Field reporting software is built in warm offices by people who have never spent a February morning on a flat roof in Aberdeen. So it assumes warm dry hands, good light, and ninety spare seconds. Real sites offer none of those. Voice notes site reports flip the problem on its head: the engineer speaks, the record writes itself, and the result is faster, fuller and more honest than anything typed in gloves.
Guide · How-to
Speak your site report.
Typing on a touchscreen in work gloves at 7am produces one kind of report. Speaking naturally into your phone while you work produces a better, more complete one.
Free forever: 20 reports a month. No card, no trial clock.
The problem
Typing fails on a cold site
Work gloves do not register on capacitive touchscreens. Cold hands lose fine motor control and typing speed drops below 10°C. Bright outdoor light makes displays hard to read. A cold engineer entering 19.5 mbar often hits 1.95, 195, or leaves it blank. It is Tuesday morning on every commercial site in the UK.
Why memory loses detail
Reports written from memory
Most reports are completed in the van or in the evening, hours after the observation. Specific figures degrade fastest. "Boiler pressure within tolerance" replaces "working pressure 1.4 bar cold, within 1.0-1.5 bar range." The engineer knows what they saw; the record does not capture it.
What speaking produces
Speak, transcribe, save
- 1
Capture at the gauge
The reading spoken into the phone is the reading on the gauge. No gap between observation and recording.
- 2
More detail, no extra effort
Speech at normal pace produces complete entries. A spoken boiler entry holds six times the detail of a typed one.
- 3
Transcribed automatically
In Quickler the voice note transcribes and saves as the entry. Accuracy above 95% for clear speech, corrected by a follow-up if needed.
Better records, less effort
In WhatsApp, before you leave site
Engineers speak their findings. Quickler transcribes automatically and generates the PDF. The honest, complete record you would have written with unlimited time.
The short version
- Speaking beats typing in field conditions by a wide margin, especially once gloves are involved.
- Voice notes captured during the job produce more accurate records than entries typed from memory hours later.
- Spoken descriptions hold far more detail than typed entries written under time pressure.
- Voice transcription field reporting now runs above 95% accuracy for clear speech. Engineers do not need to review transcripts.
- The final record is a written text entry, legally equivalent to a typed one, captured from a voice input.
- A speak to report app like Quickler runs hands free site inspection through WhatsApp, with no new interface to learn.
The problem
Typing fails on a cold site
Work gloves do not register reliably on capacitive touchscreens. Some work intermittently. Most do not work at all on standard smartphone glass. Take the gloves off and you have already added an inconvenience before typing a single character.
Cold hands lose fine motor control. Typing speed drops below 10°C and accuracy drops further. An engineer entering a pressure reading of 19.5 mbar while cold and squinting at a grey sky hits 19.5 some of the time. The rest of the time it lands as 1.95, 195, or nothing.
Bright outdoor light makes displays hard to read. Direct sun makes them near impossible. An engineer who cannot see the question defaults to the nearest plausible answer. None of this is unusual. It is Tuesday morning on every commercial site in the UK.
Why memory loses detail
Reports written from memory leak the facts
Most field reports are not finished on site. They are finished in the van, at the end of the day, or in the evening. The engineer saw the condition at 9am and writes it up at 6pm, after six more jobs, eighty miles, a layby sandwich and a call about last week's comeback.
Memory is not a recording medium. Specific figures degrade fastest: pressure readings, fault codes, crack widths, clearance distances. Context goes next. Vague impressions survive longest and help least.
So you get "boiler pressure within tolerance" instead of "working pressure 1.4 bar cold, within 1.0 to 1.5 bar operating range." You get "crack noted in eastern wall" instead of "hairline crack, 0.2mm width, running diagonally from window head to soffit over roughly 600mm." You get "tyres ok" instead of "near-side front tread 4mm, off-side front 3.5mm, both within action threshold." The engineer knows what they saw. The record does not.
What speaking produces
Speak at the gauge, capture the gauge
Speak a voice note at the point of observation and the record changes in a measurable way. The reading you say into the phone is the reading on the gauge. The crack width you describe is the one you are crouched beside. No gap between seeing and recording.
Speech also carries more than typing under pressure. A typist abbreviates to save time. The same engineer, speaking at normal pace, describes everything they see for no extra effort. Compare a cold typed entry with a spoken one:
Typed, gloves off, in a hurry: "Boiler checked ok. Flue clear."
Spoken, standing at the boiler: "Boiler's a Vaillant 831, about 7 years old. Tested to operating pressure 1.3 bar, comfortable. Gas rate within tolerance. Flue terminal at the rear, clear, good draw. CO reading 3ppm at the terminal. Annual service sticker from 2024 still on the case."
Both took roughly the same time. The spoken entry holds six times the information. It is the better record by every measure that matters.
Accuracy
How transcription handles trade vocabulary
The real question about voice note compliance report tools is how they cope with technical vocabulary: model numbers, readings, fault codes, trade terminology. The 2026 answer is: well enough for standard field use.
Common trade terms transcribe cleanly. "Ecotec Plus 831", "near-side rear", "operating pressure 19 millibar", "C1 code", all handled correctly. Unusual phrasing may need a quick correction, but for gas engineering, electrical inspection, van checks or structural surveys, accuracy above 95% is typical.
The engineer does not review transcripts. In Quickler the voice note transcribes automatically and saves as the entry for that field. If it is wrong, a follow-up message fixes it. Most of the time nothing needs fixing. They speak, the record saves, they move to the next question.
The honest record
Why spoken records are more defensible
Voice notes produce a more honest record, and this is the subtler case beyond speed and accuracy. An engineer typing knows they are producing a document, and that awareness invites sanitising. Borderline findings become "within tolerance." Uncertain observations vanish. The record reflects what feels safe to write, not what was actually seen.
Speaking feels less formal, so the description stays natural and stays accurate. Borderline reads as borderline: "tread's getting close, I'd say 3mm, worth keeping an eye on." Uncertain reads as uncertain: "couldn't get a clear look at the flue terminal from ground level, worth a closer inspection next visit."
These are better records. In an insurance claim or a regulatory review, a record that accurately states a borderline finding and the decision made about it is far more defensible than one retroactively tidied to say "satisfactory." Speaking during the job produces the record the engineer would have written with unlimited time and zero documentation anxiety. That is the record worth keeping.
End to end
How Quickler runs it through WhatsApp
Quickler's workflow runs through WhatsApp. At each question the engineer replies with text, a photo, or a voice note, using WhatsApp's own microphone button. No separate recorder. No new behaviour to learn.
The voice note transcribes automatically and stores as the entry for that question. The engineer does not review it unless they want to. The workflow moves on. At the end the PDF report generates from every answer, typed, photographed and spoken alike, and looks identical regardless of input. Nothing in the final PDF reveals which entries were spoken.
From the second job onwards, describing what you see out loud is simply faster than encoding it into a touch interface in the field. The records are more complete, more accurate, and take less effort to produce. The argument for voice is not complicated.
This works for any inspection or checklist, not just one trade. Rooted in UK compliance, it works anywhere WhatsApp does, across English-speaking markets including the UK, Ireland, the United States, Canada, Australia and New Zealand. Quickler works in any country.
Questions, answered
Can voice notes be used for official site reports in the UK?
Voice recordings alone are not a formal written record. Voice notes that are automatically transcribed into a structured written record are legally equivalent to typed entries. The transcription captures what the engineer observed at the moment of observation, which is often more detailed and accurate than a typed entry completed from memory later. The final record is written text, not the voice note itself.
How accurate is voice note transcription for technical field reports?
Modern transcription systems handle technical vocabulary well, particularly common trade terms. Accuracy rates above 95% are typical for clear speech in a reasonably quiet environment. Engineers can correct errors before submitting a record. The practical accuracy of a voice-transcribed record is consistently higher than a typed record completed from memory two hours after the job.
Does typing in gloves actually cause errors in site reports?
Yes. Touchscreen interfaces register incorrect inputs when used with work gloves, causing mistyped numbers and missed fields. In cold conditions fine motor control deteriorates and typing speed drops. Both effects produce entries that are abbreviated, contain errors, or are omitted entirely. A cold engineer in gloves and a hurry defaults to minimum entries, and minimum entries produce minimum-quality records.
What happens to voice notes after transcription in Quickler?
Voice notes sent via WhatsApp are transcribed automatically and stored as text entries against the relevant workflow question. The engineer does not need to review or retype the transcription. The final PDF report includes the transcribed text. The final record is a complete, structured document regardless of whether individual fields were answered by text or voice.
Related guides
Field compliance