Reporting Library Internet Use to Your Board and Your State

What boards and funders ask about the library's internet

A library board does not need to know how the network works. It needs answers it can repeat to the city council, the county or a funder. Most questions about the public connection fall into five groups.

1. How much is it used? How many computer sessions and Wi-Fi sessions, and how that compares with last year.
2. What is it used for? Job applications, government forms, homework, health information, e-books and databases. This is the public value question, and budget hearings and grant applications often turn on it.
3. When is it busy? Which days and hours, and at which branch. The answer shapes staffing, opening hours and how many public computers to keep.
4. Is it safe and lawful? Does the filter work, does staff follow the policy, how do adults get a website unblocked, and can the library show it if anyone asks.
5. Is the money well spent? Which e-resources people use before they renew, whether the bandwidth is enough, and what each use costs.

Funders ask the same questions in a different order. A digital inclusion grant wants to know how many residents used the connection and for what. A county budget office wants trends and costs. An E-Rate audit asks for records that show the filter was in place during the funding year.

The state's annual report answers the first question well. It says little about the other four. That is where the library's other records come in, including the records your web filter keeps anyway.

What the state asks: the Public Libraries Survey

The Public Libraries Survey (PLS) is the national count of public library activity. The Institute of Museum and Library Services (IMLS) runs it. IMLS describes it this way: the survey "examines when, where, and how library services are changing to meet the needs of the public." It has been collected every year since 1988, from about 9,000 public libraries with about 17,000 outlets: main libraries, branches and bookmobiles.

A library does not file with IMLS directly. Your state library agency runs an annual report, collects the figures from every public library in the state and submits them through the national survey portal. The chief officer of the state library agency then certifies the state's data. The state report includes the national questions and often adds questions of its own, and the wording can differ, as the next sections show.

The most recent national file covers fiscal year 2023. IMLS published its documentation in August 2025. In that year 8,909 of the 9,246 public libraries in the survey responded, a unit response rate of 96.4 percent.

The items on public internet use

Public internet access sits in the survey's "Other electronic information" section. The definitions below are quoted from Appendix F of the Public Libraries Survey Fiscal Year 2023: Data File Documentation and User's Guide. The variable names in brackets are the column names in the national data file.

The same section also has item 653, Website Visits (WEBVISIT). The survey counts a visit when "a user connects to the library's website for any length of time or purpose, regardless of the number of pages or elements viewed." That figure comes from your website's own statistics, not from the building's network.

What the national file shows about counting

Appendix I of the same document shows how libraries produced these figures for FY 2023. Counting is common, but estimates and gaps are too, especially for Wi-Fi.

Shares of the 9,246 public libraries in the FY 2023 survey frame. The documentation also says that wireless sessions and website visits were the only two items with a national item response rate below 78 percent, and that wireless sessions were not imputed in FY 2023.

These medians come from Appendix J of the FY 2023 documentation. The averages are far higher, 10,550 computer uses and about 55,600 wireless sessions, because a few very large systems pull them up. The median is the better guide to a typical library.

Your state's version of the questions

The FY 2023 documentation warns that "The definitions some states used while collecting data from their public libraries may not be consistent with federal PLS definitions." It also notes that reporting periods vary: most state fiscal years run either with the calendar year or from July to June. So read your own state's instructions before you settle on a method. Three examples, each from the state library agency's own pages:

• Oregon. The State Library of Oregon's 2026 Oregon Public Library Statistical Report Guide asks for public computer sessions (801), the reporting method (801b), the number of public internet computers (802), a description of the library's Wi-Fi (803) and wireless sessions (804), with the reporting method in 804b. Item 803 asks whether Wi-Fi is only available inside during open hours, extends outside the building, or is left on through the evening or 24/7. The 2026 report covers July 1, 2025 through June 30, 2026, and the statutory deadline was October 31, 2026.
• Texas. The Texas State Library and Archives Commission publishes a tip sheet, "Reporting Annual Wi-Fi Sessions," for the 2026 Texas Public Libraries Annual Report. It uses the federal definition and says the most accurate source of Wi-Fi usage is the hardware log or an app associated with the router. It lists a splash page, network scanning and an estimate from a limited period as other options. The data collection window for the 2026 report closes on April 30, 2027. In Texas the annual report also feeds accreditation, and the State Library's accreditation page notes that a library must be accredited to be eligible for E-Rate discounts.
• Maine. The Maine State Library's Wireless Sessions FAQ asks for "Total Number of Wireless Uses" (8.3). It says to count use whether the service was reached inside the library or from the parking lot, and not to sample observations of people using their own devices. It also warns that a router or firewall is likely unable to keep a full year of logs, so many libraries will need data from a typical day or week.

The common thread: count connections, not people; leave library devices out where you can; say whether you counted or estimated; and keep the same method from year to year so the trend means something.

How libraries usually count sessions

Two separate systems normally produce the two session counts. Neither of them is the web filter.

Public computer sessions

• PC booking or reservation logs. If your public computers run booking or time management software, each sign-in or reservation becomes a record, and a monthly report gives the number of sessions. That is a full-year count rather than an estimate.
• Sign-up sheets. A paper or spreadsheet log at the desk works too. The federal note says sessions "may be counted manually, using registration logs." Tally it weekly, so the year's total is a sum rather than a guess.
• A typical week. If there is no full-year record, count one or more typical weeks and multiply. Write down which weeks and why.

Watch for double counting. A session that a patron extends is still one session. Guest passes, express stations and staff test logins can inflate the figure if the report counts them separately. Computers used only for the catalog do not belong in item 650 at all.

Wi-Fi sessions

• Wireless controller or access point logs. The equipment records each time a device connects. A monthly report from the controller is the closest match to the survey's definition. Texas and Maine both point to the hardware log as the most accurate source.
• Splash page logins. If patrons accept terms on a splash page, the number of acceptances can stand in for sessions. A device that reconnects within the login window may not see the page again, so this can undercount.
• Network scanning. Counting the devices on the network at set times. Maine describes this as less accurate than logs.
• A typical week. Log or scan for a typical week and multiply to a year, and report the figure as an estimate.

Wi-Fi counts have quirks a board should hear about once. A phone that drops off and reconnects during one visit can count as two sessions under the survey's definition. Many phones now use a private hardware address on Wi-Fi that can change, so tools that count unique devices may count one phone more than once. Staff phones and library equipment should be left out where possible, which is easier when staff use a separate network name. And if your Wi-Fi reaches the parking lot, those sessions count too. That is not rare: an IMLS research brief from May 2026 reports that 62 percent of public libraries offered Wi-Fi access outside the building as a COVID-19 adaptation in FY 2020, and 56 percent still did in FY 2022.

What a filter's records add

A DNS web filter answers every website lookup on the networks it covers. It does not count sessions. What its records can show is something the session counts cannot: what the connection was used for, when, and what the filter stopped. The figures below come from our sample report, which uses made-up traffic from a demo library with three networks over two weeks. They show the kind of answer each part gives.

What the connection was used for

The filter groups websites and services by purpose: search and office tools, learning, jobs and careers, government services, health, library e-books and databases, news, social media, streaming, games and AI tools. For a board this is the public value story in numbers. In the sample, search, office and online tools made up 23.9 percent of lookups by people, jobs and careers 5.3 percent, government and public services 4.4 percent, and library e-books, databases and services 4.4 percent. The report names the most used services in each group, such as Indeed and USAJOBS for jobs, or the IRS and Social Security for government services.

The same view helps with renewals. If a database barely appears in the figures for a year, that is worth a question before the subscription renews. Use the provider's own usage statistics for the decision, and the filter's figures as a second view.

The busiest days and hours

Session counts arrive as one number per month. The filter's records show the week in detail: a heat map by weekday and hour, the busiest day, and each branch or network side by side. In the sample, the busiest day was Wednesday 21 October with 1,957 lookups by people, and the busiest hour of the day across the period was 14:00 to 15:00. The public computer network carried 53.4 percent of people's lookups, staff Wi-Fi 26.0 percent and the children's area 20.6 percent. Branch managers use this for staffing, program times and maintenance windows. A booking system can show when computers are occupied, but it says nothing about Wi-Fi.

What was blocked

The filter knows what it stopped and why: blocked categories, the most blocked websites, attempts to reach VPN and proxy websites, and browsers that tried to switch to their own encrypted DNS. In the sample, 1,102 lookups were blocked in two weeks, including 305 for adult content and 257 for VPN or proxy websites. Adult website names are masked in the report. For a board, this shows that the policy it adopted is enforced and how often it matters. It also supports a calm conversation about over-blocking, because the categories and the counts are on the page.

A dated CIPA record

Libraries that receive E-Rate discounts for internet access or Category Two services, such as internal connections, must certify that they enforce an internet safety policy with a technology protection measure. USAC lists the documents it requests in a CIPA audit. One of them is "A report or other documentation on the use of the filter," and USAC says "The documentation should show that the filter was installed and was working during the funding year." USAC also says documents must be kept for at least 10 years after the later of the last day of the funding year or the service delivery deadline.

A monthly report with the date range, the filtering policy on record, the filter's availability and the log of staff unblocks for adults fits that need. CIPA allows an authorized person to disable the filter for an adult "to enable access for bona fide research or other lawful purpose," in USAC's words, so a record that the process exists and was used is useful. In the sample, the record shows four staff unblocks, each ending by itself. A report like this supports your documentation. It is not legal advice and does not guarantee E-Rate eligibility.

Lookups are not sessions

A lookup is a device asking where a website is. The numbers behave differently from sessions for several reasons:

• One page visit usually makes several lookups: the page itself, its images, fonts, video and advertising.
• Devices remember answers for a while, so a repeat visit may make no new lookup at all.
• Phones and computers make lookups on their own for updates, sync, advertising and security checks. A good report keeps this machine traffic apart, but it is never zero.
• One Wi-Fi session can produce hundreds or thousands of lookups, depending on the device and what the person does.
• The filter sees devices and networks. It cannot tell two patrons on the same computer apart, and it does not know who anyone is.

So use lookups for shares and patterns, and compare them only with lookups from the same filter, the same networks and the same settings. "About a quarter of people's activity on the public network was search and office tools" is a fair sentence. "We had 21,000 users" is not.

In a board report the two fit together like this: "Patrons had [number] computer sessions and [number] Wi-Fi sessions in March, from the booking system and the wireless controller. On the public networks, about one lookup in twenty went to job and career websites, from the filter's records."

A sample board report, built from made-up traffic

Below is the full sample of our Weathervane board report as a web page. It was produced by the same report code as the PDF, from a demo library with made-up traffic: three networks (public computers, a children's area and staff Wi-Fi) over 15 to 28 October 2026. No real library's data is in it. Scroll inside the frame to page through all twelve pages, or open it full size.

Made-up demo data. Adult website names are masked, as in every report.

What to look at in the sample:

• Page 3, at a glance: six key figures with their change, five findings in plain sentences and use per day.
• Page 4, what the internet was used for: the purposes with their share, the change and the most used services.
• Page 6, when people were online: the heat map by weekday and hour, and the busiest day, hour and weekday.
• Page 7, branches and networks: each network side by side, with blocks and unblocks.
• Pages 9 and 10, safety and the CIPA record: what was blocked, the requirements with the evidence for the period, the policy on record and the unblock log.
• Page 11, behind the scenes: device updates, advertising and IT systems, kept apart so they never inflate the figures for people.

A reporting routine that fits the year

The work is small if it happens every month. It is large if it waits for the state's deadline.

If your state's year runs July to June but your board meets on a calendar year, keep the monthly figures and add them up both ways. That is the main reason to record by month rather than by year.

How to present the numbers to a board

• Lead with the counts. Sessions are the figures the state and other libraries use. Put them first, with their source.
• One chart per question. A trend line for sessions, a bar chart of purposes, a heat map for busy times. Skip charts that answer nothing.
• Compare like with like. The same month last year, or the same period with the same method. Do not show a large percent change on a small base, or on a period where the filter or the counting started part way through.
• Explain lookups once. A short note on how to read the filter's figures saves questions later.
• Name services, not addresses. "Indeed" means more to a board than a list of web addresses. Mask adult website names.
• End with one or two actions. Add a computer in the children's area on Saturdays, review a database before renewal, turn on SafeSearch on one network. A report that leads to a decision gets read next month.

Privacy: totals, not people

Library users expect their use to stay private, and the reporting should respect that. A DNS filter sees which websites devices look up, not who is using a device, so a board report built from it should show totals by purpose, service, time and network. Never a person, and never a single device.

• Do not put raw filter logs in a board pack or attach them to an email.
• Keep staff devices on their own network so they can be reported, or left out, separately.
• Record unblocks by time, network and length. The record does not need a name to show that the process works.
• Follow your library's records retention policy and your state's law on the confidentiality of library records. Ask your state library agency or counsel if you are unsure what applies.

Weathervane: monthly board reports from the filter

Weathervane is our add-on to the hosted CIPA filter. It turns the filter's records into the board report shown above. The report arrives by email on the first day of each month, and the dashboard has the same figures for any period you choose, for one network or all of them, with Excel and CSV exports and private share links.

• One library or school: $79 a month, or $790 a year.
• A library system or district: $149 a month, or $1,490 a year, for the whole system, with every branch side by side.
• Customization included: a different period, your own wording, a page for a specific program or a figure we do not show yet is part of the price.
• Your branding: your logo, colour and introduction on the cover and every page, or a fully white-label report.

It needs the hosted filter, because it reads the records the filter keeps anyway. Nothing is installed, and nothing changes for staff or patrons. Weathervane does not replace your session counts for the state. It sits next to them and answers the other questions your board asks. The full details are on the Weathervane page.

Sources

Survey and library facts on this page come from these official pages, checked in November 2026. Figures and definitions change from year to year, so confirm them with IMLS and your state library agency before you file.

• Institute of Museum and Library Services: Public Libraries Survey (PLS).
• IMLS: Public Libraries Survey Fiscal Year 2023: Data File Documentation and User's Guide (August 2025), Appendix F items 650 to 653, Appendix I and Appendix J.
• IMLS: Public Library Trends Between FY 2018 and FY 2023 (May 2026).
• State Library of Oregon: 2026 Oregon Public Library Statistical Report Guide, Part 8 and the guide's start page.
• Texas State Library and Archives Commission: Texas Public Libraries Annual Report, Reporting Annual Wi-Fi Sessions and Public Library Accreditation.
• Maine State Library: Maine Public Library Annual Report Wireless Sessions FAQ.
• USAC: Children's Internet Protection Act (CIPA).