Gas Leak Survey Software Problems: The 7 That Cost Real Money, and How to Avoid Them

Why does the build get scoped around routes instead of pipe?

The biggest scope failure here is building the system around the artefact the crews already use. Routes are how the work is dispatched, how surveyors think, and how the paperwork has always been organised, so a developer sitting with a survey supervisor will hear routes described first and will model them faithfully.

The problem is that the compliance question is not asked in routes. An inspector asks whether every main and service in a district was surveyed within its required cycle, and that is a question about pipe. A route is a proxy for pipe that was accurate on the day it was drawn and decays quietly afterwards, because the system changes constantly and route definitions do not.

Digitising routes therefore produces a faster version of the same blind spot. The record looks better, coverage still cannot be proven, and the reconstruction exercise survives with nicer inputs.

The fix is to make the pipe segment the unit of coverage and treat the route as a work assignment on top of it. Segments carry effective dates, so a main replaced in June creates newly uncovered footage automatically, and an abandonment drops out of the denominator on the date it was abandoned. Then the annual scramble becomes a report you can run today. A useful test before you commit to anything: take one district, export the segments from your GIS, and try to prove last cycle's coverage against them. However long that takes is the size of the problem you are buying out of.

What goes wrong when GIS pipe data is the foundation?

Segment based coverage is only as good as the pipe data beneath it, and most distribution systems carry more uncertainty there than anyone says out loud in a project kickoff.

Service lines are the usual weak point. Many are recorded as approximate locations or as a connection with no geometry, particularly older ones installed before as built discipline improved. Main records can carry material and vintage that were inferred rather than observed. Abandonments are sometimes recorded as a status change and sometimes not recorded at all, so pipe that no longer exists still sits in the denominator, and pipe that was never mapped sits outside it. Then there is the boundary problem: segments that cross district or jurisdiction lines, and segments that were split or merged during a GIS update, which quietly breaks any coverage history keyed to the old identifiers.

The fixes start before the software. Run a data readiness assessment as a discrete piece of work rather than assuming it, and price the remediation separately, because a project that discovers this in week six loses its schedule. Key coverage records to a stable segment identifier that survives splits and merges, and record the lineage when a segment is split so history follows both halves. Where service line geometry is genuinely unknown, represent that as a known gap with an owner rather than as coverage, because an honest gap is a work item and a false coverage record is a finding.

Why do the detection and work management integrations break after launch?

Modern detection is the best thing to happen to this work in decades, and it is also where the integrations drift.

Providers deliver indications in their own structures on their own cadence. A mobile survey contractor sends a batch after a run, an aerial provider sends results for a flown area on a different schedule, and a walking crew's instrument produces a log in a manufacturer format. When a provider updates their platform, the export changes shape, and the failure is usually partial rather than total: indications keep arriving and one attribute stops populating, so confidence scores or concentrations quietly go missing while everything looks healthy.

Work management integration breaks for organisational reasons more than technical ones. A repair order created in the work system by a dispatcher who did not start from the leak record has no leak reference on it, so the clock and the job are two different objects. Then somebody changes a work order type or adds a status, and the closure event your leak record was listening for stops arriving.

The controls that hold this together are the same ones that hold any field integration together. Validate every inbound feed against an expected shape and alert on missing attributes, not just on failed transfers. Reconcile counts per provider per area on a schedule, because a mobile run that returned no indications and a mobile run that failed to deliver look identical from inside your system. Make the leak record the parent of the repair order rather than the sibling, so a closure cannot happen without the leak knowing.

What happens when re-checks, regrades and multi state rules are not covered?

Three gaps recur, and each converts a well run programme into a compliance risk nobody can see.

Re-checks are the first. A leak that is not repaired immediately has to be re-checked on a schedule while it waits, and those re-checks are the part most often tracked in a separate spreadsheet. When a re-check produces a worse reading, the leak needs to be regraded and its clock needs to shorten, and if that escalation depends on a person noticing, it will not always happen.

Grading consistency is the second. Grades turn on readings, position relative to structures and migration potential, and two competent surveyors can reach different conclusions from the same site. On paper in a district office, that inconsistency is invisible. Capturing the observations the definition depends on, proposing a grade from them, and recording overrides with reasons turns a pattern of one crew grading consistently low into a training conversation instead of an audit finding.

Multi state operation is the third. Survey cycles, grading definitions and repair timelines vary by jurisdiction, and a system that hard codes one state's rules produces confident deadlines that are wrong across the line. Each jurisdiction needs to be its own rule set, and a segment needs to know which one applies to it. Grade 1 leaks are the one universal: immediate and continuous action until the hazard is eliminated, and no rule engine should ever be able to schedule one for later.

Should you build custom or configure what you already own?

If you are a small municipal system with a few hundred miles, one survey crew, one state and a manageable backlog, do not build. A disciplined spreadsheet and a good supervisor genuinely cover it, and the budget does more good buying better detection equipment.

Before assuming custom at larger scale, check what is already available to you. Heath's OPTIMAIN is the closest packaged survey management product in the market, and it is a reasonable fit if your programme runs largely through one provider's instruments and services. Your GIS platform may already support field data collection against segments, which is a meaningful portion of coverage proof without a new system. Your work management system may already allow a custom field for a leak reference, which fixes the orphan repair order problem for the price of a configuration change.

The build case appears when two or more of these are true. Coverage cannot be demonstrated by segment without manual reconciliation. You run two or more detection technologies whose outputs sit in separate systems. You operate across more than one state and therefore more than one rule set. Repair deadlines are not visible to whoever schedules crews. Grading inconsistency exists and nobody can quantify it. Or replacement prioritisation runs on vintage alone because leak history is not usable at segment level.

How do hidden costs get into the quote?

Leak survey quotes go wrong in five familiar places.

• GIS remediation. Unmapped services and approximate locations have to be addressed before segment based coverage means anything, and this is a separate project with its own budget.
• Number of state jurisdictions. Each set of survey cycles, grading definitions and repair timelines is its own rule model, not a lookup table.
• Detection vendor diversity. Every provider delivers indications in its own structure on its own cadence, and each is real integration work.
• Offline field capability. Reliable offline operation with conflict handling on sync costs meaningfully more than an online form, and skipping it guarantees abandonment.
• Work management integration. Straightforward technically and slow organisationally, because it crosses departments with different priorities and different change windows.

Digital Heroes delivery experience puts a first release covering GIS based coverage proof, offline field capture with grading logic, repair clocks by grade and instrument log ingestion at $70,000 to $140,000 over 12 to 16 weeks. A full platform adding detection data fusion, unaccounted for gas and methane reporting, repair work management and replacement prioritisation runs $180,000 to $420,000 across 6 to 12 months.

What separates a build that works from one that fails here?

Working builds are usable for a full shift with no signal. Survey happens in places where connectivity is unreliable, and a tool that stalls in a basement or on a rural run gets abandoned in a fortnight and replaced by paper that gets keyed in later, which is the process you were trying to leave.

They make the current job easier, not just the reporting. Showing a surveyor that this address had a Grade 2 last year at the moment they arrive changes how they investigate, and it is the single feature that most reliably wins field adoption. Tools that only serve headquarters get worked around.

They put deadlines where scheduling decisions are made. Leaks are repaired late not by decision but because nobody saw the clock when the work was slotted. The useful view for a supervisor is the next fourteen days sorted by deadline and grade with crew availability beside it, not a list of open leaks.

And they settle ownership in writing before kickoff, covering the repository, the database, the cloud accounts and a working export path. Leak records are regulatory evidence with a long retention life and they feed replacement decisions for decades, so access to them cannot depend on a vendor relationship staying healthy.