How a Mid-Sized City Reduced FOIA Response Time from 18 Days to 4 Days with AI-Powered Document Discovery

Every month, the city's clerk's office received approximately 220 Freedom of Information Act requests. Every month, an average of 60% of those requests were answered late — past the 10-day statutory deadline, exposing the city to escalating complaint risk, potential litigation, and a growing backlog of increasingly frustrated requesters. The problem wasn't resources. The city had four dedicated FOIA staff. The problem was the documents.

The City

The city is a mid-sized municipality with a population of approximately 340,000, employing around 4,200 full-time equivalent staff across 18 city departments. The city operates a standard council-manager form of government, with the city clerk's office serving as the central FOIA processing function for all city departments.

The city's document management environment had evolved organically over two decades without a unified strategy. By 2024, responsive documents for a given FOIA request could exist in any combination of 12 separate systems: a shared network drive structure spanning 18 departments, a Microsoft SharePoint Online environment implemented in 2019 for newer documents, a legacy Documentum installation still housing records from 2004–2018, department-specific Microsoft Teams channels used informally since 2020, a Laserfiche records management system used by the City Clerk's office itself, four department-specific legacy databases (Public Works, Planning, Police Records, Finance), an email archive system, and a physical records log for pre-2004 paper records that had been partially digitised.

A FOIA clerk responding to a request for, say, all records related to a specific zoning variance decision from 2021 would need to search SharePoint, the network drives for three or four relevant departments, the Laserfiche system, and potentially the legacy Documentum archive — each with a different search interface, different access credentials, different query syntax, and different export procedures. For a moderately complex request, the document discovery phase alone could consume two to four hours of dedicated staff time.

The Compliance Crisis

Illinois' Freedom of Information Act requires a response within five business days for routine requests, extendable to ten business days with written notice. The city's practice of issuing the standard five-day extension notice on all requests gave the office a working baseline of ten business days — but even with that extension, the average response time was running at 18 business days, nearly double the statutory maximum.

The compliance failure was not for want of effort. FOIA staff were working full schedules. The backlog was growing because document discovery was genuinely time-consuming — not because staff were slow or inattentive. Three full-time FOIA clerks and one FOIA Coordinator were collectively spending an estimated 75% of their working hours on document discovery and retrieval across the 12 source systems. The remaining 25% covered request intake, legal review, redaction, response preparation, and correspondence.

"We had four people doing the work. The problem was that most of their time was spent searching — logging into one system, then another, then another. By the time you had found everything, you had half a day left to actually process the request. We needed a way to search everything at once."

— City Clerk

Evaluating a Solution

The City Clerk issued a Request for Information in August 2024, seeking vendors with experience deploying workplace search solutions in local government environments, specifically with the capability to connect multiple document systems — including legacy on-premises systems — through a unified search interface.

Three vendors were invited to demonstrate. The evaluation criteria were straightforward: the solution had to connect to SharePoint, the legacy Documentum system, the network drive structure, and the Laserfiche installation. It had to return results across all connected systems in a single query. It had to support relevance ranking by date, author, department, and document type. It had to produce an exportable result set for each query — a list of potentially responsive documents that could be used as the basis for the legal review and redaction phase. And it had to respect the existing permission structures in each source system — a staff member in Public Works should not be able to retrieve HR documents through the search interface.

Keyspider's Workplace Search product was selected following a three-week proof of concept in which all three vendors were given access to a sandboxed subset of the city's document environment. Keyspider's POC demonstrated 96% recall on a test set of 50 known documents drawn from across the 12 systems — the highest of the three vendors — and was the only solution that successfully connected to the legacy Documentum installation without requiring a custom middleware layer.

Deployment

Implementation began in October 2024. Keyspider's connectors for SharePoint, network drive structures, and Laserfiche were configured within the first week. The legacy Documentum integration, which required working with a specialist connector rather than a standard API, took an additional five days. The full-text email archive integration was the final component, completing at the end of week three.

By the end of week four, all 12 document systems were connected and searchable through a single Keyspider Workplace Search interface accessible to authorised FOIA staff. The interface allowed staff to search by keyword, date range, document type, department, and author — and to save search configurations for recurring request types. Query results displayed document titles, source system, last modified date, and a content snippet showing where the search terms appeared within each document.

A two-day training programme was conducted with all four FOIA staff members in the final week before go-live. The training covered the search interface, the export-to-review workflow (exporting potentially responsive documents to a structured folder for legal review), and the process for logging search queries as part of the FOIA process audit trail.

Results at 60 Days

Response Time

At the 60-day mark, the average FOIA response time had fallen from 18 business days to 4 business days — a 78% improvement. The 10-day statutory limit was being met for 96% of requests, up from 40% before deployment. The remaining 4% of late responses were complex, multi-department requests requiring extensive legal review and redaction — categories that had always required more time and were not meaningfully changed by improved document discovery.

Document Discovery Time

The most significant operational change was in the document discovery phase. FOIA staff reported that a typical moderate-complexity request — one requiring searches across three to five systems — now took 4 to 6 minutes to complete the discovery phase, compared to the prior average of 2 to 4 hours. For simple, single-department requests, discovery was complete in under 2 minutes.

This compression of the document discovery phase had a cascading effect on the entire FOIA workflow. With discovery now taking minutes rather than hours, staff were able to begin the legal review and redaction phase on the day a request was received. Requests that previously required multiple days just for discovery were now completing end-to-end in a single working day for straightforward categories.

Staff Capacity

The reduction in document discovery time freed approximately 60% of FOIA staff capacity — equivalent to approximately 2.4 full-time equivalents across the four-person team. Rather than reducing headcount, the city clerk redirected this capacity to proactive records management: building a FOIA reading room of commonly requested documents, indexing and tagging the backlog of un-indexed network drive files, and developing a proactive disclosure programme to reduce preventable FOIA volume in 2025.

Open Records Compliance: A Note on Audit Trails

One concern raised by the city attorney during the evaluation was the audit trail requirement: Open Records laws require that the city be able to demonstrate due diligence in its document searches — that a good-faith effort was made to locate all responsive documents. A manual search across 12 systems, while time-consuming, was at least demonstrably thorough when staff logged their search activities.

Keyspider's Workplace Search addressed this requirement through its search log functionality. Every query run through the system is logged with timestamp, staff identifier, query terms, filters applied, systems searched, and result count. This log is exportable as a structured report and can be attached to a FOIA response file as evidence of the search methodology. The city attorney reviewed the log format and confirmed it satisfied the evidentiary requirements for demonstrating good-faith search compliance under Illinois FOIA.

Cost Savings Calculation

The city's budget office conducted a post-deployment ROI analysis based on actual staff time data. The analysis calculated the annualised value of recovered staff time as follows: four FOIA staff averaging 6 hours/day on document discovery pre-deployment, reduced to approximately 1 hour/day post-deployment — a recovery of 20 staff-hours per day. At the city's average loaded staff cost for this grade level ($58/hour including benefits and overhead), the annual savings in staff time were calculated at approximately $290,000.

Against an annual Keyspider Workplace Search subscription cost, the city calculated a return on investment of approximately 4.2x in year one — and a higher ratio in subsequent years as the reading room programme reduces FOIA volume. The city's procurement team noted that the ROI calculation did not include the value of reduced legal exposure from compliance improvement, which they considered a significant additional benefit that was difficult to quantify but real.