WFK AuditVerified · Evidence-based
Redgate Freight (illustrative)
Regional logistics & courier — ~40 staff
A trusted, people-strong operator with a quote-to-win process that quietly leaks time, money and jobs — none of it a technology problem, all of it fixable.
UX maturity
58/ 100
AI maturity
31/ 100
Band
Emerging
Enquiry-to-quote journey2/5
Response time & follow-up2/5
Status & tracking comms3/5
Internal workflow & tools2/5
Data readiness2/5
AI appetite & governance1/5
01
Scope & method
What we assessed
- A 90-minute discovery session with the owner and office manager, plus three live quote build-throughs observed end to end.
- The enquiry-to-quote journey across phone, web form and both shared inboxes; a two-week sample of inbound enquiries and their outcomes.
- The internal toolset (spreadsheets, email, the dispatch board) and the current AI/automation appetite.
Caveats & limits
- Response-time figures are from a two-week sample, not a full year — directional, not audited. The pattern is clear; the exact percentage will move.
- Win/loss data does not exist in structured form, so lost-enquiry impact is estimated from the team’s account, flagged as such throughout.
- No customer interviews in this scope — findings on customer experience are inferred from observed journeys and staff report, not direct research.
02
Executive summary
- Every quote is hand-built from scratch. The owner estimates ~11 hours a week across the team goes into re-keying details that already exist elsewhere.
- Roughly 1 in 3 enquiries never gets a quote back inside 24 hours — and those are the ones most likely to go to a competitor who replied first.
- The business runs on trust and personal service. That is the moat, and nothing here recommends automating it away — only removing the admin friction around it.
03
What's working
- Genuine personal service — customers name specific staff in testimonials. This is real and defensible.
- Deep niche credibility in time-critical regional freight; repeat clients dominate revenue.
- A tight, willing team that already works around the tooling gaps by hand — the appetite to improve is there.
04
Findings
UX-01
CriticalQuotes are rebuilt from zero every time
UX › Enquiry-to-quote journey
- Evidence
- Observed three quotes built live: each pulled customer, route and pricing detail by hand from email, a spreadsheet and memory. No reusable template or saved customer record.
- Impact
- ~11 staff-hours/week on re-keying; quote quality varies by who builds it; the owner is the bottleneck for anything non-standard.
- Recommendation
- Introduce a structured quote form backed by a saved customer/route record, so a repeat quote is a lookup, not a rebuild. Internal-first, human-approved before it goes out.
Effort to fixMedium
OPS-02
CriticalThe pricing spreadsheet is unbacked-up
AI › Data readiness & resilience
- Evidence
- The master pricing spreadsheet lives on one machine with no version history or backup. Loss would be near-unrecoverable.
- Impact
- One drive failure or bad edit could take out the ability to quote at all. Highest-severity, lowest-effort fix in the report.
- Recommendation
- Move it to versioned cloud storage today. A 20-minute fix that removes an existential risk.
Effort to fixSmall
UX-02
HighNo first-response inside the window that wins the job
UX › Response time & follow-up
- Evidence
- Enquiries arrive by phone, web form and two inboxes with no shared queue. ~30% waited >24h for a quote; several had no acknowledgement at all.
- Impact
- Lost work goes uncounted — the enquiries that vanish are invisible in the numbers. First-to-respond usually wins in time-critical freight.
- Recommendation
- One shared enquiry queue with an automatic acknowledgement ("we’ve got it, quote by X"). Buys time and stops silent drop-off — the single highest-ROI fix.
Effort to fixSmall
AI-01
HighNo data foundation for the automation they want
AI › Data readiness
- Evidence
- Pricing logic lives in the owner’s head and a personal spreadsheet. No structured history of quotes-won-vs-lost. Keen on "AI quoting" with nothing for it to learn from.
- Impact
- Any AI/automation bought today would have no reliable ground to stand on — a common, expensive false start.
- Recommendation
- Capture structured quote + outcome data first (a by-product of fixing UX-01). Six months of clean history is the prerequisite; revisit AI-assisted pricing after, not before.
Effort to fixLarge
UX-04
HighThe quote request is unusable on a phone
UX › Mobile & forms
- Evidence
- On a 375px viewport the web enquiry form overflows horizontally; the "Get a quote" button sits below the fold and two fields are impossible to tap accurately.
- Impact
- Over half of inbound web traffic is mobile. A form that fights the thumb is lost enquiries you never see.
- Recommendation
- Make the form single-column, full-width, with 44px tap targets and the CTA above the fold.
Effort to fixSmall
UX-08
HighAccessibility failures block some users outright
UX › Accessibility (WCAG 2.2 AA)
- Evidence
- Body text and the primary CTA fall below 4.5:1 contrast; the enquiry form has no visible focus state and unlabelled fields; images lack alt text (full table in §05).
- Impact
- Keyboard and screen-reader users cannot complete an enquiry — lost customers, and a legal/compliance exposure for public-sector-adjacent contracts.
- Recommendation
- Fix contrast, restore focus states, label every field, add alt text. Detail and criteria in the conformance table.
Effort to fixMedium
OPS-01
HighAll pricing knowledge lives in one head
AI › Data readiness & resilience
- Evidence
- Non-standard quotes route to the owner because the pricing logic is undocumented. When they’re away, complex quotes stall.
- Impact
- A single point of failure on the core revenue process — a business-continuity risk, not just an efficiency one.
- Recommendation
- Externalise pricing rules into the structured quote record (UX-01) — the same fix that also unlocks data.
Effort to fixMedium
UX-03
MediumStatus chasing eats the office day
UX › Status & tracking comms
- Evidence
- Customers phone to ask "where is it?" because there is no proactive update. Staff interrupt dispatch to answer, then lose their place.
- Impact
- Repeated interruptions fragment the team’s focus and push quoting later — feeding straight back into UX-02.
- Recommendation
- Proactive milestone updates (booked / collected / delivered) on the jobs that generate the most "where is it?" calls. Start with the top 20% of routes.
Effort to fixMedium
AI-02
MediumAppetite outruns oversight
AI › Governance & trust
- Evidence
- Interest in an "AI that just replies to customers." No view yet on what it would be allowed to send, who checks it, or how a mistake is caught.
- Impact
- Customer-facing automation without an approval gate is the fastest way to damage the personal-service trust the business is built on.
- Recommendation
- Any customer-facing AI stays internal-draft, human-approved before send. Draft the reply; never let it speak unattended. This is a design rule, not a tool.
Effort to fixSmall
UX-05
MediumThe form asks for data you already hold
UX › Enquiry-to-quote journey
- Evidence
- Returning customers must re-enter company, address and account details every time; there is no recognition of a known customer.
- Impact
- Friction on exactly the people most likely to book — repeat clients, who are most of the revenue.
- Recommendation
- Recognise returning customers and pre-fill from the saved record introduced in UX-01.
Effort to fixMedium
UX-06
MediumNo confirmation after submitting an enquiry
UX › Response time & follow-up
- Evidence
- Submitting the web form returns a bare page reload — no "we’ve got it" message, no reference number, no expectation set.
- Impact
- Customers assume it failed and phone or resubmit, adding load and eroding confidence before you’ve even quoted.
- Recommendation
- Show a clear confirmation with a reference and a "quote by X" promise — pairs with UX-02.
Effort to fixSmall
UX-07
MediumThe homepage buries the thing that wins the work
UX › Positioning & content
- Evidence
- The niche (time-critical regional freight) and the named-staff personal service — the two things customers cite in testimonials — are three scrolls down, under generic stock copy.
- Impact
- The differentiator that actually closes work is hidden; visitors bounce before they reach it.
- Recommendation
- Lead with the niche and the people. Put the proof (testimonials) above the fold.
Effort to fixSmall
AI-03
MediumNo baseline to prove any automation worked
AI › Value measurement
- Evidence
- No current measure of quote turnaround, win rate, or hours spent — so any future tool’s impact would be unprovable, and unsellable internally.
- Impact
- You could improve things and never be able to show it — which also makes the next investment harder to justify.
- Recommendation
- Capture three baseline numbers now (turnaround, win rate, admin hours) so the "before" exists.
Effort to fixSmall
AI-04
MediumThe team reads "AI" as "redundancy"
AI › Adoption & change
- Evidence
- Two staff independently voiced worry that automation means job cuts. No one has framed it as removing admin, not people.
- Impact
- Quiet resistance kills adoption — the best tool fails if the people route around it.
- Recommendation
- Frame and communicate every change as "less re-keying, same team, more customer time." Involve them in the design.
Effort to fixSmall
05
Accessibility conformance
WCAG 2.2 AAThe public enquiry journey was checked against the WCAG 2.2 AA success criteria most relevant to completing a quote request. The blocking failures (contrast, focus, labels) are all low-to-medium effort.
| Criterion | Level | Result | Note |
|---|---|---|---|
| 1.1.1 Non-text Content (alt text) | A | Fail | Fleet and hero images have no alt text. |
| 1.4.3 Contrast (Minimum) | AA | Fail | Body text ~3.1:1; primary CTA ~2.9:1. |
| 1.4.11 Non-text Contrast | AA | Fail | Form field borders too faint to perceive. |
| 2.1.1 Keyboard | A | Pass | — |
| 2.4.7 Focus Visible | AA | Fail | No visible focus ring on links or inputs. |
| 3.3.2 Labels or Instructions | A | Fail | Two enquiry fields rely on placeholder text only. |
| 4.1.2 Name, Role, Value | A | Pass | — |
| 2.5.8 Target Size (Minimum) | AA | Fail | Mobile CTA and radio targets under 24px. |
06
AI risk & governance register
R-01
Customer-facing AI sends an inaccurate or off-brand reply unattended.
Likelihood HighImpact High
MitigationInternal-draft, human-approved before send. No unattended customer messages (see AI-02).
R-02
AI-assisted pricing trained on thin or biased history quotes badly.
Likelihood MediumImpact High
MitigationDefer assisted pricing until six months of clean, structured quote+outcome data exists (AI-01, AI-03).
R-03
Over-automation erodes the personal-service differentiator.
Likelihood MediumImpact High
MitigationAutomate admin, never the relationship. Explicit "do not automate" list agreed with the owner.
R-04
Vendor lock-in on an off-the-shelf "AI quoting" tool.
Likelihood MediumImpact Medium
MitigationOwn the structured data first; keep it portable so tools can be changed without losing the asset.
07
Agent opportunities surfaced
Workflows where an assistant could do the drafting and your team keeps the decision. Each is a standalone pilot — a review screen we design that sits on top of a tool you already have. Nothing here runs unattended; a person approves every action.
AO-01
Acknowledging every inbound enquiry inside the window that wins the job
- What we saw
- Enquiries land in two shared inboxes and by phone with no shared queue; ~1 in 3 waited over 24 hours and several got no reply at all (UX-02, UX-06).
- The interface we’d design
- A single review queue where each new enquiry arrives with a drafted acknowledgement ("we’ve got it, quote by X") and a suggested priority. Your office manager scans, tweaks a line if needed, and sends — seconds per enquiry, nothing silent.
- You stay in control
- No message leaves until a person clicks send. The draft is a starting point, not an autopilot.
- Runs on top of
- Your existing shared inboxes — no new email system, no migration. The queue reads what already arrives.
Pilot4-week pilot to design and stand up the acknowledgement queue, measured on first-response time against your current two-week baseline. Fixed fee, then a small monthly to run and tune.UX-02UX-06
AO-02
Rebuilding a repeat customer’s quote from scratch
- What we saw
- Every quote is hand-built by pulling customer, route and pricing detail from email, a spreadsheet and memory — ~11 staff-hours a week, quality varying by who builds it (UX-01, UX-05, OPS-01).
- The interface we’d design
- A quote-draft screen: pick the customer, and it pre-fills a proposed quote from the saved record and the inbound detail, with the pricing logic shown so it’s checkable. Your team confirms or adjusts before it goes out — a lookup-and-approve, not a rebuild.
- You stay in control
- Every quote is reviewed and signed off by a person before it reaches the customer. The judgement stays yours; only the re-keying goes.
- Runs on top of
- The structured quote record introduced in the Quote-to-Win build (UX-01) — this interface sits on that, it does not replace your pricing.
PilotScoped as the next step after the structured quote record exists. Fixed-fee build of the draft-and-approve screen, then monthly to tune as edge cases surface.UX-01UX-05OPS-01
AO-03
Answering "where is it?" calls that interrupt the office day
- What we saw
- Customers phone for status because there’s no proactive update; staff break off dispatch to answer and lose their place, pushing quoting later (UX-03).
- The interface we’d design
- A daily review screen that drafts proactive milestone updates (collected / in transit / delivered) for the routes that generate the most chasing calls. Your team approves the batch in one pass; the updates go out before the phone rings.
- You stay in control
- You approve the batch each day — the assistant proposes the messages, a person releases them. Start with the top 20% of routes and expand only if it’s working.
- Runs on top of
- The status information already on your dispatch board — the interface surfaces and drafts from it, it doesn’t track the jobs itself.
PilotSmall pilot on your busiest routes, measured on volume of inbound "where is it?" calls. Deliberately narrow — prove it on 20% before touching the rest.UX-03
08
The biggest opportunity
The quote-to-win bottleneck is one problem wearing three hats: slow response, hand-built quotes, and no data to improve either. Fix the structured quote once and it pays back three times — faster replies, consistent pricing, and the clean history that makes real AI possible later.
09
What not to automate yet
- The personal phone relationships. This is the differentiator — automate the admin around it, never the human on the end of it.
- Customer-facing AI, for now. No data foundation and no oversight model means it is a liability, not a feature. It comes after, and only behind an approval gate.
10
Prioritised roadmap
Now
Stop the silent leaks and remove the existential risk — days, not months.
OPS-02UX-02UX-06AI-02
Next
Fix the quote engine, accessibility and comms; start capturing clean data.
UX-01UX-05UX-08OPS-01AI-03
Later
Once the data foundation is real, revisit assisted pricing and adoption.
AI-01AI-04
11
Do this first
- 1Put every enquiry into one shared queue with an automatic acknowledgement (UX-02).
- 2Write the "internal-draft, human-approved" rule for any AI that touches a customer (AI-02).
- 3Design the structured quote record — the fix that pays back three times (UX-01).
12
Proposed engagement
A four-week Quote-to-Win build — the shared enquiry queue, the structured quote record, and the data capture that unlocks everything after. One scoped engagement, not a five-year transformation.
What you get
- A single shared enquiry queue with automatic acknowledgement, live within week one (addresses UX-02).
- A structured quote record with saved customers and routes, so a repeat quote is a lookup, not a rebuild (UX-01).
- Clean quote + outcome capture switched on from day one — the six months of history that make assisted pricing possible later (AI-01).
- A one-page "internal-draft, human-approved" rule for any customer-facing AI, so the personal-service trust is protected by design (AI-02).
- A 30-minute handover walkthrough with the team, plus a written runbook.
Timeline
Four weeks, fixed. Weekly check-in; nothing goes live without your sign-off.
Investment
£6,500 fixed — no day-rate creep, no agency overhead.
Cost of inaction
Left as-is, the ~11 hours/week of re-keying and the ~1-in-3 enquiries that go cold keep compounding — quietly, and invisibly, because the lost work never shows up in the numbers.
A
Appendix — full issue log
Every finding in one place, ordered by severity — the working behind the report.
| ID | Finding | Severity | Effort |
|---|---|---|---|
| UX-01 | Quotes are rebuilt from zero every time | Critical | Medium |
| OPS-02 | The pricing spreadsheet is unbacked-up | Critical | Small |
| UX-02 | No first-response inside the window that wins the job | High | Small |
| AI-01 | No data foundation for the automation they want | High | Large |
| UX-04 | The quote request is unusable on a phone | High | Small |
| UX-08 | Accessibility failures block some users outright | High | Medium |
| OPS-01 | All pricing knowledge lives in one head | High | Medium |
| UX-03 | Status chasing eats the office day | Medium | Medium |
| AI-02 | Appetite outruns oversight | Medium | Small |
| UX-05 | The form asks for data you already hold | Medium | Medium |
| UX-06 | No confirmation after submitting an enquiry | Medium | Small |
| UX-07 | The homepage buries the thing that wins the work | Medium | Small |
| AI-03 | No baseline to prove any automation worked | Medium | Small |
| AI-04 | The team reads "AI" as "redundancy" | Medium | Small |