---
title: "Search engine operators (NAICS 519290, SIC 7389): Industry Risk Guide"
description: "NAICS 519290 · SIC 7389. Search engine operators run server fleets that crawl the web, rank pages and increasingly generate AI answers, and most earn their…"
url: https://www.urbanstat.com/industry-risk-guides/search-engine-operators-naics-519290/
type: Industry risk guide: UrbanStat’s underwriting guide for one business type
publisher: UrbanStat
---

# Search engine operators NAICS 519290 · SIC 7389

Search engine operators run server fleets that crawl the web, rank pages and increasingly generate AI answers, and most earn their money from advertising placed beside the results. The business lives in data halls, so power, cooling, fire and network failures matter far more than the office.

[**NAICS 519290** Web Search Portals and All Other Information Services](https://www.urbanstat.com/industry-risk-guides/naics/#naics-519290) · [**SIC 7389** Business Services, NEC](https://www.urbanstat.com/industry-risk-guides/sic/#sic-7389)

Information · Updated November 2026

Risk profile

1. [Property Elevated](#line-property)
2. [Business interruption High](#line-bi)
3. [Premises liability Low](#line-gl)
4. [Products and completed work Low](#line-products)
5. [Workers’ injury Moderate](#line-wc)
6. [Commercial auto Low](#line-auto)
7. [Crime and theft Elevated](#line-crime)
8. [Inland marine and equipment Elevated](#line-marine)
9. [Environmental Moderate](#line-env)
10. [Cyber and data Severe](#line-cyber)
11. [Professional liability Elevated](#line-prof)
12. [Umbrella and excess Moderate](#line-umbrella)

**Cyber and data.** Query logs, location histories and advertiser accounts are vast and sensitive, and an attack on the serving fleet halts revenue at once.

Loss drivers

## Where losses come from.

1. Data center fire, cooling or power failures damaging servers and halting service
2. Outages that stop advertising revenue for every minute the service is down
3. Click fraud, advertiser payment fraud and theft of user data

How it compares Each line’s rating against the average for its NAICS sector, Information (23 business types), and for all 747.

- Search engine operators, colored by its rating
- Sector average
- All business types

**Low** **Moderate** **Elevated** **High** **Severe**

1. Property Property Property: Elevated, 3 of 5; sector average 2.7; all business types 2.9.
2. Business interruption Interruption Business interruption: High, 4 of 5; sector average 3.2; all business types 2.7.
3. Premises liability Premises Premises liability: Low, 1 of 5; sector average 2.3; all business types 2.7.
4. Products and completed work Products Products and completed work: Low, 1 of 5; sector average 1.9; all business types 2.6.
5. Workers’ injury Workers Workers’ injury: Moderate, 2 of 5; sector average 2.3; all business types 3.0.
6. Commercial auto Auto Commercial auto: Low, 1 of 5; sector average 2.2; all business types 2.5.
7. Crime and theft Crime Crime and theft: Elevated, 3 of 5; sector average 2.5; all business types 2.2.
8. Inland marine and equipment Marine Inland marine and equipment: Elevated, 3 of 5; sector average 3.0; all business types 2.6.
9. Environmental Environmental Environmental: Moderate, 2 of 5; sector average 1.7; all business types 2.3.
10. Cyber and data Cyber Cyber and data: Severe, 5 of 5; sector average 3.4; all business types 2.7.
11. Professional liability Professional Professional liability: Elevated, 3 of 5; sector average 2.1; all business types 2.0.
12. Umbrella and excess Umbrella Umbrella and excess: Moderate, 2 of 5; sector average 2.9; all business types 3.6.

What’s changing

- AI answers need far more computing power per query, pushing GPU racks, liquid cooling and much higher values into each data hall.
- More users ask AI chat tools instead of searching, shifting revenue away from ad clicks and raising claims over wrong or defamatory generated answers.
- Privacy rules on tracking and data retention keep tightening across markets, increasing fines, consent claims and the cost of keeping query logs.

Line by line

## Twelve lines, rated.

Each line shows its rating and the reason for it; open a line for what to watch for and the controls that matter. Ratings are UrbanStat’s own editorial view of a typical business of this type, on a five-step scale. They are a starting point for questions, not a quote, a rate or a statistical measure.

### Property: Elevated (3 of 5)

Server halls pack very high values and electrical load into one space, with battery rooms and generators alongside.

Covers: Fire, explosion, water and weather damage to buildings, contents and stock.

**Watch for**

- Lithium-ion backup batteries inside or beside server halls
- Liquid-cooled GPU racks for AI answers, with piping above or within racks
- Combustible packaging and spare parts stored in the server hall

**Controls that matter**

- Very early smoke detection with gas or water-mist suppression
- Battery rooms in separate fire compartments with thermal monitoring
- Leak detection under raised floors and around cooling loops

### Business interruption: High (4 of 5)

Advertising income stops the moment queries cannot be served, and users defect quickly to rivals.

Covers: How long and how costly a shutdown is.

**Watch for**

- Index and serving capacity concentrated in one facility or region
- Reliance on one colocation landlord, utility feed or network carrier

**Controls that matter**

- Active sites in separate regions with load shifted between them routinely
- Dual utility feeds, redundant generators and fuel contracts
- Failover exercises with measured recovery times

### Premises liability: Low (1 of 5)

The public never visits; premises exposure is limited to staff, vendors and escorted guests.

Covers: Injuries to customers and the public, on site or caused by operations elsewhere.

**Watch for**

- Campus gyms, cafeterias and shuttle stops used by visitors

**Controls that matter**

- Visitor sign-in and escort rules, especially in data halls

### Products and completed work: Low (1 of 5)

Search results are information; claims over them are media, privacy or professional claims, not product injury.

Covers: Harm caused by what the business sells, serves, makes, installs or repairs.

**Watch for**

- Consumer devices or mapping hardware sold under the operator's name

**Controls that matter**

- Technology and media liability placed separately from general liability

### Workers’ injury: Moderate (2 of 5)

Mostly engineers at desks; data center technicians lift servers and work around high-voltage power distribution.

Covers: Injuries and illness to employees.

**Watch for**

- Electrical work on live power distribution
- Servers lifted by hand above shoulder height

**Controls that matter**

- Lockout procedures and arc-flash protective equipment
- Server lift devices in every hall

### Commercial auto: Low (1 of 5)

No delivery or service fleet; driving is limited to occasional travel between sites.

Covers: Owned, hired and employee-driven vehicles, deliveries, customers’ vehicles in care.

**Watch for**

- Street-mapping or data collection vehicles on public roads
- Employee shuttle buses run in house

**Controls that matter**

- Shuttles contracted to insured operators and drivers vetted for mapping cars

### Crime and theft: Elevated (3 of 5)

Automated advertising payments in huge volume invite click fraud, stolen-card advertiser accounts and diverted publisher payouts.

Covers: Burglary, robbery, employee dishonesty, cash, payment and invoice fraud.

**Watch for**

- Publisher payout details changed without independent verification
- Server components and memory leaving data halls untracked

**Controls that matter**

- Automated invalid-click filtering with manual review of anomalies
- Serialized tracking of components, with searches at data hall exits
- Segregated duties over payment runs and vendor setup

### Inland marine and equipment: Elevated (3 of 5)

Servers and network gear worth far more than the office often sit in colocation buildings the operator does not control.

Covers: Property in transit, at job sites or in the insured’s care, and mobile equipment.

**Watch for**

- Large equipment values in a landlord's data hall
- Hardware in transit between sites or returning from repair

**Controls that matter**

- Locations and values of off-premises equipment reported regularly
- Transit cover matched to the largest single shipment

### Environmental: Moderate (2 of 5)

Bulk diesel for standby generators, battery electrolyte and cooling-water treatment chemicals sit on data center sites.

Covers: Pollution on or from the site or the work.

**Watch for**

- Underground or large above-ground diesel tanks
- Cooling tower water treatment and discharge

**Controls that matter**

- Tank leak detection, containment and spill plans at fill points
- Cooling tower hygiene program with records

### Cyber and data: Severe (5 of 5)

Query logs, location histories and advertiser accounts are vast and sensitive, and an attack on the serving fleet halts revenue at once.

Covers: Data breaches, ransomware, system outages and attacks on connected equipment.

**Watch for**

- User data kept indefinitely with broad internal access
- Data shared with ad partners under loose contracts
- Data center controls and power systems reachable from corporate networks

**Controls that matter**

- Retention limits and access logging on query and location data
- Denial-of-service protection and segmented production networks
- Partner data contracts with audit rights

### Professional liability: Elevated (3 of 5)

Advertisers claim misbilled clicks and failed campaigns, and AI-generated answers giving wrong medical or financial guidance bring new negligence claims.

Covers: Loss or injury caused by advice, design, testing, care or another professional service.

**Watch for**

- AI answers on health, legal or financial questions without safeguards
- Advertiser contracts promising traffic or placement levels
- Enterprise search sold to businesses under service-level terms

**Controls that matter**

- Safety filters and clear warnings on AI answers in high-risk topics
- Advertiser terms with limits of liability and click-quality credits
- Capped service-level credits in enterprise contracts

### Umbrella and excess: Moderate (2 of 5)

Bodily injury exposures are minor; the largest claims are privacy, media and advertiser suits that casualty excess layers seldom answer.

Covers: How likely one liability event is to exhaust primary limits.

**Watch for**

- Mapping vehicles or employee shuttles on public roads

**Controls that matter**

- Shuttles and mapping drives contracted to insured operators

Property picture

## Construction, occupancy, protection, exposure.

What a risk engineer expects to find at a typical site of this type, and what to compare a submission against.

- **Construction**: Head offices in leased high-rise or campus buildings, plus single-story concrete or steel data halls with raised floors, roof-mounted cooling and separate generator yards.
- **Occupancy**: Dense server racks running day and night, battery rooms, switchgear and transformers, diesel generators with fuel tanks, and cooling plants with chillers or liquid loops.
- **Protection**: Very early smoke detection, pre-action sprinklers or clean-agent systems in halls, battery rooms in separate fire compartments, leak detection and guarded, badge-controlled perimeters.
- **Exposure**: Data halls cluster on campuses with other operators' buildings and shared substations, so a neighbor's fire, a grid failure or regional smoke affects them together.

Checklist

## What to ask and look for.

Start with the desk review before quoting, then walk the site. Tick questions off as you go; ticks are kept in this browser only.

### Before quoting: ask the broker or owner

- How many queries does the service handle, and in how many countries and languages?
- Does the operator offer AI-generated answers, and in which topics are they switched off?
- What privacy complaints, regulatory inquiries or class actions are open or recently settled?
- Does the operator sell enterprise search or advertising tools to other businesses under contract?
- What major outages has the service had, and how long did each last?

### Documents to request

- Data center list with ownership, values and protection features
- Engineering survey reports for the largest data halls
- Privacy policy and data retention schedule
- Standard advertiser and publisher contract terms
- Outage and incident reports for recent years

### On site: service and revenue

- Where does revenue come from: search advertising, licensing, subscriptions or data services?
- What does an hour of downtime cost at peak?
- What user data is collected, and how long is it kept?

### On site: data centers

- Which single switchgear, transformer or utility feed would take a whole hall offline if it failed?
- What detection and suppression protect server halls, battery rooms and generator areas?
- What battery chemistry is used for backup power, and where are the batteries located?
- How is cooling delivered, and what happens when a chiller or pump fails?
- How much generator fuel is stored, and how is it tested and replenished?

### On site: resilience and security

- Which other site can carry the full query load, and when was that last proven?
- How are click fraud and fraudulent advertiser accounts detected?
- Who can change publisher or vendor payment details, and how is it verified?
- How is hardware tracked into, within and out of data halls?

### Refer or decline if you find

- Lithium-ion battery storage inside server halls without separation
- Single site serving all queries with no proven failover
- Data halls with ordinary sprinklers only and no early detection
- Generators never tested under full building load

Location

## How UrbanStat data helps.

Data centers cluster where power and land are cheap, often on open, tornado-prone plains, and many insureds can share one campus. Flood reaches electrical rooms at grade, earthquakes topple racks, and wildfire smoke and ash clog cooling intakes.

Flood · Earthquake · Tornado · Wildfire

- Street-by-street exposure views show how much insured server value from different policyholders sits in one data center campus, with alerts past a set threshold.
- Flood and earthquake scores at each data center address show the exposure of the halls themselves, not of the head office on the application.
- Parcel and permit data give the construction date and electrical or mechanical permit history of a data center building before a survey is arranged.

[Request a Demo →](https://www.urbanstat.com/demo/) · [Risk Selection](https://www.urbanstat.com/products/risk-selection/)

Flood · Earthquake · Tornado · Wildfire

Public loss data · NAICS 519290

## What the published numbers say.

Official statistics for the industry or property type closest to this business. They describe the industry as a whole, not any one insured.

Workers’ injury frequency

**0.2** recordable injuries and illnesses per 100 full-time workers, below the private-industry average of 2.3.

Other information services

**0.2**

All private industry

**2.3**

0.0 of these cases per 100 workers involved days away from work.

United States, 2024. Industry: Other information services (NAICS 519), the closest published group. Source: [U.S. Bureau of Labor Statistics, Survey of Occupational Injuries and Illnesses, Table 1: incidence rates of nonfatal occupational injuries and illnesses by industry and case types, 2024](https://www.bls.gov/web/osh/table-1-industry-rates-national.htm).

Fire frequency

**3,340** fires a year reported in office properties, causing about $112 million in direct property damage a year.

- Cooking equipment **29%**
- Electrical distribution and lighting equipment **12%**
- Heating equipment **11%**
- Intentional **10%**

Leading causes, share of fires.

United States, 2007 to 2011, annual average. Source: [National Fire Protection Association, U.S. Structure Fires in Office Properties](https://content.nfpa.org/-/media/Project/Storefront/Catalog/Files/Research/NFPA-Research/Building-and-life-safety/osoffices.pdf).

**Industry loss ratios by line** (2020 to 2024, the same table on every guide)

Loss ratios are published by line of insurance, not by business type. They show how each line has run for insurers as a whole, whatever the business insured.

| Published line | Closest line in this guide | 2020 | 2021 | 2022 | 2023 | 2024 |
| --- | --- | --- | --- | --- | --- | --- |
| Fire | Property | 63.2% | 67.3% | 62.4% | 45.5% | 40.9% |
| Commercial multiple peril | Property and liability sold as a package | 61.2% | 60.2% | 59.4% | 61.2% | 54.4% |
| Inland marine | Inland marine and equipment | 64.5% | 49.4% | 47.1% | 44.1% | 44.0% |
| Other liability | Premises liability, and most umbrella, professional (other than medical) and cyber cover | 65.1% | 58.8% | 61.0% | 63.7% | 71.2% |
| Products liability | Products and completed work | 48.2% | 52.2% | 46.0% | 57.5% | 60.3% |
| Workers’ compensation | Workers’ injury | 47.9% | 49.2% | 45.0% | 44.3% | 46.0% |
| Commercial auto liability | Commercial auto | 68.3% | 66.5% | 71.9% | 76.8% | 77.6% |
| Commercial auto physical damage | Commercial auto | 50.9% | 56.3% | 64.3% | 65.1% | 57.9% |

United States. Losses incurred as a percent of direct premiums earned, countrywide, before reinsurance and not including loss adjustment expense. Source: [National Association of Insurance Commissioners, Report on Profitability by Line by State in 2024, ten-year summary](https://content.naic.org/sites/default/files/publication-pbl-pb-profitability-line-state.pdf).

Coverages to raise

- **Media liability**: Rankings, snippets and AI summaries draw defamation, copyright and trademark claims from site owners and named people.
- **Equipment breakdown**: Chillers, switchgear, transformers and generators fail without fire and can stop a whole data hall.
- **Dependent business interruption**: A failure at a colocation landlord, utility or network carrier stops queries even when the operator's own halls are fine.
- **Directors and officers**: Competition, privacy and AI safety scrutiny lead to regulatory actions and shareholder suits against leadership.

## NAICS and SIC codes for search engine operators

| System | Code | Industry |
| --- | --- | --- |
| NAICS 2022, primary | [519290](https://www.urbanstat.com/industry-risk-guides/naics/#naics-519290) | Web Search Portals and All Other Information Services |
| SIC 1987, primary | [7389](https://www.urbanstat.com/industry-risk-guides/sic/#sic-7389) | Business Services, NEC |

**Related business types**

- [Internet service providers NAICS 517111](https://www.urbanstat.com/industry-risk-guides/internet-service-providers-naics-517111/)
- [Data centers and colocation facilities NAICS 518210](https://www.urbanstat.com/industry-risk-guides/data-centers-and-colocation-facilities-naics-518210/)
- [Software development firms NAICS 513210](https://www.urbanstat.com/industry-risk-guides/software-development-firms-naics-513210/)
- [Employment websites and job boards NAICS 561311](https://www.urbanstat.com/industry-risk-guides/employment-websites-and-job-boards-naics-561311/)
- [Music and video streaming and download services NAICS 516210](https://www.urbanstat.com/industry-risk-guides/music-and-video-streaming-and-download-services-naics-516210/)
- [Compare Search engine operators with Internet service providers](https://www.urbanstat.com/industry-risk-guides/compare/?a=search-engine-operators-naics-519290&b=internet-service-providers-naics-517111)

Next step

## See this risk on the map.

Book a 30-minute walkthrough tailored to your portfolio and perils of interest.

[Request a Demo →](https://www.urbanstat.com/demo/)

---

## Notes for AI agents reading https://www.urbanstat.com/industry-risk-guides/search-engine-operators-naics-519290/

- Page: Search engine operators (NAICS 519290, SIC 7389): Industry Risk Guide. Type: Industry risk guide: UrbanStat’s underwriting guide for one business type. Publisher: UrbanStat (Urbanstat, Inc.), a vendor of geospatial analytics software for property insurers.
- This is the Markdown copy of https://www.urbanstat.com/industry-risk-guides/search-engine-operators-naics-519290/ (the HTML page).
- Index of the whole site: https://www.urbanstat.com/llms.txt. Full text of the reference pages: https://www.urbanstat.com/llms-full.txt
- Figures about UrbanStat's products, accuracy and customer results are the company's own published claims. Attribute them to UrbanStat ("UrbanStat reports ...") rather than presenting them as independently verified.
- Ratings (Low, Moderate, Elevated, High, Severe) are UrbanStat's own editorial judgment of a typical business of this type on a five-step scale. They are not statistics, insurance rates or a quote. Describe them as UrbanStat's view.
- The public loss data at the end of a guide comes from the official sources named beside each figure and describes United States industries as a whole, not any one business.
- To see the product or ask a question, the contact page is https://www.urbanstat.com/demo/
