI optimizes budget allocation through Google Smart Bidding and Meta Advantage+, while people design target KPIs and guardrails.

  2. Reporting - Google Analytics Intelligence and Looker Studio's AI features generate reports automatically, while people interpret the data and develop strategies.

  3. Creative - ChatGPT, Claude, and Midjourney generate ad copy and banners at scale, while people ensure brand fit and final quality.

  4. Analysis - AI offers performance analysis and budget optimization recommendations, while people extract insights and provide business context.

Related reading

How to use Looker Studio: Making the most of a free data visualization tool

A detailed explanation of how to create effective reports and build dashboards with Looker Studio.

AI agents that automate ad operations, such as Cascade, can provide even more advanced support for bringing ad operations in-house.

Practical checklist

Use this checklist for each stage of bringing ad operations in-house to track progress and establish a reliable operating structure.

When considering the transition

  • Clearly defined the objectives of bringing ad operations in-house: cost reduction, expertise development, speed, or data use.

  • Set specific KPIs, such as CPA, ROAS, and operating costs.

  • Compared current ad agency fees with expected in-house costs (staffing + tools).

  • Shared the objectives with all stakeholders, from executives to frontline staff.

When building the operating structure

  • Secured a team of at least 2 people to prevent dependence on one person.

  • Documented operating procedures, naming conventions, and report formats.

  • Decided how to cover the required skill sets: strategy, creative, and technical.

  • Developed a hybrid model with a plan for joint operations with the ad agency.

During the transition

  • Developing practical skills while working alongside the ad agency.

  • Holding weekly/monthly knowledge-sharing meetings.

  • Confirmed that the company owns its ad accounts.

  • Gradually expanding the scope of operations.

After operations are established

  • Expanding first-party data use through CRM/SFA integration.

  • Using AI tools to make analysis, bidding, and creative production more efficient.

  • Regularly evaluating KPI improvements compared with the period before the transition.

  • Maintaining protocols to prevent dependence on one person: documentation, team structure, and knowledge-sharing meetings.

For ways to reduce the workload of in-house operations, What is ad operations automation? Tasks, platform-specific features, and how to get started outlines which tasks can be automated and which require human judgment.

For how to choose tools that support in-house operations, see Ad operations tools: Types, comparisons, and selection.

For an option that lets you run in-house operations without manual work, also see What are AI agents for ad operations? Capabilities, differences from traditional tools, and implementation steps.

For how to distinguish ad spend from operating costs, a prerequisite for bringing ad operations in-house, see Google Ads costs: Pricing models, minimum ad spend, and setting a budget.

FAQ

What is the minimum team size for in-house ad operations?

At least 2 people are recommended. With only 1 person, the risk of dependence on that individual is high, and operations may stop during vacations or after their departure. Ideally, a team of 3 covers strategy, operations, and creative, but people covering multiple roles can also work well.

Should we terminate our ad agency contract immediately?

An immediate, full transition is not recommended. It is safer to start with a hybrid model, work alongside the ad agency, and gradually build skills before moving fully in-house.

At what level of ad spend do in-house costs become lower than ad agency fees?

The larger your ad budget, the easier it is to achieve cost advantages even after accounting for staffing and tool costs. However, consider the value of building expertise and making decisions faster, not just the monetary amount.

Can we bring ad operations in-house using only AI tools?

AI is a powerful support tool, but strategy development, brand decisions, and understanding business context remain human responsibilities. Dividing roles appropriately between AI and people is key to successful in-house operations in 2026.

What preparation is needed to use first-party data?

You need to build a system that connects existing databases, such as CRM, MA, SFA, and core ecommerce systems, with ad platforms. Establishing a structure for collaboration with engineers and data teams in advance is important.

Summary

Bringing ad operations in-house is the most reliable path to turning marketing from an "expense" into a "strategic asset."

In 2026, tighter third-party cookie restrictions and rapid advances in AI technology make bringing ad operations in-house even more valuable. These 7 points are particularly important:

  1. Recognize the strategic value - Cost reduction is only the beginning.

  2. Prevent dependence on one person - Address the biggest risk from day one.

  3. Transition in stages - Build skills reliably through a hybrid model.

  4. Build a cross-functional team - Ensure coverage of multiple skill sets.

  5. Set clear objectives and KPIs - Measure results quantitatively.

  6. Use first-party data - A source of competitive advantage.

  7. Use AI technology - A way to solve longstanding challenges.

Build a sustainable in-house structure to develop your company's ad operations capabilities as a strategic asset and establish a long-term competitive advantage.

Cascade sets
your marketing free

Cross-channel measurement, daily optimization, and reporting. Cascade connects the pieces of ad management that used to live in silos.

Cascade sets
your marketing free

Cross-channel measurement, daily optimization, and reporting. Cascade connects the pieces of ad management that used to live in silos.