How Much Does Utility Network GIS Software Cost in 2026?

What a utility network GIS build actually costs

The Esri, Smallworld or Bentley licence is a separate negotiation with the platform vendor and it is not what this page prices. This page prices the custom layer that sits around the platform: the editing behaviours, the job and posting workflow, the validation gates, and the feeds that carry your model out to OMS, ADMS, planning and work management. Across the 2,000-plus projects Digital Heroes has delivered, that layer falls into three bands.

• Single slice: $45,000 to $90,000, 6 to 10 weeks. One problem only. Redline intake that reads a photographed marked print and proposes an edit set, or one downstream export projection with a run log and a nightly diff report for the consuming team. Right when posting is basically working and one handoff is eating the hours.
• Workflow layer: $90,000 to $200,000, 14 to 20 weeks. Job and version management on branch versioning, as-built posting with a gate that refuses a version carrying topology errors on a live feeder, redline intake, triage and reimplementation of the twenty to thirty legacy customizations editors actually touch, and the first two downstream exports.
• Full program: $250,000 to $700,000, 9 to 18 months. The workflow layer plus conversion tooling with repeatable per-feeder error reporting, offline field capture producing proposed edit sets rather than direct topology edits, subnetwork exports for OMS and ADMS, a CIM projection, and integration into Maximo or whatever work management system holds your asset register.

Utilities usually call us describing the full program and need the workflow layer first, because conversion tooling has nothing stable to convert into until posting is fast.

What pushes the number to the top of the band

• The size of your customization inventory. Auto Updaters have no direct equivalent on the utility network, so each surviving behaviour becomes an attribute rule in Arcade plus server-side validation, with a different failure mode that has to be tested. Utilities we work with typically find between 60 and 300 individual customizations. Triaging that list honestly is the cheapest hour in the project.
• Gas alongside electric. Pressure systems and the UPDM model are a second domain with their own tracing semantics, not a second layer on the electric build. Scope it as its own increment or the estimate will drift by six figures.
• Multiple operating companies. Two construction standards books means two compatible sets of editing behaviour, two validation rule sets, and a permanent argument about which one the shared tooling defaults to.
• Model verification debt. If a third of your secondary was never digitised or phase was populated three different ways by three past conversion vendors, you are buying a field verification program with a software project attached. No amount of tooling invents connectivity that was never recorded.
• The number of downstream consumers. Each of OMS, ADMS, CYME or Synergi Electric, CIM and work management wants a differently shaped projection of the same topology. Every additional consumer is an integration with its own acceptance testing and its own team who will call at 2am when the extract is wrong.

What pulls the number down

• Deleting customizations instead of porting them. Roughly half of a typical inventory has not been used since the editors who requested it retired. Every one you retire is build and test time you do not pay for.
• Piloting on two feeders and one district. Running the whole loop, a real job posted and a real OMS handoff, on a small footprint before scaling costs less than converting everything and then discovering the posting gate blocks a workflow nobody documented.
• Leaving offline editing out of phase one. Field capture that produces a proposed edit set for a technician to validate is far cheaper than crews editing topology in a version, and it is the pattern that holds up anyway.
• Owning your own asset package. If your target schema is already defined against an industry data model, discovery drops by weeks compared with designing it inside the project.

A worked example that adds up

An electric utility with roughly 180,000 meters, one operating company, an as-built backlog running about 90 days, ArcFM Solution behind it and 140 catalogued customizations of which 26 survive triage. Scope is the workflow layer with two downstream exports.

• Discovery, customization inventory and triage with the editing team: $18,000
• Job and version queue with states, assignment and aging: $42,000
• Redline intake with vision-assisted extraction from photographed prints: $34,000
• Validation gate and error inspector integration on affected feeders: $22,000
• Export service with OMS and CYME projections, run log and diff report: $46,000
• Pilot on two feeders, technician training and acceptance: $16,000

Total $178,000 over 18 weeks. That sits in the upper half of the workflow band because the customization count is high and two exports are in scope. Drop the CYME projection and pilot with fewer surviving behaviours and the same utility lands near $135,000.

How the spend phases across the project

• Discovery and triage, 8 to 12 percent. Paid up front, and the phase that decides whether the rest of the estimate holds.
• Job workflow and posting, 35 to 45 percent. The largest block, because it carries the validation logic your operations group will lean on.
• Downstream exports, 20 to 30 percent. Scales linearly with consumer count, so this is the line you control by sequencing.
• Pilot, training and cutover, 10 to 15 percent. Underspent more often than any other line, and the reason some migrations technically ship without changing the backlog.

The ongoing costs nobody quotes

Budget 15 to 20 percent of the build cost per year for the custom layer alone. Utility network tooling carries recurring work that a fixed-price build does not cover.

• Platform release compatibility. ArcGIS Pro and ArcGIS Enterprise ship new releases on a regular cadence, and add-ins, attribute rules and trace configurations get retested against each one you adopt. Utilities that skip two releases pay for the catch-up in one lump.
• Export contract maintenance. Your OMS or ADMS vendor upgrades and the projection that fed it cleanly for two years starts rejecting records. Someone owns that, and if it is not in a support agreement it is an emergency.
• Hosting and compute. Validate topology on a large network is not free, and dirty area processing across a full service territory has a real infrastructure cost separate from your Esri licensing.
• Editor training as staff turn over. The workflow layer exists so a new technician is productive on day two. That only stays true if the training material is maintained when behaviours change.
• Data remediation that never ends. Phase and normal state cause more downstream arguments than every other attribute combined, and keeping them clean is a standing job, not a project task.

What the build price does not include

Four lines sit outside every custom quote in this category, and each one has arrived late in a budget conversation somewhere.

• Platform licensing. Named user and server licensing for Esri, Smallworld or Bentley is negotiated with the platform vendor and is usually the larger recurring number. A custom quote that appears to include it has misunderstood the question.
• Field verification. If your secondary was never fully digitised or phase was recorded inconsistently, someone has to go and look at the system. That is a crewed field programme with its own schedule, and no editing tool substitutes for it.
• Data conversion labour, unless it says so. Ask directly whether iterative conversion with per-feeder error reporting is in the price. It is often the larger half of a migration program and the easiest line for a bidder to exclude quietly and win on.
• Your own people. A GIS supervisor, a standards engineer and at least one senior editor need real hours across the whole build. Their absence is the most common reason a delivered workflow layer ships on time and the as-built backlog does not move.

When not to spend this money

If you are a co-op under roughly 40,000 meters with a stable system, one GIS technician and a posting backlog measured in days, do not build. Buy the platform, buy a configured vertical solution, and put the money into verifying what is actually in the field, because a perfectly engineered workflow around a model that was guessed is an expensive way to be wrong faster.

Also skip it if your customization inventory is small and mostly cosmetic. Some utilities discover during triage that they have eleven behaviours their editors care about, all of which are expressible as attribute rules in a few weeks. That is a configuration engagement, not a build, and any developer telling you otherwise has not counted.

How to sanity check a quote

Ask what the quote assumes about your customization count and whether anyone has actually inventoried it. A number produced without that count is a guess wearing a spreadsheet. Ask what the build does when a technician tries to post a version with unresolved errors on a live feeder, and what the OMS receives that night; if the export simply runs, the quote is missing the validation work that makes the rest worth having. Ask whether conversion is inside or outside the price, because data conversion is usually the larger half of the program and it is the easiest line for a vendor to leave out and win on price. Finally, get code and configuration ownership in writing before kickoff. Utility GIS dependencies outlive the people who signed them.

If you want a second opinion before signing anything, Digital Heroes builds and runs its own products, so the people choosing your architecture live with those decisions on their own revenue. Nothing about that commits you to the build.