A product behavior or export was observed in an authorized account. Direct does not mean complete, passed, or recommended.
Field service software evidence
Every conclusion has a visible boundary.
Inspect the test scope, observed result, market, last check, and unresolved proof gate before relying on a review or shortlist.
Register reviewed / Sep 20, 20265 direct records3 full scenario registers. 0 numerical scores.
How to read the ledger
Evidence state is not a rating.
A polished demo, an official feature page, and an observed exception are different forms of evidence. This register keeps them separate.
Current primary sources define the claim and the test plan. No hands-on product result is inferred.
An eligibility, region, account, or device gate prevented the intended direct run. The block is evidence about access, not product quality.
Product evidence records
Observed result, then the next proof required.
Latest underlying check in this register: . Each row retains its recorded check date. The editorial review is not a new product test or a confirmation that every vendor fact remains current.
Scroll sideways to see all five columns. On a keyboard, focus the table area and use the left and right arrow keys.
| Product and state | Scope and observed result | Next proof gate | Market and last recorded check | Record |
|---|---|---|---|---|
| Jobber Direct | S01 through S08, 48 controls 7 Pass, 24 Conditional, 0 Fail, 17 Blocked. No score. | Resolve material role, device, messaging, finance, support, and exit blocks before any score. | Canada | Open record ↗ |
| ServiceM8 Direct | S01 through S08, 48 controls 2 Pass, 21 Conditional, 0 Fail, 25 Blocked. No score. | Use controlled field devices, a second role, and safe finance and export sandboxes. | International account | Open record ↗ |
| Tradify Direct | S01 through S08, 48 controls 1 Pass, 24 Conditional, 0 Fail, 22 Blocked, 1 Not run. No score. | Run the safe deposit control and resume field checks only with controlled iOS or Android access. | European account | Open record ↗ |
| Zoho FSM Direct | Mobile notes, offline cache, and EUR invoice flow Observed note sync, cached offline reading, and a manual EUR 1 partial payment. Not a full 48-control run. | Test two-user handoff, offline write conflict, refund or write-off correction, and complete exit. | European account | Open record ↗ |
| Jotform Direct | 12 intake, delivery, and portability controls 10 Pass, 1 Conditional, 1 Fail. One administrative alert and one PDF and CSV record passed. No score. | Test a live sandbox handoff, real-device offline use, role boundaries, delivery failures, and complete account exit. | Global web account | Open record ↗ |
| GorillaDesk Documented | Official pricing, trial, offline, accounting, import, and change-log sources mapped to a controlled workflow protocol One-route pricing and documented product limits were rechecked. No hands-on workflow result, score, savings, or customer outcome is claimed. | Run the 25-stop Basic and Pro routing comparison, weak-connectivity replay, accounting recovery, component export, and post-cancellation retrieval checks. | North America | Open record ↗ |
| Housecall Pro Access blocked | Official documentation and observed eligibility gate The access boundary was observed. No authenticated workflow result or score exists. | Obtain a vendor sandbox, demo, or authorized US or Canada contributor. | US and Canada | Open record ↗ |
| Workiz Access blocked | Official documentation and observed +1 phone gate The signup gate was observed. No authenticated workflow result or score exists. | Obtain vendor clarification or a supported-country contributor without using inaccurate details. | United States | Open record ↗ |
| ServiceTitan Documented | Official onboarding, export, pricing, and terms sources Desk 1.0 only. No hands-on result, normalized price, savings claim, or score exists. | Obtain a dated normalized quote and an authorized sandbox before adding price or product conclusions. | North America | Open record ↗ |
Pass means the named control produced the required retained evidence. Conditional means a documented plan, role, add-on, device, or workaround boundary changed the result. Blocked means the control could not be completed safely or comparably. None of these counts is a universal product score.
Bounded direct observations
Small facts, full context attached.
An observation proves only the recorded action and result. It does not fill an untested device, role, market, finance, or exit gate.
Offline cache and note recovery
Observed on a connected mobile account, then checked after returning online. Full device coverage remains unresolved.
Read the Zoho recordEUR 1 manual payment
The invoice balance moved from EUR 125 to EUR 124. Gateway payment and refund paths remain untested.
Read the Zoho recordInitial AI branch omission
The initial AI-generated form omitted the emergency follow-up branch. That observation does not prove a later correction.
Inspect the bounded resultOne-row CSV control
A fictional customer export was reconciled. Attachments and a complete archive remain outside the observed result.
Read the Tradify recordCite a bounded observation
These summaries describe retained tests, not new runs. Original account records remain private. Editorial review: .
Observed /
The initial AI build omitted the emergency branch
- Test context
- One generated plumbing service-request form in a verified Starter account. Synthetic data only; phone-width browser preview, not a native-device field test.
- Observed result
- Selecting Emergency did not reveal the requested follow-up questions. An empty submission produced nine required-field errors.
- Limitations
- This records the initial omission. It does not prove that the emergency branch was subsequently corrected, or that every branch and device works.
- Evidence reference
- Retained initial-generation, branch-selection, and validation notes. The missing branch remains an open control in the test record. Read the full test context.
Suggested citation: ServiceTech Signal, “The initial AI build omitted the emergency branch”, observed Aug 23, 2026, editorial review Sep 20, 2026. Permanent observation link.
Observed /
One administrative alert was received and reconciled
- Test context
- One additional authorized fictional request and the configured administrative mailbox. Form and submission identifiers plus field contents correlated the records.
- Observed result
- The request was accepted at 19:54 and the matching administrative email reached the mailbox at 19:55. Jotform's email preview had a six-hour timestamp discrepancy against the other records.
- Limitations
- One received message is not a delivery-rate benchmark. Customer autoresponder receipt, bounce recovery, and repeated failure cases remain untested.
- Evidence reference
- Retained submission, sent-log, and received-message reconciliation. Private identifiers, recipient details, and original messages are not published. Read the full test context.
Suggested citation: ServiceTech Signal, “One administrative alert was received and reconciled”, observed Sep 10, 2026, editorial review Sep 20, 2026. Permanent observation link.
Observed /
One CSV had 18 columns but no embedded record identifiers
- Test context
- Validation of the fictional 23 August record: one UTF-8 CSV with 18 unique columns and one readable, one-page A4 PDF.
- Observed result
- All populated values reconciled. The CSV omitted form ID, submission ID, and time of day. The PDF filename retained the submission ID, but the document did not retain the form ID.
- Limitations
- File provenance remains necessary. These single-record outputs do not prove a complete backup, attachment export, or successful account exit.
- Evidence reference
- Retained file parsing and field-level reconciliation. Original exports remain private; this is a sanitized editorial summary, not a downloadable raw test artifact. Read the full test context.
Suggested citation: ServiceTech Signal, “One CSV had 18 columns but no embedded record identifiers”, observed Sep 10, 2026, editorial review Sep 20, 2026. Permanent observation link.
Use the evidence
Move from claim to controlled decision.
Start with the category owner, test finalists on the same scenario, and normalize a dated written quote. These tools do not turn an incomplete record into a recommendation.
Contribute evidence
Challenge the record with reproducible facts.
Corrections, current pricing, and regional observations enter a manual editorial queue. Vendor identity and commercial relationships must be disclosed.