# Document management and the cost of information silos

**Document management information silos** describe a familiar scene. An owner looks for a critical document, an old quote, a technical detail of a job or a message from a client, and the search repeats across email, cloud drives, the management system, messaging apps and paper archives. The search takes 20 to 30 minutes. Multiplied by team, days and year, the cost becomes large. The problem is not a lack of documentation. The difficulty lies in locating the right document when it is needed.

A custom document management system responds to that fragmentation. According to the Digital Innovation Observatories of Politecnico di Milano, about 60% of document system implementations fail because the existing operational flows were never understood, rather than because of technology limits. The real question is how to map the document relationships a business already has into an architecture that strengthens them.

## The myth of one system for everyone

A common mistake is searching for the perfect software that suits any organisation. Companies implement generalist packages and then adapt their processes to the software, reversing the correct order. Each business has its own operating logic, shaped by sector, size and history. Generic metadata cannot capture those specifics: a construction company does not categorise documents like a law firm, and a restaurant does not share the flows of a manufacturer. The structural rigidity of standard systems collides with the fluidity of daily operations. In software development, the concept of dynamic context discovery follows the opposite principle: instead of loading all available context, the system retrieves only what the current task needs. A custom document system applies the same idea to files and records.

## What fragmentation costs

The numbers are concrete. A professional spends on average 20% of the working day searching for documents, about 90 minutes out of eight hours. For a team of five people at an average hourly cost of EUR 30, that exceeds EUR 54,000 per year in search time alone. Coopers & Lybrand put the average cost of recreating a lost document at EUR 220. The invisible costs add up beyond that: delayed decisions, errors caused by outdated versions, and commercial opportunities lost to slow responses. The Italian digital document management market reached EUR 2.3 billion in 2025, up 13% from 2021 according to the Digital B2B Observatory, yet only 42% of companies have adopted a document management system, and half invest less than 1% of revenue in these projects.

## Metadata built on real workflows

The difference between a generic archive and a custom system is operational. A generic system offers predefined folders and tags, standard categories that do not reflect the business and a search based on file name, date or generic keywords. A custom system builds metadata on the specific operating logic of the company, models the structure of area, sector and case on the real workflow, and uses a relational database that knows the business. A construction company looking for documentation on waterproofing for condominium Rossi gets hundreds of files from a generic search and three relevant documents in two seconds from a system indexed by work area, civil construction sector, client and intervention type. Intelligent preselection beats generic search, and the gain is contextual precision rather than raw speed.

## Why custom is not a luxury

The predictable objection is cost. The arithmetic says otherwise. A generic system charges EUR 80 to EUR 150 per user per month, and over years adds the hours lost to ineffective searches, the errors from wrongly retrieved documents and the opportunity cost of delayed decisions. The total usually exceeds the initial investment in custom development, amortised over three to five years. The advantage goes beyond cost: the processes do not adapt to the software, because the system adapts to the way the company already works. When the business evolves, the metadata grow with it, and there is no migration to a more powerful package every three years.

## What makes it accessible now

Three converging factors bring custom document management within reach of mid-sized and small companies. Hardware and computing costs are falling: Nvidia announced Vera Rubin chips with a 90% cost reduction over the previous generation, and AMD introduced processors designed for smaller corporate data centers rather than hyperscalers alone. Mature relational databases such as MySQL and PostgreSQL deliver high performance on custom architectures at contained cost. And the supplier market is consolidating, with a USD 20 billion funding round for xAI involving Nvidia, Fidelity and Cisco signalling greater stability. The result is that custom document systems are viable for companies with revenue from EUR 500,000 upward, with a measurable return within 18 to 24 months.

The cost of silos is rising with the tools that create them, and [the signals of the software shift](/en/ecorner/2026/software-industry-shift.html) explain why the bill keeps growing.

## Frequently asked questions

**Why do document projects fail?**
About 60% fail because the existing operational flows are not understood, rather than because of technology limits. Mapping how the company works comes before choosing software.

**What does fragmentation cost?**
Team members spend about 20% of the working day searching for documents. A team of five at EUR 30 per hour loses more than EUR 54,000 per year in search time.

**What changes with a custom system?**
Metadata are built on the company's real workflows, so a search returns the relevant documents in seconds instead of hundreds of generic files.

**Is custom development affordable for a small company?**
Falling hardware costs, mature databases and a consolidating supplier market make it viable from around EUR 500,000 in revenue, with a return within 18 to 24 months.

## Sources

Politecnico di Milano — Digital Innovation Observatories — https://www.osservatori.net/
Politecnico di Milano — Digital B2B Observatory — https://www.osservatori.net/it/ricerche/osservatori-attivi/digital-b2b
Nvidia — Vera Rubin platform — https://www.nvidia.com/
AMD — data center processors — https://www.amd.com/
PostgreSQL — relational database — https://www.postgresql.org/