Part of the Insights series on the Mobtag platform.

In short: Feature checklists that ignore field conditions and one ID per item are how you get a back-office go-live and a site that reverts to the clipboard inside a quarter.

RFPs love integration bullets. Spend serious time on gloves, rain, poor signal, and thirty seconds to the right record. If a vendor’s happy path assumes perfect Wi-Fi and a QWERTY desk, you will pay for shadow processes you cannot defend in a review.

When the demo is better than the job

Demos are choreographed; a shutdown in the rain is not. Before you debate roadmaps, look for friction:

  • Hand the phone to someone in PPE and ask them to find a known serial without your demo script.
  • Run a form with required photos on a connection you throttle to a bad day—not conference Wi-Fi.
  • Ask where pass/fail lives if the PDF is only a snapshot: structured data should drive search and trends, not a late OCR project.

The non-negotiables to test in a demo (not a slide)

  1. Open a record the way a tech actually works: standing, PPE, maybe one hand. Is the “must see” data above the fold?
  2. One front door to the item: a stable URL or QR, not five routes into five half-empty tables.
  3. Forms and photos on the object, not a side channel that gets re-keyed on Friday night.
  4. Exports to decisions—CSV, PDF, or a partner system—without rebuilding the official number in Excel.
  5. Map or site context if leadership ever says “we don’t know where the spare is” in a meeting you care about.
Question to ask the vendor Why it matters
“Show me a cold start on site, no Wi-Fi, then sync.” Offline and patience are not edge cases.
“What happens to history when a cert is re-issued?” You need supersedence, not delete-and-mystery.
“Who can change a closed job?” Audit is about who did what after the fact.
“Show me the same truth on mobile and web for one serial.” Parity or you inherit two systems of record.
“What breaks if we add twenty percent more assets next year?” Scale surprises land in year two, not on the PO line.

TCO: what the licence price hides

Licence rows rarely include re-keying, reconciliation, or the person who stays late to fix exports before a board pack. Price data quality, training, and governance the way you price downtime—they are not “soft” if a regulator or insurer asks once.

  • How many people touch a record between site and report? Every hop is tax.
  • Do integrations need a perpetual middleware project, or are events and IDs boring enough to own?
  • Can the same team admin the system without inventing a second full-time role?

Design a pilot that can fail honestly

A pilot should end with a number you trust—time to open a record, share of jobs closed in-system, re-key events per week—not a slide that only says "positive sentiment."

  1. Pick one site or one work type, not “all generators” on day one.
  2. Name success metrics in advance and publish them to sponsor and field lead alike.
  3. Run long enough to include a bad weather week and a normal one.
  4. Put the sponsor on a rota for feedback, not only in the steerco deck.
flowchart LR P[Scope: one site or work type] --> M[Measure time quality re-key] M --> W[Include bad weather week] W --> D[Scale cut or double down]

Red flags in the sales story

You hear Worry
“Unlimited users” with no governance model Accidental over-sharing or uncontrolled fields you cannot retract.
“We’ll fix that on the roadmap” for core field work You are funding their product backlog, not buying yours.
“Just export to Excel for that report” Excel becomes a second system of record.
Every integration is “custom” Capex that never stops.
“Training is in the first month only” Field habit is a year-one cost, not a one-off event.

Summary: The right system survives an ordinary bad Thursday on site and still behaves like a database when the board and the regulator ask in the same week.