The Engineer's Admin Problem: Why Field Service Teams Hate Their Job Software

Almost everything written about field service software is written for the person signing the cheque. Scheduling gains, first-time-fix rates, utilisation dashboards — all addressed to an operations director at a desk.

Nobody writes for the engineer.

So this is written from the other side of the van: the side where you are on a flat roof in February, one hand on a ladder, trying to close a job on a phone that has decided it needs your fingerprint again.

I wanted evidence rather than anecdote, so I pulled every public review I could legitimately retrieve for 14 platforms UK engineers actually carry, and counted the complaints myself.

The finding I did not expect

I set out to count complaint categories. The most interesting number came out sideways.

Two completely different kinds of people were writing these reviews. Some talk about my engineers, our business, we rolled it out. Others talk about my company makes me, forced to use it, no choice.

So I coded for it. Any review of 40 characters or more using buyer language but no mandated-user language went in one bucket; the reverse went in the other. Then I compared star ratings.

A 1.80-star gap on the same fourteen products, from the same pool, on the same day.

Be careful with what that proves. The sub-samples are small, role is inferred rather than verified, and people who review mandatory work software are an angry, self-selecting group — which is why the whole-sample mean is only 1.86 stars. No absolute figure here is a satisfaction score. But the comparison survives all of that, because both groups come from the same skewed pool. Whatever pushes mandated users down to 1.53 should push buyers down too. It doesn't.

One reviewer put it better than my regex could. A 1-star review of BigChange JobWatch, verbatim:

That is the whole problem in one sentence, written by someone who was not trying to make a point about software design. He was just annoyed.

What engineers actually complain about

Every review of 1 to 3 stars carrying at least 25 characters of text — 1,557 of them — coded against a fixed list of themes. A review can carry several, so percentages do not sum to 100.

Reliability, not usability, is the main event. 616 reviews — 40% of the sample — say the app crashed, ran too slowly to use, failed to sync, or simply did not work. Only 338, 22%, complain about workflow design: too many taps, too many forms, poor search, repeated logins.

I did not expect that ratio. The industry conversation is overwhelmingly about workflow — fewer fields, simpler job sheets, fewer taps. That is real, and it is Group B. But it is roughly half the size of the problem nobody puts in a webinar title: a great deal of this software is not stable enough to finish a job on.

Nor is it a legacy issue. Among the 759 reviews dated 2024 or later, crashes still lead at 14.8% and functions not working at 12.3%. It holds for the British products too: the eight UK-built platforms give 190 reviews in which crashes lead at 21.1%.

A warning before the next chart: do not use store scores as a shopping list. Google Maps scored 3.2 on the same day and storefront, and the twelve field service apps with a published Play score averaged 3.17 — barely below it. Every argument here rests on what reviews say, not on stars. One number is still hard to look away from, though: Dynamics 365 Field Service sits at 1.6 stars from 715 Google Play ratings and 1.32 from 72 UK App Store ratings.

The mobile UX failure pattern, corrected

Here I have to correct a story I have told myself, because the data does not support it.

The received wisdom is that field service software fails because it was designed in a warm office by someone who has never worn gloves in the rain. Wet screens, cold fingers, bright sunlight.

Across 2,814 reviews, gloves are mentioned once. Rain and weather, once. Sunlight and glare, once. That is not because the van is comfortable — it is because engineers never get far enough into the app to complain about the weather. You cannot be annoyed that the screen is hard to read in sunlight if the app has not finished loading.

Where the office-versus-van mismatch does show up, it is as specific, fixable decisions — all verbatim from reviews I retrieved:

• Biometric unlock that assumes bare hands. Totalmobile, 4 stars, June 2020: "facial recognition would be useful as can't use fingerprint whilst wearing PPE gloves." He liked the app. He still could not get into it in PPE.
• Crashing when the phone meets the van. Joblogic, September 2021: "Constant crashes every time I plug my phone into my van, and every time I unplug from the van." Nobody in an office plugs their phone into a Transit forty times a day.
• Safety paperwork blocking the job it protects. Joblogic, 2 stars, November 2022: "the app won't let me show that im onsite. it keeps saying I have to fill out the dynamic risk assessment but I have already done it."

That last is the pattern in miniature. Gating "I am on site" behind a form that will not register as complete turns a safety control into an obstacle, and teaches the engineer that the fastest route through the app is around it. That is how risk assessments end up done in a batch, from memory, in the van park on Friday — the same failure mode as Friday-afternoon CRM updates in field sales, with worse consequences.

Where the day actually goes — and why nobody can tell you

I wanted to open with the standard statistic: engineers spend some percentage of the day on tools and the rest on travel, paperwork and waiting. You have seen the number. I have quoted versions of it myself.

I could not verify it. Not one primary, methodologically transparent source.

The canonical survey is the Service Council's Voice of the Field Service Engineer, and both the 2024 and 2025 editions are member-only — no figure, sample size or method is public. Field Service News research sits behind a $45-a-month subscription (about £34; indicative, at $1.33 to the pound on 26 July 2026). The open scholarly literature contains no peer-reviewed wrench-time benchmark. Every freely available version traces back to a vendor blog citing another vendor blog.

So I will not give you a wrench-time percentage, and you should be suspicious of anyone who does without naming a report you can open.

What is verifiable is narrower: ONS records median total paid hours of 40 a week for electrical service and maintenance mechanics and repairers (ASHE 2025, Table 14.9a, 23 October 2025). Those are paid hours. Nobody publishes how many reach the tools.

What that costs, if you want a number

Since nobody will give me a measured figure, here is a transparent model. Change any input you disagree with.

ONS median gross annual pay for the same occupation is £41,662 (ASHE 2025, Table 14.7a). Add employer National Insurance at 15% above the £5,000 secondary threshold (HMRC, 2026-27 rates, updated 5 June 2026) — £5,499. That is £47,161 before pension, van, fuel, tools or phone: across roughly 1,840 productive hours, about £25.60 an hour, understated because it excludes the vehicle.

Now assume ten engineers, five jobs a day each, and fifteen minutes per job lost to the frictions in Exhibit 2: 2,875 hours a year, roughly £73,700. Fifteen minutes is modest — a Dynamics 365 reviewer claimed the app added "a minimum of an extra 30 minutes" to every work order, and a Salesforce reviewer "about 10 to 15 minutes onto every job." Engineers' estimates, not measurements.

The point is not the £73,700. It is that this cost appears in no system you own — not the licence, not payroll, not the P&L. It shows up as engineers finishing late, jobs closing thin, and second visits that should not have been needed: exactly the hidden manual cost nobody counts.

Offline is the dividing line

Only 3.0% of complaints explicitly name offline or signal. That looks small until you remember it is a precondition. An engineer with no signal and no offline mode does not write a review about offline mode — he writes one about the app not working.

The physical reality is worse than most office-based buyers assume. 83.58% of UK landmass has good 4G from all four operators; 96.21% from at least one, leaving 3.79% with none at all (Ofcom, Connected Nations update: Spring 2026, published 13 May 2026, data as at 1 January 2026). In Scotland only 71.11% has all four and 9.18% has none. In-vehicle 4G across UK roads runs at 86% to 92% by operator. The widely quoted 88% is not geographic coverage — it is each operator's Shared Rural Network obligation, and the two get conflated constantly.

So your engineer loses signal regularly, by design of the country — and what the app does about it is documented thinly. Of the eight platforms whose offline behaviour I tried to check, exactly one publishes specific, public documentation of what does and does not work offline: Microsoft. Simpro's help guide requires a customer login; Salesforce's help centre would not render; Joblogic's and Commusoft's knowledge bases did not resolve. On their own homepages Salesforce markets an "Offline-first Mobile Experience" and Commusoft "seamless offline sync", while Totalmobile, Klipboard, Joblogic and BigChange do not mention offline at all (all retrieved 26 July 2026).

Here is the irony. Microsoft — whose app is rated worst in my sample — is the only vendor telling you the truth in public, and the truth is unflattering. Its documentation states offline is not on by default: "An administrator has to configure the offline profile for the app before technicians can download the data." Its limitations page lists what breaks — "Inventory validation doesn't run without network connectivity", "Access to knowledge articles isn't available in offline mode", "The Export to PDF option isn't available while the application is in offline mode" (Microsoft Learn, updated 23 July 2026). Most striking, on the page about working offline: "The new mobile experience doesn't support working offline" (Microsoft Learn, updated 9 December 2025).

Beside it, the engineer's version — 1-star, UK App Store, 2 July 2026: "Absolute rubbish, doesn't work offline, or it seems to be, but then you realise when the signal comes back that you need to restart and fill in every job report you just did."

That is the dividing line. Not "does it have an offline mode" — nearly everything claims one. Three questions separate real offline from marketing offline: is it on by default; which functions stop working; and what happens to work already entered when the connection returns. Ask them in writing before you sign.

Credit where the data supports it

It would be easy to write this as a straight kicking, and that would be lazy. Of the 1,478 negative and mixed Google Play reviews in my sample, 476 (32%) received a public vendor reply — and the spread is dramatic:

A reply is not a fix, and several I read were boilerplate. But there is a real difference between answering 94% of your unhappy engineers and answering 4%, and the reply rate on negative reviews is a free thirty-second signal about whether the people building the app have contact with the people using it. Some products visibly got better, too. Simpro Mobile, 4 stars, UK App Store, October 2024: "The app was originally buggy and dropped data. It is now much better, all of our engineers find it easy to use."

What actually fixes it

Each item below is something the reviews explicitly ask for, and several are things a product in this set already does well.

Cut the fields before you buy anything. Most job sheets carry fields that exist because someone asked in 2019 and nobody has since checked whether the answer gets read. For each, name the report it feeds and the decision that report drives. Fields failing that test are pure tax on the engineer. It costs nothing — ghost workflow hunting applied to a form. Where engineers still work from paper or a manufacturer's PDF, parse the document into the back-end system rather than building a form for it.

Pre-fill everything the system already knows. The engineer should never type the site address, asset serial, contract reference, their own name or the date. If the job was dispatched to them, the system knows all of it. One Salesforce reviewer complained the app "doesn't remember basic things such as filling in your name." Any field typed twice is a defect.

Voice first, typing last. The highest-leverage change, and not theoretical — one product here is praised for it specifically. SAP Field Service Management, 5 stars, UK App Store, July 2020: "Ability to use speech to write notes and add multiple attachments enables quick data capture." Ninety seconds talking into a phone produces a richer record than six minutes typing on site. Same mechanics as voice-to-CRM in field sales; the payload is a job sheet instead of a call note.

Photo-first evidence, uploaded in the background. Photographs are how engineers document work, and upload is where it breaks. Multi-select must exist — a Dynamics 365 reviewer asked "who why would you have the only option to upload photos 1 at a time?" — and uploads must queue and retry in the background, never blocking.

Sensible defaults, and one sign-in per shift. Most dropdowns have an answer right 80% of the time; default to it. And login friction is 7.8% of complaints — bigger than photos, bigger than taps. Sessions should be measured in shifts, with face unlock offered, because fingerprints do not work through gloves.

Methodology and sources

Everything original here comes from one dataset built on 26 July 2026.

Gathering. The 14 products, named with per-product counts in Exhibit 2, are the engineer-facing apps of platforms a UK mid-market operation would realistically shortlist. Google Play reviews were retrieved from the GB storefront in English, newest first, paginated to exhaustion or 12 pages per app; Apple reviews came from embedded page data on each app's GB and US listings. Total: 2,814 — 2,669 from Google Play, 145 from the App Store.

Analysis set. Every review rated 1–3 stars with at least 25 characters of text: n = 1,557, spanning 12 August 2013 to 24 July 2026, with 49% dated 2024 or later (that recent subset, n = 759, is reported separately). Ratings without text say nothing about why; 4- and 5-star reviews were read separately for the fairness and "what fixes it" sections rather than coded.

Categorisation. A fixed set of pattern definitions, one per theme, run against every review, so a review naming three problems counts in three. Symptom themes (crashes, slowness, battery) match the complaint word itself. Object themes (offline, photos, search, forms) require the object within 70 characters of a failure cue — "photos" plus "won't upload", not any mention of a photograph — because an early version counted positive mentions and, embarrassingly, matched "hot" inside "photos". 65% carried at least one coded theme; 35% did not, being almost entirely undifferentiated abuse naming no mechanism, and they remain in every denominator.

Limitations. The sample is not UK-only — Google Play review text is not geo-restricted — but the UK-built subset (n = 190) produces the same theme ranking, which is the main reason I trust the ordering. It is unbalanced: Salesforce Field Service alone is 631 of 1,557, so percentages skew towards high-volume enterprise apps. Reviewers self-select heavily, biasing every absolute figure downwards, hence the reliance on within-sample comparisons. Role is inferred from wording on sub-samples of 70 and 88 — the most interesting number here and the least robust. Coding is automated then spot-checked, not hand-adjudicated.

What I could not access. I intended to include G2, Capterra and Software Advice, which carry more structured pros-and-cons text than app stores. All blocked automated retrieval: G2 returned HTTP 403, Capterra UK 410, Software Advice 404, Trustpilot 403. Apple's legacy reviews RSS feed returns zero entries and appears defunct. I used no secondary summary of those platforms, because a review count I have not read is one I cannot stand behind — so this dataset skews towards mobile failure and away from commercial complaints.

On naming vendors. Every quotation was retrieved from the review it is attributed to and reproduced verbatim, including spelling. Reviews are unverified accounts by anonymous users — evidence of what engineers report, not established fact. I excluded reviews alleging vendor conduct I cannot check.

What to do on Monday

Ride along. Pick your most experienced engineer and your least, and spend a day in each van. Time one thing: the gap between "work finished" and "job closed in the system". You will learn more in two days than from any demo, and you will have a baseline nobody can argue with.

Send the three offline questions in writing, to your incumbent and anyone you are evaluating: is offline on by default; which functions stop without signal; what happens to data already entered when the connection returns.

Read your own reviews. Search your platform on Google Play and read the one-star reviews. Half an hour, no cost. If your engineers are in there, your agenda is already written.

For the arithmetic on your specific team rather than my worked example, that is what the hidden waste audit produces. If you would rather talk it through, book a call.

Frequently Asked Questions

Why do field service engineers dislike their job management software?

Mostly because it does not work reliably. In 1,557 negative and mixed reviews I coded in July 2026, 40% complained about basic reliability — crashes (16.6%), functions not working (12.1%), slowness (11.8%) and sync failures (6.6%) — against 22% about workflow design such as too many taps or forms.

Is the software genuinely bad, or are engineers resistant to change?

The evidence points at the software. Reviews written in buyer or owner language average 3.33 stars while reviews by engineers who say they are made to use the same products average 1.53 — a 1.80-star gap within the same review pool. An engineer avoiding an app that crashes and loses their photos is behaving rationally.

Which field service apps genuinely work offline?

Fewer than the marketing suggests, and it is hard to check. Of eight platforms I tried in July 2026, only Microsoft publishes specific public documentation of offline limitations — stating offline must be configured by an administrator, that inventory validation, knowledge articles and Export to PDF do not work offline, and that its new mobile experience does not support offline at all.

How much of the UK has mobile coverage for field engineers?

83.58% of UK landmass had good 4G from all four operators as at 1 January 2026, 96.21% from at least one, and 3.79% had none (Ofcom, Connected Nations update: Spring 2026, published 13 May 2026). Scotland is markedly worse at 71.11% and 9.18%.

What proportion of an engineer's day is spent on tools rather than admin?

Nobody publishes a credible figure. The canonical surveys — the Service Council's Voice of the Field Service Engineer and Field Service News research — sit behind membership and subscription walls, and no peer-reviewed benchmark exists. Treat any specific wrench-time percentage with suspicion unless it names a report you can open.

What is the cheapest way to improve field service app adoption?

Deleting fields. Most job sheets carry fields nobody reads, and removing them costs nothing. After that: pre-filling what the system knows, voice capture, background photo upload and one sign-in per shift. But fix reliability first — a redesigned form inside an app that crashes on save makes things worse.

Related Reading

• Why Field Sales Teams Won't Use the CRM (And What Actually Fixes It)
• Ghost Workflows: The Manual Tasks Nobody Documented
• Voice-to-CRM: The Complete Guide for UK Field Sales Teams
• Job Sheet Parsing: Turning Paper and PDFs into ERP Data
• The Cost of Bad CRM Data in Field Sales
• UK Field Force Automation Market 2026
• Why the UK Field Sales Business Model Is Breaking