Public Works Software: Custom Build or Off-the-Shelf

What Cityworks, Cartegraph and Lucity actually do well

Every argument for building starts by being unfair to these products, so let us not. Trimble Cityworks was designed around Esri from the beginning, which means your ArcGIS layers are the asset register rather than a copy of it, and for a department whose geodatabase is in decent shape that is a real head start. OpenGov Cartegraph is the friendlier of the two for a smaller streets and parks operation and gets a condition-based programme running without a consultant living in your office. CentralSquare Lucity and Brightly Asset Essentials both cover the maintenance management ground competently.

They give you a work order lifecycle, a preventive maintenance calendar, an asset hierarchy, inspection forms, and reporting your director can put in front of a council committee. They also carry things a custom build has to fund from scratch: an accessible citizen portal, hosting somebody else patches, and a user community that has already argued about your problem on a forum.

Most departments should buy. Under about 25 staff with a single division, or anywhere your county or state has standardised on a product and an exception would cost two years of procurement and information technology arguments, stay bought. And if your asset data is genuinely poor, buy nothing new at all this year. Spend the money reconciling asset identifiers between your work order history and your geodatabase, because that reconciliation is the prerequisite for every option including the one we sell.

Where they stop: the field discovery nobody can record

The gap is not features. It is direction of travel.

Cityworks and Cartegraph both integrate with Esri, and the integration assumes the geographic information system is authoritative while the field is a consumer. At 2:40am your crew excavates a main break and finds six inch cast iron from 1961 where the map says eight inch ductile from 1988. The crew leader is on hour fourteen and the close-out form wants eleven fields, so nobody records it. Six months later engineering designs a replacement sized for the wrong pipe. The software did not cause that. It made recording the truth harder than not recording it, which amounts to the same thing.

The second gap is intake. Requests arrive by phone to the clerk, through a SeeClickFix or Accela portal, by email to a shared inbox, and through a council member forwarding a constituent complaint. Each channel produces a differently shaped record, and every vendor's answer is to put everything through their portal, a sentence that has never survived contact with a resident who wants to speak to a person. So the same pothole becomes four tickets and your response time metrics are fiction.

The third gap shows up six days a year and costs the most. A four inch snow event or a hundred year rain produces four hundred requests in nine hours, dispatch falls back to a whiteboard and a group text, and none of it enters the system. When Public Assistance paperwork comes due, somebody reconstructs those events from photos and memory. A Category A or Category B claim needs documented labour hours, equipment hours against the applicable equipment rate codes, and location. If you cannot produce it, you eat the cost.

Fourth, the compliance record. Your Municipal Separate Storm Sewer System permit requires inlet inspection and cleaning evidence. Sign management under the Manual on Uniform Traffic Control Devices requires a maintained retroreflectivity method with records to match. Both live in spreadsheets in most departments because the licensed system does not reach the crews doing the work.

The arithmetic: named user seats against a one-time build

This category is priced per named user, and that single fact drives most of what goes wrong.

Do the sum with your own figures. Count everyone who should be touching a work order: the streets superintendent, the foremen, the sewer crew, the sign crew, the fleet shop, the dispatcher, the geographic information system analyst. Now count how many seats you actually bought. In most departments the second number is a third of the first, which is why the sign crew keeps a workbook and the foreman rekeys paper for forty minutes every morning. That forty minutes, times the number of foremen, times 250 working days, is the real licensing cost and it does not appear on any invoice.

Then the visible money. Take your annual licence, add the consultant retainer you keep to maintain configurations, and multiply by five. A department paying $50,000 a year in licence plus $30,000 in configuration support is at $400,000 across five years. Against that, a custom first release runs $60,000 to $130,000 with support at 15 to 20 percent a year, so a $110,000 build costs roughly $203,000 over five.

The crossover has two forms and either one is enough. On seats, it lands around 60 to 70 people who genuinely need access, because that is where per-user pricing stops being cheaper than owning software with free seats. On volume, it lands around 8,000 work orders a year across three or more divisions, because below that a well-configured product absorbs the load. The tell that you are already past it: you pay for a tool and still run the real operation in Excel or on a whiteboard.

What a custom build actually costs

A focused first release covering channel-agnostic intake with deduplication, the asset register with two-way geographic information system sync, offline-capable mobile work orders and dispatch runs $60,000 to $130,000 and ships in 12 to 16 weeks. A full platform adding fleet and equipment, inventory and parts, the public request portal, preventive maintenance scheduling against real crew capacity, event mode for declared disasters and finance integration runs $150,000 to $400,000 phased over 6 to 12 months.

Then the two lines that arrive after the quote. Migration runs 10 to 25 percent of build cost and sits at the top of that range here, because the work is not extraction, it is reconciling asset identifiers between your work order history and your geodatabase. Cityworks stores its data in SQL Server and you can query it directly, so the export is the easy part. Budget the reconciliation as its own phase before application development starts. Year two onward runs 15 to 20 percent of build cost annually for hosting, patching, schema changes on the Esri side and the enhancements that follow every winter.

What drives the number up: offline-first mobile is the single largest multiplier, because a truck in a river valley or a vault has no signal and a sync engine that survives two crews editing the same asset offline is genuine engineering rather than a checkbox. Esri integration scales with how clean your layers are. Financial system integration to Tyler Munis, Springbrook or Workday is knowable, but only once finance commits to a chart of accounts mapping in writing, and that conversation takes longer than the code.

The four situations where building wins

• Regulatory fit. Federal Emergency Management Agency reimbursement fields captured during the storm rather than reconstructed after it, stormwater permit evidence tied to the inlet record, and sign retroreflectivity history attached to the asset. These are records with deadlines, and a system that cannot produce them costs you money in a specific year.
• Scale economics. Past roughly 60 to 70 people who need access, or 8,000 work orders a year across three divisions, per-user licensing plus configuration retainer passes the build inside three years. Count the people you did not buy seats for.
• A workflow that is your competitive advantage. Public works has an equivalent: whatever is structurally unusual about your operation. A combined utility, a shared services agreement with three neighbouring townships, a snow operation that is forty percent of your winter, an internal fleet shop billing other departments. Off-the-shelf tools are built for the median department, and every year configuring around that gap is a year you pay software to make the job harder.
• Integration sprawl across three or more systems. ArcGIS, the financial system, the fleet fuel and telematics feed, and a citizen portal. Four interfaces each needing a vendor change request is where packaged licensing stops being the cheap option.

One signal alone is rarely decisive. Two together usually is, and the pairing we see most often is unlicensed field crews plus an unusual operating structure.

How to decide in a week

Do not write requirements. Pull data.

Export last year's citizen requests and run a crude duplicate check: same asset class, within fifty feet, within fourteen days. Whatever percentage comes back is the amount your response time reporting is currently overstating, and in departments we have measured this sits between twenty and thirty five percent. That single number changes the conversation with your council more than any demonstration will.

Second, pick one asset your crews touched in the last month and try to answer four questions from your systems alone: what is it, what was actually found when it was last opened, which ward and stormwater basin it sits in, and what it has cost over three years. Time yourself. Most departments can answer the first and none of the other three, because cost is stored per work order rather than per asset over time.

Third, give every developer on your shortlist your inlet layer and ask how they would model an inlet that belongs to a basin, sits in a ward, falls on a snow route and was replaced in 2019 under a capital project. A flat asset table with custom fields means they have never built this. Then ask what happens when two crews edit the same asset offline for six hours and both sync at 4pm. If the answer is last write wins, keep interviewing.

Then buy a paid discovery phase that ends with a written specification you own outright: the asset data model, the offline conflict rules, the chart of accounts mapping, records retention and portal accessibility criteria. Take it to every other firm on your shortlist. It is what makes four quotes comparable.

Digital Heroes will not write code without a signed product requirements document. We contract through India LLP, US LLC and UK LTD entities so intellectual property assigns under law your city attorney already reads, and the department owns the repository and the cloud accounts from the first commit. Over fifty specialists, more than 2,000 projects, a named team you meet before signing, checkable on Clutch, Trustpilot, Fiverr Vetted Pro and D-U-N-S.

We are the wrong firm for a department under 25 staff with one division, and wrong when the real problem is that nobody uses the system you already own. New software does not solve an adoption problem. It resets the clock on it.

Book a 30-minute call with Digital Heroes and get a written plan and a fixed quote within 48 hours.