In 2023 to 2024 HM Revenue & Customs received 37 million telephone calls, and the department stated that 66% of those calls could have been handled online, according to the House of Commons Public Accounts Committee report on HMRC customer service. That's the central lesson for smaller organisations too. Good solutions for customer service don't start with buying a chatbot. They start with deciding which contacts should never need a phone call, which ones always need a person, and how a customer moves between the two without repeating themselves.
Six Customer Service Solutions and What Each One Fixes
In 2025, 78% of UK consumers said they were very or somewhat likely to abandon a brand after one poor customer-service experience, according to TCN's survey of 1,000 UK adults. That figure changes the buying question. You're not shopping for features. You're fixing the experience that makes a customer leave.
Think of the six categories as parts of one working kitchen. The CRM is the recipe book that holds every customer's history. The contact centre is the order window that sends each request to the right cook. AI assistants run the quick counter for routine orders, workflow automation moves tickets between stations, and analytics shows the manager which problems keep coming back. Managed IT support keeps the kitchen itself running, secure and open for service.
Start From the Pain, Not the Product
You can name your biggest pain quickly. It's usually lost context between emails and calls, missed calls at peak times, or slow handoffs between colleagues. Name yours first, then check which category addresses it. A firm losing context needs a CRM before it needs AI. A firm missing calls needs routing before it needs dashboards.
The comparison below sets out what each category does, who it suits and what it'll cost you in effort.
| Category | Core job | Best suited to | Main trade-off |
|---|---|---|---|
| CRM (Dynamics 365 Customer Service) | Holds customer records, cases and history in one place | Teams losing context between calls, emails and visits | Needs clean data and ongoing admin time |
| Contact centre (telephony and omnichannel) | Routes calls, chats and emails to the right agent | Growing inbound volumes and queues | Licence and hardware costs, plus agent training |
| AI and Copilot assistants | Answers routine questions and drafts replies for agents | High-volume, low-complexity requests | Poor handoff makes customers repeat their story |
| Workflow automation (Power Automate) | Moves cases, approvals and alerts without manual steps | Repetitive hand-offs between teams | Broken flows can fail silently without monitoring |
| Analytics (Power BI) | Shows volumes, resolution times and repeat contacts | Leaders who need evidence for decisions | Only as reliable as the data captured at source |
| Managed IT support | Keeps platforms running, secure and integrated | Firms without in-house IT capacity | Less direct control over day-to-day priorities |
Practical rule: Buy for the recurring problem you can already describe in one sentence. If you can't describe it, audit your contacts before you spend anything.
Why Self-Service and Human Support Must Work Together
In 2024, 85% of UK consumers considered having a choice of customer-service engagement methods important, according to ServiceNow's Consumer Voice Report. Choice doesn't mean every channel for every task. It means each channel does its own job well, and the customer can reach a person when the task needs one.
The common failure runs in a loop. A customer can't find an answer online, so they phone. The queue grows, staff rush, and answers get shorter. More customers then phone back because the first answer didn't settle the issue. Removing the phone line doesn't break that loop. Fixing the online journey does.
The Building Blocks That Break the Loop
Five pieces work together, and missing any one of them weakens the rest.
- A knowledge base built around top questions: Write short pages that answer one task each, in the words customers use when they search.
- Guided online forms: Ask for order number, account reference and a clear description upfront, so staff don't spend the first reply asking for basics.
- Case tracking: Give customers a reference and a way to see progress, and you'll cut chase contacts that add nothing.
- Clear escalation routes: Name who picks up urgent, complex or accessibility-related cases, and how quickly.
- Reporting on what's left: Track which questions still reach staff after self-service, because those pages need rewriting.
A Test You Can Run This Week
List the ten most frequent reasons people contact you. Mark each one as answerable online, answerable by form, or needing a person. Then check two things. Does every item needing a person have a defined route to one? And if a customer starts on web chat and moves to the phone, does their history move with them?
If the answer to either question is no, that's your first project. It's cheaper than new licences, and it protects staff time for the cases where a person makes the difference.
Choosing Between a CRM and a Contact Centre Foundation
Readers often confuse these two, so let's separate them plainly. A CRM keeps the record. It holds who the customer is, what they've ordered, which cases they've raised and what you promised them last time. A contact centre controls arrival. It decides how calls, chats and emails queue and which agent receives each one. One is memory, the other is traffic control.
Most small teams feel missing memory first. An agent answers the phone, has no history in front of them, and asks the customer to explain from the start. That's why a CRM is usually the foundation. If you'd like the concept unpacked further, our guide on what customer relationship management means in practice covers the basics. Add contact centre routing when inbound volume creates queues, missed calls or unfair workloads, not before.
Where Dynamics 365 Fits
F1Group offers Dynamics 365 CRM and Customer Service with full customisation, so your products, case types and approval steps can be reflected in the system rather than forcing your team into a generic template. It also connects to Microsoft 365 email and Teams, so agents work from one screen instead of copying notes between Outlook, a spreadsheet and a phone log.
The Costs Buyers Overlook
Licences are only part of the bill, and you should ask for every quote in pounds sterling (GBP). If a supplier shows US dollar pricing, ask for the GBP equivalent and check whether it's per user, per month, and what happens when part-time staff or seasonal cover need access.
Two hidden costs matter more. First, migration. Moving records out of spreadsheets takes cleaning time, and it must follow UK GDPR data minimisation and retention rules, so you carry across only what you need and can justify keeping. Second, administration. Someone has to keep fields, queues and security roles accurate as the team changes. Budget a named owner, even if it's a few hours a month.
A Worked Example
Say a team of about twelve agents shares one inbox. Two people reply to the same customer, nobody owns a tricky case, and holiday cover means reading weeks of threads. Replace that inbox with CRM cases and each enquiry gets an owner, a status and a history the next agent can read in seconds. The migration effort, cleaning contacts and agreeing case categories, is usually worth it within the first busy month.
Before you choose, answer two questions this week. How many channels do customers use? And where does each customer's history live today? If history lives in personal inboxes, start with the CRM.
Where AI, Copilot and Automation Fit Safely
In 2025, 45% of UK consumers said they were comfortable interacting with chatbots or virtual assistants, while 36% were uncomfortable, according to MaxContact's survey of more than 1,000 UK consumers. An AI-first design therefore puts off a large minority before you've solved anything. The principle that works is AI-assisted, human-available.
Used that way, the tools are useful. Copilot can draft a reply for an agent to check and summarise a long case before a call. Power Automate can route a case straight to a person when it's flagged as urgent, complex or complaint-related. Our overview of customer service automation options shows where these pieces sit. None of that requires a customer to argue with a bot.
Draw the Boundary by Request Type
AI handles narrow, predictable requests well. It handles emotion, risk and exceptions badly. The table below draws that line for the requests most SMEs receive.
| Request type | AI handles it? | Hand off to a person when |
|---|---|---|
| Order status | Yes | Delivery exception or lost parcel |
| Appointment or opening hours | Yes | Customer needs to change a booking that clashes with another |
| Account balance or password reset | Partly | Identity check fails or the account is locked |
| Product information | Yes | Customer asks about compatibility, safety or warranty |
| Refund or payment dispute | Partly | Amount exceeds the agreed limit or the customer has contacted before |
| Complaint | No, acknowledge only | Always, and within the complaints procedure timescale |
| Financial hardship or vulnerability | No | Immediately, with a named contact |
Three Design Features You Should Require
- History carried across: The full conversation lands in the human queue with the case, so the customer never restarts their story.
- Clear disclosure: Customers are told they're speaking to AI, in line with your UK GDPR duties on transparency and data handling.
- One-step human route: Every AI answer offers a visible way to reach a person, not a loop back to the start.
Vulnerable customers are never routed to AI-only handling. Build that in as a rule, not a judgement call for the software.
Adding AI without integrated records and a human fallback can increase contact volume rather than reduce it, because failed automated chats become phone calls.
Start with a pilot. Pick two or three high-volume, low-risk requests, run them for a month, and check fallback rate and repeat contacts before you widen the scope. AI-assisted replies still need a named person accountable for what goes out.
Measuring Service Quality with Complaint Data and KPIs
The Consumer Council for Water (CCW) recorded 20,744 business-customer complaints to retailers across all contact channels in 2023 to 2024, equal to 75 complaints per 10,000 supply points, in its business customer complaints report for 2023 to 2024. That is a regulated utilities benchmark, not an SME average. Its value lies in the method: a rate per customer lets you compare fairly, while a raw total grows as you win more customers.
A raw count is like a shop counting unhappy customers without counting total shoppers. The rate gives you a like-for-like signal. The chart below shows that normalisation point alongside three service measures worth tracking from the start.
Capture the Right Fields on Every Case
Complaint data only helps if it is structured the same way every time. Capture customer and account identifiers, channel, product or service, failure category, severity, vulnerability indicators, timestamps, owner, resolution code, compensation and root cause. Most fields are dropdowns an agent completes in seconds, so the discipline sits in the design, not the typing.
Feed those fields from your CRM, telephony, email and knowledge tools into Power BI. You can then separate a billing defect from an individual performance issue, instead of blaming whoever handled the angriest call. Set automatic alerts when repeat contacts rise or cases pass an agreed age. Problems then surface before the monthly meeting, when there is still time to change a queue, a script or a handoff rule.
The short video below walks through how structured case data supports this kind of reporting in practice.
Close the Loop, Then Measure
Classify each case, route it by severity and skill, monitor resolution time against your service level, and send recurring causes to the team that can fix them. A dashboard nobody acts on is decoration.
| Measure | What it shows | How to read it |
|---|---|---|
| First-contact resolution | Share of issues solved in one contact | Shows whether channels resolve issues, not just answer them |
| Median response time | Typical wait before a useful reply | Use the median, because averages hide long waits |
| Fallback rate | Share of AI conversations passed to a person | A high rate signals gaps in intent coverage |
| Repeat contact rate | Customers returning about the same issue | Flags weak fixes and poor handoffs |
| Post-escalation satisfaction | Rating given after a person takes over | Confirms the handoff itself works |
Baseline repeat contact rate first this month. It is the measure that exposes weak fixes across every channel at once. Pair it with fallback rate to see whether the problem sits in self-service design, in the handoff rule, or in missing case data. A high fallback rate is not proof that AI failed. It often shows that intent coverage was too narrow or that the customer's history did not travel with the case.
Selection Criteria and a Roadmap for Implementation
We apply three tests before recommending any solution. It must suit the customer's needs, meet the customer's risk acceptance, and fall within the customer's budget. A platform that fails one test will cause trouble later, however strong its feature list looks in a demonstration.
The diagram below summarises those tests and the technical review that follows them.
Score Each Option Against One Checklist
- Needs fit: Does it fix the recurring pain you identified, in your team's words?
- Risk fit: Can you control data access, retention and AI disclosure to your standard?
- Budget fit: Is the full cost in GBP clear, including licences, migration and training?
- Integration: Does it connect to Microsoft 365, your telephony and third-party applications?
- Single point of contact: Can one support route cover every system, including third-party apps?
- Reporting effort: Can a manager produce the weekly figures without exporting to a spreadsheet?
- Training need: How long before a new starter handles a case confidently?
Our support services can be customised to encompass all systems, including third-party apps, with a singular contact point, which matters when a fault sits between two suppliers. For project detail, see our notes on Dynamics 365 Customer Service implementation.
The Six-Month Roadmap
- Audit four weeks of contact data by reason and channel.
- Set baselines for your chosen measures.
- Fix the knowledge base for your top ten questions.
- Configure case records and routing.
- Pilot AI or automation on two low-risk requests.
- Define handoff rules and test them with real staff.
- Review monthly and widen scope only when fallback and repeat rates improve.
| Phase | Roadmap steps | Timing |
|---|---|---|
| Audit and baseline | Steps 1 and 2, audit contact data and set baselines | Weeks 1 to 4 |
| Knowledge and case set-up | Steps 3 and 4, fix top ten answers, configure cases and routing | Weeks 5 to 10 |
| Pilot and handoff | Steps 5 and 6, pilot two low-risk requests and test handoff rules | Weeks 11 to 16 |
| Review and widen | Step 7, monthly review, widening scope only when rates improve | Months 5 to 6 |
Pace depends on data quality and team size, so treat the timings as a shape, not a promise. Choosing a platform before auditing demand is a common reason projects overrun. The audit is what tells you which licences you'll use.
East Midlands Support and a Worked Case for Local SMBs
Many East Midlands SMBs across Nottingham, Leicester, Derby and Lincoln still run customer contact through shared Microsoft 365 mailboxes and spreadsheets. That works until a network or device problem blocks the whole team at once, and then you need a partner who can visit the site, not just reset a password remotely. Our support plans include unlimited phone and remote support, onsite visits when needed, director-level contact and 24×7 monitoring, so you get the same help whether the issue is a Teams outage or a Dynamics 365 configuration question.
An Illustrative Case
This example is hypothetical, and it carries no promised figures. Say a fifteen-person equipment repair firm near Derby takes warranty queries through a shared inbox and fields constant repeat questions about order status. Nobody can tell which queries are waiting, and customers chase by phone.
The firm audits four weeks of contacts and finds order status and warranty claims make up most of its volume. It moves customer records into Dynamics 365 Customer Service, so every warranty query sits against the equipment history. A Power Automate flow then opens a case as soon as a warranty claim is logged, and assigns it to the right technician.
Finally, the operations manager reviews a Power BI view of repeat contacts each week. The firm judges the change against the baselines it set at the start, not against a supplier's promise. Local support through the first month matters most, because that's when a team under pressure drifts back to the shared inbox it knows.
Book a Free Review of Your Customer Service Setup
A free IT review starts from how work moves between people and systems. F1Group's review identifies the IT solutions worth implementing and what may be done to keep IT services running smoothly. Support plans are designed to suit most budgets.
Bring a short pack before you write in. Your ten most common contact reasons, even as a rough tally from four weeks, show where the volume sits. Systems in use (mailboxes, telephony, spreadsheets and any CRM) show where a case can stall. Users and licences, including part-time cover, show who is available when the usual person is out. Channels customers already use (phone, email, web forms, chat or walk-ins) show whether the same facts are captured once or typed again on every contact.
Those four items decide more than the software name on the invoice. A handoff with no written owner works like a parcel left in a corridor: the next person may never see it. If the phone queue never updates the record the email team uses, context drops between channels. Data capture needs one place the history lives, and a rule for what must be written down before a case moves.
Test that before you buy anything else. Over about six months, count contact reasons and note the channel each one arrived on. Agree who takes a case when it leaves a shared mailbox, and what the record must contain. Only then match a Microsoft foundation to that pattern, and bring it in at a pace staff can absorb.
If context is slipping between calls, emails and spreadsheets, email info@f1group.com and ask for the review around your working day. Visit F1Group to see how the East Midlands team supports Dynamics 365, Microsoft 365 and Power Platform for local organisations, and start with a free review of your current setup.