Explore fresh, expert perspectives on customer engagement and innovation

Scorecards do more than quality assurance. The same mechanic, built as a rating, pass/fail, or observation question, can predict customer outcomes, meet UK regulatory requirements like FCA Consumer Duty, surface commercial opportunities, and build a data-backed case for where to introduce AI agents. This guide covers 13 ways to use one, plus how to build each use case in MaxContact.

In This Article

• How Scorecards Work
• Predicting Customer Outcomes
• Compliance and Regulatory Risk
• Commercial and Sales Performance
• Operational and Product Insights
• Getting Started
• FAQs

For years, the case for automated QA rested on one number: most contact centres only review 2-5% of their calls. That's no longer the gap. According to MaxContact's 2026 UK Conduct Risk Report, 46.7% of regulated contact centres now use AI to review every call, nearly matching the 49.5% still sampling manually. Coverage, in other words, is largely solved.

What isn't solved is what happens next. The same research found that 54% of contact centres had a compliance breach or harm issue surface outside their routine QA process, and 16.5% found more than one. Reviewing every call tells you what happened. It doesn't tell you what's about to happen. It won't flag who's about to churn, which calls should never have happened at all, or which agent needs help before their next review cycle.

That's the real opportunity in a scorecard. Not just scoring compliance after the fact, but turning every call into a signal you can act on, for coaching, for commercial performance, and for risk. Here are 13 ways to do it.

How Scorecards Work

Before diving into the use cases, a quick note on mechanics. It's what makes all of this possible in MaxContact Conversation Analytics, rather than needing a different tool for every job.

A MaxContact scorecard question can be built as an observation (a few lines of AI commentary on what it found), a rating (a score on any scale you choose, such as 1 to 5, or a simple true/false), or a pass/fail. The same underlying mechanic, "based on this transcript, did X happen?", powers everything from a compliance check to a churn prediction.

Scorecards can also run two ways. An always-on schedule scores every new call as it lands, continuously. A one-time schedule runs against a historical sample instead, useful for testing a new use case against last quarter's calls before switching it on for good.

Results then roll up into question-level reporting, so you're not digging through individual calls to see which questions are failing, where, and for whom.

Predicting Customer Outcomes

Predicted CSAT and NPS

A scorecard question can ask, in plain terms: based on this transcript, what would this customer's CSAT score likely be? Because it runs on every call, you get a sentiment read on 100% of conversations, not the 5% or so that respond to a post-call survey.

The catch: this is only credible once you calibrate it against your real survey results. Run the question alongside your existing survey data, check how closely the predictions track the real scores, and adjust until they're reliably close. Skip that step and all you've got is a sentiment score, not predicted CSAT. You have a sentiment score wearing a more exciting name. Once calibrated, the same logic extends to NPS.

Repeat Contact and FCR Likelihood

Was the issue actually resolved on this call, or is the customer about to ring back? A scorecard can score first-contact-resolution (FCR) likelihood directly from the transcript. Did the agent confirm the issue was fixed, or leave it open, vague, or dependent on another team?

This is arguably more actionable than CSAT, because failure demand has a direct, calculable cost. At roughly £5 per service call, every repeat contact doubles the cost of that interaction. That's before you've factored in agent time or the churn risk a frustrated second call brings with it.

Promise-to-Pay Likelihood

For collections teams, a scorecard can rate how firm a payment commitment was. Not just whether a promise to pay was logged, but whether the language was definite ("I'll pay Friday") or hedged ("I'll try to sort something"). The question can be scoped so it only applies when an arrangement-to-pay or promise-to-pay outcome has been set, rather than running on every call.

Because the strength of that commitment is directly tied to revenue collected, this gives collections leaders a leading indicator: a queue of weak promises worth a follow-up call, before the due date passes and it becomes a broken promise instead.

Customer Churn Intent

A scorecard question can rate whether a call shows real churn risk: cancellation language, competitor mentions, or complaints that keep repeating across contacts. Score it as a simple true/false, or as a numeric risk rating to see customers nudging from low to high risk before they act on it.

Either way, the output is the same: a retention queue built from what customers are saying, not from account age or a generic risk model.

Compliance and Regulatory Risk

Vulnerability Detection

Under FCA (Financial Conduct Authority) Consumer Duty, firms have to identify and respond appropriately to vulnerable customers, and increasingly, prove they did. The FCA's own framework names four drivers of vulnerability: health, life events, resilience, and capability. A scorecard question can be built to listen for the language that signals each of these. It scores whether the agent picked up on it, and flags whether they fully acknowledged it, only partially addressed it, or missed it entirely.

This is the strongest UK regulatory fit on this list, for a simple reason: firms already know they struggle with it. Last year, the FCA reviewed 180 board reports and found the most common failure wasn't a lack of effort. It was an inability to evidence outcomes for vulnerable customers specifically. A scorecard that runs on every call, not a sample, is a direct answer to that evidence gap.

Complaint Detection

Under FCA rules, an expression of dissatisfaction counts as a complaint whether or not the customer ever uses the word "complaint." A scorecard can be built to catch that distinction, flagging calls where a customer expressed real dissatisfaction, even phrased as a passing comment, so they get logged and handled as complaints rather than closed out as routine queries.

This is a clear compliance win with a direct cost attached. MaxContact's 2026 UK Conduct Risk Report found the average regulatory fine paid by UK contact centres in the past 12 months was £81,000, across everyone surveyed, including businesses that paid nothing at all. Nearly a third paid £50,000 or more. Since February 2026, the stakes have risen further still. The Data Use and Access Act raised the maximum PECR (Privacy and Electronic Communications Regulations) fine from £500,000 to £17.5 million or 4% of global turnover, whichever is higher. For the first time, company directors can also be held personally liable, for up to £500,000 individually.

Fraud and Social Engineering Signals

A scorecard can be trained to listen for the patterns that indicate fraud risk on a call: pressure tactics, inconsistent answers to identity questions, or signs that a third party is coaching the caller in the background. Scored consistently across every call, a pattern that's easy to miss in a single conversation becomes visible at scale, and auditable after the fact.

Fines aren't the only cost at stake here either. The same research found non-compliant sales cost UK contact centres an average of £22,000 in lost revenue over the past year through refunds and cancellations. Just over half of respondents (50.2%) said they'd lost a sale, a contract, or a customer entirely over a compliance issue.

Commercial and Sales Performance

Missed Sales Opportunities

A scorecard question can flag buying signals an agent didn't act on: a customer who asked about an upgrade, mentioned a competitor's offer, or hinted at a bigger need the agent never followed up on. Run across every call, this builds a clear, evidenced coaching list, rather than a manager's hunch about who's leaving revenue on the table.

Sales Methodology Scorecards

A scorecard can also check whether an agent followed a specific sales methodology, not just whether they spotted an opportunity. A BANT scorecard, for example, scores a call against the four areas the framework is built on: Budget, Authority, Need, and Timeline. Each becomes its own rating question, usually scored on a 0-5 scale in this example (ratings can run on any range you choose), with a minimum score needed to pass (3 out of 5 was the bar in the original demo), plus an observation question at the end to pull the whole picture together.

The same approach works for any other sales methodology your teams are trained on, including FAB, AIDA, or SPIN. Because every call gets measured against the same criteria, you get a consistent read on how closely the team is actually following the methodology, not just whether revenue came in.

Script and Campaign Effectiveness

Because scorecards score consistently and at scale, they're a ready-made engine for A/B testing. Run the same scorecard against calls using different script versions or campaign approaches, and compare outcomes directly: which version holds objections better, which closes faster, which gets flagged for compliance issues less often. The scorecard becomes the measurement layer for commercial experiments you're already running, instead of a side project in a spreadsheet.

Operational and Product Insights

Automation Candidacy

This is where a scorecard stops being purely a QA tool and starts building your automation roadmap. Start with one scorecard and one pass/fail question: is this call type a good candidate for automation? Pair it with an observation question asking what, specifically, makes it one. Run it across your calls and your candidates start to emerge: complaints, vulnerability checks, simple status updates, whatever they are for your business.

From there, write a dedicated scorecard with one question per candidate type, each scored pass/fail. Run that as a one-time scorecard against a historical sample, and you get a hit rate per candidate: this type of call could have been automated 87% of the time, that one only 20%. That gives you a data-backed, evidenced pipeline for deciding what to automate first, not a guess about where AI Agents could help, but a ranked list built from your own call data.

Contact Reason and Failure Demand

A scorecard can categorise why a customer called, and more usefully, flag which of those calls shouldn't have needed to happen at all. A customer chasing an order update because a text alert never arrived is failure demand: a problem created somewhere else in the business that lands on your contact centre's call volume. This is valuable well beyond QA. It's a direct line into your ops and product teams about where the underlying process is broken.

Knowledge Gaps

Agent uncertainty, excessive holds, and unnecessary transfers are all symptoms a scorecard can detect and score consistently. Rather than treating this as an individual performance issue, use it to point at the knowledge-base articles or training modules that need fixing. The same hesitation showing up across multiple agents on the same topic is a knowledge gap, not a coaching gap.

Getting Started

Here's how to turn any of these into a live scorecard in MaxContact.

1. Start with one use case, not all 13. Pick the one with the clearest payoff for your team right now. Vulnerability detection if Consumer Duty evidence is your biggest exposure, churn intent if retention is the priority.

‍
2. Write or import the scorecard. Describe the question in plain English and let the scorecard builder turn it into evaluation criteria, or import an existing scorecard if a similar methodology already exists elsewhere in the business.

‍
3. Choose your question type. Decide whether the use case needs an observation (a few lines of commentary), a rating (true/false, or any numeric scale, such as 1 - 5), or a straight pass/fail.

‍
4. Scope it, if it needs scoping. Some questions only make sense in context. A promise-to-pay question, for example, should only fire when an arrangement-to-pay outcome has actually been logged on that call. Scorecards can be scoped to specific products, outcomes, or campaign types so you're not scoring irrelevant calls.

‍
5. Test it against a historical sample first. Run it as a one-time scorecard against last month's or last quarter's calls before switching it on live. This matters most for anything predicting an outcome: CSAT, churn, automation candidacy. Check the results against what you already know before trusting it at scale.

‍
6. Switch to an always-on schedule. Once you're confident in the results, move the scorecard to an always-on schedule so it scores every new call automatically, with no manual sampling involved.

‍
7. Review through question-level reporting. Don't just watch an aggregate score. Use question-level reporting to see which specific questions are failing, for which agents, teams, or call types, so the next action is obvious rather than buried in an overall pass rate.

FAQs

Does this replace manual QA entirely? Not necessarily. Many regulated contact centres run scorecards alongside manual spot-checks, especially while a use case is new or being calibrated against real outcomes. What it replaces is reliance on a 2-5% sample as your only evidence base.

How accurate is predicted CSAT? It depends entirely on calibration. Run it alongside real survey data first, check how closely the predictions track actual scores, and adjust the scorecard question until they align. An uncalibrated predicted CSAT score is just a sentiment score with a different name.

Can I run a scorecard on calls that already happened? Yes. One-time scorecards can run against a historical sample, which is useful for testing a new use case or auditing a past period, rather than only scoring calls as they come in.

Do I need a different tool for each use case? No. Every use case on this list runs on the same scorecard mechanic: observation, rating, or pass/fail questions, inside Conversation Analytics. What changes is the question you ask, not the platform.

Does automation candidacy mean an AI agent takes over immediately? No. It identifies and ranks which call types are good candidates for automation based on real call data, giving you an evidenced starting point for deciding where to introduce AI Agents, rather than a guess.

Also known as an auto dialler, an outbound dialler automates the process of making calls to customers and contacts and routes the answered calls to your available agents.

Manual vs auto dialling

Dialling can be done manually on a mobile or business phone, with an agent placing calls one after another from a list. Many contact centres use automated dialling for high-volume outbound campaigns, as agents spend less time dialling and waiting for calls to be answered and more time having conversations.  

If you’re considering incorporating dialler technology into your contact strategy here are the dialler features you should look out for when choosing a provider.

7 outbound dialler features to look for

1. Multiple dialling modes available

Most auto-diallers have three dialling modes that teams switch between depending on the type of outbound campaign you are doing. These modes are predictive, progressive and preview.

Predictive dialling

Predictive dialling places calls ahead of agents becoming free based on the software’s predictions of agent availability. It dials multiple numbers simultaneously, so that when agents finish one call, they can be instantly connected to the next to maximise talk-time. It suits high-volume campaigns with experienced teams and mature contact lists.

Progressive dialling

Progressive diallers have a slower pace, only dialling a number when an agent is ready and available to take the call. Dialling is instant and automatic, so call volumes stay relatively high, but it is best suited to regulated campaigns and collections, where consistency matters more than volume.

Preview dialling

A preview dialler shows the agent the contact record first, and after a short preview period dials automatically. This delay lets the agent prepare for the call, using information typically taken from the company CRM system, which is often integrated into the dialler. Preview dialling mode suits high-value or complex calls where context changes the outcome.

For more detail on different dialling modes, read our predictive, progressive and preview dialler modes guide.

2. Answer Machine Detection (AMD)

Answer machine detection (AMD) identifies voicemails and answering machines before they reach an agent. This means agents only spend time on live conversations, saving them valuable time and boosting productivity.

If the system can't confidently tell within a couple of seconds whether a call is live, it passes the call to the agent, so real conversations aren't lost to false positives. MaxContact's dialler gives up to 40% less time lost to voicemails and no-answers.

3. Built in compliance controls

Automated outbound dialling is an area that’s closely regulated in the UK, with silent and abandoned calls often a talking point due to the fines and reputational damage if they are handled wrong.

When you’re comparing dialler platforms, look for controls that are built in rather than bolted on, so compliance doesn't depend on someone remembering to configure it.

Here are some key controls to look out for:

• Pacing that aims for zero abandoned calls. A call is abandoned when a customer answers and no agent is free. Ofcom's target is 0%, not the old 3% figure some diallers are still set to.
• 72-hour rule handling. A number that received an abandoned call should only be recalled within 72 hours if an agent is guaranteed, for example by switching it to preview mode.
• Minimum ring time, in line with Ofcom's 15-second guidance.
• Valid CLI on every call. See our guide to CLI compliance.
• Do Not Call and TPS screening, so registered numbers are removed automatically.
• Calling windows, configurable for each campaign.
• Audit trails, with call recording and data capture, so you can evidence compliance if Ofcom or the FCA ask.

For more on how the rules work in practice, see our guide to Ofcom and PECR regulations.

4. Number recycling and retries

Many calls go unanswered the first time, so a good dialler doesn't just move on, as that means lost leads. Instead, it puts unanswered numbers back in the queue to be retried, so more leads are reached, and it keeps how often each contact is called within official limits, so they're never pestered. Look for control over how many attempts each number gets and how long the dialler waits between them, so you can retry at different times without overdoing it. It should also handle customers who ask to be called back later, so those conversations aren't lost.

5. CRM integration

A powerful dialler is even better when it works hand in hand with your existing systems. It gives an agent access to customer history, preferred contact methods, and past feedback – all within the dialler interface.

This allows for personalised conversations that address specific needs, leading to happier customers and improved call outcomes. Easy integration applies to after-sales and debt resolution teams too.

By connecting your dialler with other systems, you can put all relevant information at agents’ fingertips, reducing the time spent searching for information. MaxContact connects with Salesforce, HubSpot, Zoho and AutoConvert, with open APIs for other systems.

6. Outbound skill-based routing

Skill-based routing isn't only for inbound calls. Outbound skill-based routing matches each lead to the agent best placed to handle it, based on skills and proficiency, and it works in preview, progressive and predictive modes. It stops high-value or sensitive leads reaching an agent who isn't equipped for the conversation. Read how outbound skill-based routing works.

7. Reporting and dashboards

You can only improve contact centre performance when you can measure it. Look for reporting and analytics dashboards that give you complete visibility around call metrics and agent performance, as well as scheduled reports so you can set targets and track them across campaigns, teams, and agents.

Choosing an outbound dialler

Use this list as a checklist when comparing providers, whether you're running sales, debt collection or customer service campaigns. To see how MaxContact's dialler handles these features, explore our outbound dialler software or book a demo.

Compliance used to be something you could, quietly, budget for. A fine here, a sample there, a QA process that reviewed a small slice of calls and hoped the rest looked the same.

That maths no longer works.

In 2026 the cost of getting it wrong has changed by an order of magnitude, and regulators have moved from asking whether you’ve done the work to asking you to prove it. In our recent webinar, Kayleigh Tait and Conor Bowler walked through what’s changed, shared new research from more than 300 UK contact centre leaders, and showed how MaxContact’s Auto-QA closes the gap. Here’s the summary.

‍

The 2026 compliance landscape

Two things have shifted at once.

First, the numbers. PECR - the regulation governing outbound calls, texts and marketing communications - used to cap fines at £500,000. Since the Data (Use and Access) Act came into force in February 2026, that cap has risen to £17.5 million, or 4% of global turnover, whichever is higher. That’s a 35x increase in maximum exposure. And for the first time, company directors can be held personally liable, up to £500,000 each.

Second, the FCA’s stance. Consumer Duty has been law since 2023, but the regulator has moved from “have you put this in place?” to “prove it with evidence.” Of the 180 board reports the FCA reviewed last year, the most common failure wasn’t that firms hadn’t done the work - it was that they couldn’t evidence the outcomes for vulnerable customers.

At these levels, a fine isn’t a line item you plan for. It’s a risk you have to design out.

‍

What the research found

To ground this in reality rather than headline numbers, MaxContact commissioned independent research across 300+ UK contact centre leaders, 285 of them FCA-regulated. A few findings stood out.

Coverage has matured - but gaps remain. Asked how they review calls today:

• 49.5% still sample manually
• 46.7% already use AI to review every call
• 2.8% review nothing systematically

That’s a more mature picture than the “2–3% sampled” figure often quoted in industry research. But it also confirms that a real share of organisations still aren’t solving the problem with technology.

Sampling leaves you exposed. More than half - 54% - said a compliance breach or harm had occurred outside their routine QA process. 16.5% said it had happened more than once. The point is simple: if you only look at a sample, the problems tend to live in the calls you didn’t look at.

And it costs real money. Across everyone surveyed - including firms that paid nothing - the average regulatory fine in the last 12 months was £81,000, with 31% paying £50,000 or more. Fines weren’t the whole story either: the average lost revenue from non-compliant sales (refunds, cancellations, deals that fell through) was £22,000, and 50.2% said they’d lost a sale, contract or customer over a compliance issue.

‍

‍

The confidence gap

One of the more revealing findings came from asking teams two questions. First: how confident are you that you could evidence fair treatment if the FCA came knocking? Confidence was high across the board.

Then we flipped it: have you actually found a breach outside your QA sample? The gap between the two is the interesting bit.

It’s less about how hard a team looks and more about what they’re looking at. Sales and collections calls tend to follow scripts and structured flows, so there are only so many ways a conversation can drift out of compliance. Customer care and technical support calls are far less scripted - troubleshooting, escalations, one-off advice - so there’s more variance per call, and more chance of a breach hiding in the calls that never make it into the sample.

‍

The feedback delay problem

Even when issues are caught, they’re caught slowly. On average, it takes 3.08 days between a call happening and the person who took it getting feedback. Only 10.9% hear back within a day; over a third (37.2%) wait three days or more.

The blocker isn’t attitude. Time, cost and headcount accounted for 37.4% of the reasons given, while only 6% felt there was no genuine need for faster feedback.

That delay matters, because feedback has a shelf life. Third-party research shows employees are 3.6x more likely to say they’re motivated to do outstanding work when feedback comes daily rather than at a quarterly or annual review. Put the two together and the risk is clear: if something’s going wrong on a call and nobody flags it for the best part of a working week, it’s probably happening again and again in the meantime.

‍

AI Call Scoring vs Auto QA at Scale

MaxContact addresses this with two features inside Conversation Analytics. They’re easy to blur together, so it’s worth being precise.

AI Call Scoring is an AI scorecard builder, included in the Conversation Analytics base package. You write the scorecard in plain English, and it scores individual calls — a human still picks which calls to run it against. It cuts review time from 2–3x the length of the call down to around five minutes per call.

Auto QA at Scale takes those same scorecards and runs them on a schedule - historically and as new calls come in - so you get consistent coverage across 100% of eligible calls. Every result is backed by evidence in the transcript, and calls are grouped by outcome (pass, fail, auto-fail, not applicable) so your team can focus human review where it’s needed.

‍

What Auto-QA looks like in practice

In the demo, Conor built an FCA compliance scorecard and showed how it runs at scale. A few things worth knowing:

• Build the scorecard the way that suits you. You can draft it in a tool like Excel first - using comments and track changes to collaborate - then bring it into Conversation Analytics. Each criterion is defined across what to detect, what to listen for, and how to score it, using a decision tree where possible to keep scoring consistent.
• It handles vulnerability. The scorecard can detect drivers of vulnerability across health, life events, resilience and capability, and score whether the agent acknowledged, semi-acknowledged or missed the indicators - directly relevant to the FCA’s focus on vulnerable customers.
• Schedules do the running. Choose an always-on schedule (new calls scored as they come in) or a one-time run (for example, 5% of last quarter’s calls to check historical compliance). Add rules - minimum call length, successful outcomes only - set the sample size, activate, and it runs.
• Filter and report on what matters. Build views by result - passes, fails, auto-fails - and drill into any call to see the evaluation summary and exactly why it failed. Performance reporting breaks results down by campaign, team and user, with more question-level reporting arriving shortly.

‍

Common questions from the session

Isn’t AI scoring just swapping one compliance risk for another? No - because it isn’t a black box. Every score links straight back to the exact part of the transcript it came from, so a QA lead can check any result against the call in seconds. The AI decides what needs looking at; a human still decides what to do about it.

We already run Conversation Analytics with AI Call Scoring - is Auto-QA a big project? No. It sits on infrastructure you already have, so it’s a matter of turning the feature on. Your existing scorecards carry straight over — you’re not rebuilding anything. What changes is that scoring runs on a schedule against every eligible call, rather than a person choosing which calls to score.

Can results be shown to agents, not just QA? This is in development, and it’s permissions-based - you control the level of detail. The plan is three layers of feedback: calls scored in real time as they happen, a daily “top three to improve, top three strengths” summary, and the same across a rolling seven-day view.

‍

The takeaway

The rules have changed, and sampling no longer counts as evidence. When breaches hide in the calls you don’t review, and feedback takes three days to land, the fix is coverage that’s complete, evidenced and fast. That’s exactly the gap Auto-QA is built to close.

‍
About MaxContact

MaxContact is an AI-powered customer engagement platform that helps businesses turn every customer conversation into a revenue-driving outcome. Our platform spans contact centre software, conversation analytics, and AI agents and chatbots - working together as one connected solution. Book a demo.

‍

Auto QA scores 100% of your eligible interactions against your own criteria and links every score back to the exact exchange in the transcript. When a regulator, an ombudsman or the board asks about a specific call, the evidence already exists.

Reviewing every call turns out not to be the finish line

We surveyed 300 UK contact centre managers and directors in August. 285 of them operate under FCA regulation, and among that group, 46.7% already have AI reviewing every single call. For years the case for automating Quality Assurance was that you could only ever listen to a fraction of your conversations. For half this market, that's no longer the problem. So, we asked those same 285 people whether they'd ever found a compliance breach or harm issue outside their routine Quality Assurance sample.

54.0% said yes. 16.5% had found more than one.

Reviewing everything, it turns out, isn't the same as being able to prove anything. Consumer Duty doesn't ask whether a conversation was recorded or even scanned. It asks for an assessment against defined criteria, applied consistently, with an audit trail and a demonstrable link between what you found and what you did about it. Plenty of businesses now have automated review across all their calls. Far fewer could produce that chain for one named customer, on one named call, this week.

The part that costs money is the wait

The other finding that shaped this product was about speed. Only 10.9% of regulated – businesses get feedback to an agent within a day of the call. The mean is 3.08 days. And when we asked what stops teams reviewing more, the answers were about resource rather than capability - time, cost and headcount accounted for 67.4% between them. Three days is a long time on a contact centre floor. A missed disclosure or a mishandled vulnerability marker usually isn't a one-off; it's a habit, and it carries on across conversations nobody has flagged yet.

The end of the Quality Assurance Sample

Every eligible interaction is scored automatically, as soon as the transcript is available. There are no selection step and no queue. Scorecard criteria are written in plain English rather than built from keyword rules or complex logic, so your Quality Assurance team can create and change them quickly and without waiting on us. Every score is evidence-linked, so a result can be defended in a coaching session, an internal audit or a regulatory review.

And because it's one evidence layer rather than three, Operations, Quality Assurance and Compliance are working from the same view of the same conversations instead of separate samples and separate conclusions. Your reviewers keep the final say throughout. They can challenge a score, question an output and recalibrate criteria whenever the operation changes.

If you already use MaxContact’s Conversation Analytics, you already have AI Call Scoring. That lets your team score selected calls against your existing scorecard, taking a review from roughly thirty minutes down to five - around four days a month back for a reviewer. Auto QA is the paid add-on that removes the selection step entirely and turns that output into a complete, auditable record across every call.

Available today

Auto QA is available now as an add-on to Conversation Analytics. Existing customers can speak to their Customer Success contact about switching it on. If you'd like to see it working on your own call volume rather than ours, book a demo and we'll walk you through it.

Sales want to push performance. Compliance want to protect against risk. 77.5% of the managers and directors in our research agreed that pushing performance raises compliance risk - so this isn't a tension anyone needs convincing of. Score every call and both teams are at least arguing from the same evidence.

Let sales win. Let compliance sleep.

The full research is in Confidence Is Not Evidence: The 2026 UK Conduct Risk Report - findings from all 285 regulated respondents, plus five questions we think belong in a Consumer Duty board report this year. For more on how automated Quality Assurance works in practice, see AI Quality Assurance in Contact Centres: Why Reviewing Every Call Isn't Enough.

Tracking KPIs quarter-on-quarter tells you whether you're improving. It doesn't tell you whether the number you're improving towards is actually good. And it doesn't tell you whether your measurement is giving you an accurate picture of your operation in the first place.

That last point is where the real performance gap comes from.

Two thirds of UK contact centres are now using or piloting AI, with 99% of those operations feeling confident in its impact. Yet across the metrics that matter most, performance has barely shifted year-on-year. FCR is flat or declining. Conversion rates are down. AHT is unchanged. Agent workloads are up 10%. The sector has invested heavily in technology but the performance gap remains as wide as ever.

The reason isn't that AI doesn't work. It's that most organisations use it in response to visible symptoms such as reducing call volume or speeding up response times, which leaves poor performance problems unaddressed at the root cause.

To understand where that gap sits, MaxContact surveyed 300 UK contact centre decision makers across sales, customer service and debt collection for our 2025-26 KPI Benchmarking Insights Report. It's the second year we've run the research, so we can track how performance is shifting over time.

The measurement problem nobody's talking about

Before we get to the performance benchmarks, let's draw your attention to the issue that runs through all of them.

Most contact centres measure performance through post-call surveys, manual QA sampling and periodic reporting. Post-call surveys capture a small sample of interactions, skewed toward customers at the extremes of satisfaction. Manual QA in organisations we speak to reaches max 5% of calls. Periodic reporting tells you what happened last month.

At that sample rate, you can identify which metric is underperforming but you can't reliably identify why. And if you can't identify why, you can't fix it. You're just guessing, which is what most performance optimisations actually are, dressed up as data-driven decision making.

This matters because every benchmark in this piece is only useful if your measurement is telling you the truth. A CSAT score built on 10% of calls, an FCR rate from manually sampled interactions, a conversion rate that looks reasonable against a sector average you've never questioned. These numbers can give you false confidence about performance you haven't really achieved.

If you want to review how your operation is capturing performance data, our complete guide to call centre reporting metrics covers the full picture. Keep that in mind as we work through each benchmark.

CSAT: the conditions matter more than the score

48% of contact centre leaders rank CSAT as a top 5 KPI

CSAT is the metric most contact centre leaders care about most. It's also the one where the gap between stated priority and operational reality is widest.

There's no single sector mean for CSAT in our data. Satisfaction scores vary too much by sector, use case and measurement method to produce a meaningful average. That variability is the first signal that something's wrong with how most operations approach it.

The more useful question isn't "what's a good CSAT score." It's "what conditions in your operation are producing the score you have." And on that measure, the benchmark data reveals a problem most operations aren't looking at directly.

Agent workloads increased at 53% of contact centres last year, a 10% rise on the year before. Annual agent turnover sits at 31.2% on average, with some operations reporting churn of 41-50%. These metrics are the direct mechanism through which CSAT deteriorates. Stretched agents handle calls differently, and high turnover means less experienced people managing more complex conversations. Neither produces the kind of consistent interactions that drive strong satisfaction scores.

Only 10% of contact centres in our survey reported annual churn below 20%. Those operations aren't just performing better on a staffing metric. They're building the kind of service consistency that CSAT is actually measuring. For most, high turnover is a structural drag on service quality that no post-call survey will fix. The survey measures the symptom. The cause goes unaddressed.

What good looks like also depends on your use case. In outbound sales, CSAT is partly a lead quality and script problem. Customers who feel mismatched to a product will score you poorly regardless of how the agent handled the call. In inbound customer service, resolution and wait time are the key drivers. The levers are different, so the benchmarks should be treated differently too.

The contact centres with the strongest CSAT scores aren't waiting for survey results to tell them something's wrong. They're using Conversation Analytics to understand sentiment across 100% of calls, spotting where calls are going wrong before it shows up in a survey score and before the customer has already decided not to come back.

Speed of answer: you're prioritising a metric you're not hitting

Sector mean: 17.4 seconds. High performer target: under 10 seconds

Speed of answer is the second most prioritised KPI in our survey, with 35% of respondents ranking it in their top five. The benchmark data reveals something uncomfortable about it.

Across all contact centres, the mean is 17.4 seconds. The high performer target is 10 seconds. Nearly 40% of respondents are already hitting between 2 and 10 seconds. That's the bracket that defines good. Therefore,  most contact centres are actively prioritising a metric they're not hitting, against a standard they may not have explicitly set.

The difference between 17 seconds and 10 seconds may not sound critical. But for a customer who has already decided to pick up the phone, who has exhausted self-service and committed to the friction of a call, every additional second in a queue adds to existing frustration. Our data shows mean call abandonment sits at 4.1%. Operations answering in under 10 seconds are less exposed to that abandonment, and the customers they do lose are disproportionately the ones closest to walking away entirely.

Speed of answer is primarily an inbound metric. For outbound sales and debt collection, the equivalent is connect rate, the percentage of dialled numbers that reach a person. The sector mean is 42.8%, with high performers achieving 60-79%. If your operation is outbound-led, that's the speed benchmark driving your performance.

For inbound teams, closing the gap between 17 and 10 seconds comes down to workforce management and call deflection. Smarter IVR design, AI-assisted resolution of routine queries, and capacity planning that reflects actual demand patterns. Our report shows 38% of respondents are already using predictive analytics for capacity planning. Those operations hold speed of answer during demand spikes without the cost of permanent overstaffing.

Service level achievement: a baseline, not a definition of good

34% of contact centre leaders rank service level achievement as a top 5 KPI

Service level achievement sits third in our KPI rankings. It's also the metric where the definition of good is most taken for granted.

The standard most operations work to is the 80/20 rule: 80% of calls answered within 20 seconds. It's been around long enough to feel like a law of physics rather than a performance target. Most contact centres treat hitting it as the definition of good when it isn't. It's the definition of acceptable.

The operations genuinely performing well on service level aren't managing to 80/20. They're using it as a floor. The 39% of respondents answering calls within 2-10 seconds haven't set a different target. They've built operations that outperform the standard because their workforce management, demand forecasting and deflection are sophisticated enough to absorb volume spikes without queue times creeping up.

The biggest threat to service level is workload. Agent workloads increased at 53% of contact centres last year. When demand rises faster than capacity, service level is the first metric to suffer and the one customers feel most directly. You can have strong FCR and solid CSAT and still lose customers at the queue stage because they simply couldn't get through.

For outbound operations, service level works differently. It's less about queue management and more about campaign pacing, maintaining consistent contact rates without burning through call lists or hitting compliance thresholds. Good performance in any channel means holding your numbers when conditions get difficult, not just when demand is manageable.

The 38% of respondents using predictive analytics for capacity planning are better placed here than those still forecasting from historical averages. Demand patterns have shifted, and operations working from last year's data are likely to be caught short. Service level is where that shows up first.

80/20 is where you start. Consistently beating it is what good actually looks like.

Conversion rate: you're solving the wrong problem

Sector mean: 16%. High performer bracket: 20-29%. MaxContact platform average: 38.8%

Most contact centres trying to improve conversion rate start in the wrong place. They review scripts. They add coaching. They push agents to make more calls. The data suggests none of those things address the primary problem.

The sector mean conversion rate is 16%. Average daily calls per agent have fallen from 65.5 to 59 year-on-year. First-call close rates have dropped from 28% to 25%. That looks like a performance decline. But consumer confidence has fallen sharply over the same period. Agents aren't getting worse. They're working harder against a more resistant environment. Pushing them to make more calls doesn't close the gap. It increases attrition.

The operations holding conversion rates above 20% are doing it by being more precise about when they call, who they call, and which dialling mode they use for which campaign type. Most operations run one dialling mode across all campaign types and wonder why conversion is flat.

To show how wide that gap can be: across 7.8 million calls on the MaxContact platform, our customers achieve a 38.8% success rate against a sector mean of 16%. That difference doesn't come from better agents. It comes from smarter contact strategies. Predictive dialling for high-volume campaigns, preview dialling for high-value or sensitive conversations, and call timing built around when a specific customer is likely to answer.

For debt collection, the benchmarks shift but the underlying point holds. Promise to pay rate sits at a sector mean of 28%, and good is above 30%. Right party contact averages 27%. FCR in debt collection has dropped five percentage points to 37%, reflecting economic pressure on consumers rather than operational failure. But 15% of debt collection operations are still hitting collection rates of 40-49% in those same conditions. Reaching the right person at the right moment with an agent trained for a sensitive financial conversation produces a fundamentally different result than high-volume dialling and hoping for the best, and the operations hitting those numbers know it.

If your conversion rate is below the sector mean, the most useful question isn't "how do we improve our scripts." It's "are we contacting the right people, at the right time, in the right way" and do we have the data to answer that honestly.

FCR and AHT: the metric relationship that determines everything else

Inbound FCR mean: 40.9%. High performer bracket: 50-79%. Mean AHT: 7.8 minutes

FCR and AHT are the two metrics most likely to be misread in isolation and most likely to be managed against each other in ways that make both worse.

The pattern is familiar. Leadership sees AHT creeping up and pushes agents to handle calls faster. AHT comes down. FCR comes down with it, because customers whose issues weren't properly resolved call back. Cost per contact rises. CSAT drops. The operation has optimised for a number and degraded its actual performance in the process.

The metric that looked like it was improving was masking a problem getting worse. This is why FCR and AHT need to be read together, always.

The inbound FCR mean is 40.9%. High performers, the top 27% of respondents, achieve 50-79%. That's the bracket to aim for. Mean inbound AHT is 7.8 minutes, with over a third of respondents reporting 10-15 minutes. That higher figure isn't a problem in isolation. A 12-minute call that resolves the issue completely is a better outcome for the customer, the agent and the cost base than a 5-minute call that generates a callback.

Good isn't a low AHT. It's a high FCR with an AHT that reflects the complexity of what your operation actually handles.

The use case split matters here. In inbound customer service, FCR above 50% is the high performer threshold. The operations hitting it aren't rushing calls. They're doing it through better agent knowledge, faster access to customer history, and calls that get to the point without losing the quality of the resolution. In debt collection, FCR has dropped five percentage points year-on-year to 37%, reflecting economic pressure on consumers. The 15% of debt collection operations achieving 40-49% FCR in those conditions are differentiating on contact precision and conversation quality, not volume. For outbound sales, the paired benchmarks that matter are calls to success ratio, with a mean of 15.6 calls per sale, and handle time, where 4-5 minutes is the sweet spot.

The performance gap on FCR is also where the measurement problem bites hardest. Most operations run manual QA on 2-5% of calls. At that sample rate, you can identify that FCR is low but you can't reliably identify why, which means you can't fix it systematically. Is it a knowledge gap? A script failure? A specific objection type agents aren't equipped to handle? A process breakdown adding unnecessary handle time without improving resolution? You can't answer those questions from 2-5% of your call data.

If AHT is the metric you're most focused on right now, our guide to reducing average handle time without sacrificing customer experience covers the specific tactics for closing that gap without pushing FCR in the wrong direction.

Why the performance gap isn't closing

Two-thirds of UK contact centres are using or piloting AI, and sentiment is overwhelmingly positive. So why haven't the numbers moved?

The answer is in where the investment is going. Chatbots and virtual agents are the most common AI applications in our survey, at 57% and 56% respectively. Both are built to stop calls reaching agents in the first place. That's a legitimate goal. But deflection doesn't tell you anything about the conversations that do get through. It can't explain why your FCR is sitting at 40% when high performers are hitting 60%. It won't surface the objection that keeps killing conversions at the same point in the call. It has no view on where resolution breaks down, or why.

The contact centres with the strongest metrics aren't necessarily running more AI tools. They're running them in a different place. Conversation Analytics applied across 100% of calls gives you a fundamentally different picture than manual QA on 2-5%. You can see patterns. You can identify what high-performing agents are doing differently. You can stop guessing at why a metric is underperforming and start knowing. Most contact centres already have the technology to do this. The question is whether it's being pointed at the right thing.

Where does your contact centre performance sit?

These benchmarks aren't a verdict. An FCR of 38% or a conversion rate below the sector mean doesn't mean your operation is in trouble, it means you now have a clearer sense of where the gap is and roughly how wide it is. Most operations never get that far.

The contact centres pulling ahead aren't doing anything the average operation couldn't. Agent stability is managed deliberately rather than accepted as an inevitability. Dialling strategy is matched to campaign type. FCR and AHT are read together. And critically, performance decisions are based on what's happening across the full call volume, not the fraction that manual QA and post-call surveys ever reach.

That last point is what makes underperformance so persistent. If your measurement is only capturing part of what's happening, you're optimising against an incomplete picture. The benchmarks in this report are only useful if the data behind your own metrics is sound.

For a broader look at efficiency measurement across your operation, our guide to measuring call centre efficiency is worth a read.

Want to go deeper on the data? The full 2025-26 UK Contact Centre KPI Benchmarking Insights Report covers all 300 respondents, segmented by use case, with year-on-year comparisons across every metric in this piece.

Download the 2025-26 UK Contact Centre KPI Benchmarking Insights Report

‍

Vague objectives mean QA scorecards and reporting tend to focus on superficial metrics rather than what moves the needle. Even a well-configured platform struggles to prove ROI if nobody has defined what "working" looks like in the first place.

This article covers what a good setup looks like, where implementations tend to go wrong, and what results contact centres can expect. Whether you're evaluating speech analytics software or about to switch one on, this is the practical detail worth knowing before you commit.

How Conversation Analytics powers effective quality assurance

Conversation Analytics is MaxContact's speech analytics platform. It transcribes every call, identifies patterns in topics and phrases, and surfaces sentiment across the full call population rather than a sample.

Two capabilities sit on top of that data, and they solve different problems. AI Call Scoring lets your QA team select which calls to score against your existing scorecard. Using transcript evidence, a traditional thirty-minute review becomes a five-minute one. Auto QA at scale goes further: every eligible interaction is scored automatically as soon as the transcript is available, with no selection step at all.

There's also AI Assist, which works alongside the scorecard. Once a call is transcribed and scored, you can ask it open questions a fixed scorecard was never built to answer, such as how could this agent have shifted the sentiment of that call, or where was the actual turning point in the conversation. It's closer to having a coaching conversation about a specific call than reading a score with supporting evidence.

Scoring quality will always depend on what data feeds it. Accurate transcription, well-categorised topics, and clearly defined criteria all come out of Conversation Analytics itself, which is why the setup sequence is a critical part of implementation.

Start with outcomes, not configuration

When clients implement our Conversation Analytics and Auto QA capabilities, we don't rush into configuration. Instead, we start with one question: what does success look like for your contact centre environment?

Success for a financial services contact centre managing FCA Consumer Duty obligations looks much different compared to success for a BPO needing to reduce QA workload.  Which looks different again from a sales team trying to understand what separates their best converting calls from average ones. Conversation Analytics and Auto QA can serve all of those outcomes. But they can't define them for you.

The contact centres that get the most from the platform do this work before implementation begins. "Better quality assurance" is not a useful benchmark. A clear definition means specifying what you're trying to improve, how you'll measure it, and what meaningful progress looks like at 30, 60, and 90 days.

That definition shapes everything that follows: which scorecards you build, which call types you prioritise, how you configure Auto QA, and how you report results internally.

Why the setup sequence matters

Each stage of implementation exists to reduce risk in the one after it. Skip or rush the discovery stage and it is much harder to prove ROI later on.

Why Topic Discovery changes everything

Topic Discovery is the step most implementations underestimate.

Before any scorecard gets built, Conversation Analytics analyses existing call data and categorises what's being discussed; the topics, phrases, and patterns from actual interactions, rather than what a team assumes is happening based on memory or anecdote. Scorecard criteria built on that foundation reflects the calls you're actually having, which shows up directly in how reliable the Auto QA scoring ends up being.

Conversation Analytics: Realistic results and timelines

One question most people ask is how quickly you'll see a return. Efficiency improvements are different from quality improvements, and the two don't follow the same timeline.

Efficiency gains are immediate

Review time drops from day one. This part is structural rather than something that builds up gradually. As soon as scoring is live, review time falls from around 30 minutes per call to somewhere in the 2-5 minute range, depending on call complexity.

The results across our customer base are consistent:

• An automotive contact centre reduced QA review time from 30 minutes to approximately two minutes per call, seeing a 93% reduction
• An education provider reduced call review time from 30 minutes to three minutes, a 90% reduction
• A BPO running Auto QA across 45 client scorecards confirmed 83-84% of QA tasks as automatable

Broader QA improvement takes longer.

Broader improvements, such as compliance rates shifting or call quality lifting across the board, take longer to show up, and the exact timeline varies by customer size and sector. Based on what we've seen so far, a window of roughly 4-6 weeks is a reasonable starting expectation, though this depends on how much call volume you're scoring and how quickly coaching conversations are acted on. The platform creates visibility; but what your team does with it determines the pace.

ICX moved from time-intensive manual call reviews to surfacing compliance issues as they happened, and then used that visibility to build a smarter coaching programme. Read the ICX Conversation Analytics case study.

What mature Auto QA looks like in practice

Most operations start with the efficiency win; faster reviews, full call coverage, and that's the right place to start. The difference in mature operations is what happens to the output after scoring, not the scoring itself.

In regulated environments such as financial services, insurance and debt collection, that means treating Auto QA output as compliance evidence rather than a coaching tool that happens to also help with audits. Every scored call becomes part of a documented trail covering the whole call population, which is a different proposition to a QA reviewer's sample.

For sales and collections teams, the same principle plays out differently: Conversation Analytics surfaces objection handling patterns and sentiment shifts through a call, showing which phrases and approaches tend to show up in calls that convert.

In both cases, the output only earns its value once someone's acting on it. Scoring feeds regular coaching conversations, compliance drift gets caught before it embeds itself, and team leads use the platform directly rather than waiting on a report from someone else. That level of adoption tends to build during training, as teams get more comfortable acting on what the data's telling them.

What to look for when evaluating a Conversation Analytics platform

Not all conversation analytics and quality assurance software works the same way. The differences that matter most aren't always visible in a demo. These are the questions worth asking before you commit.

Is scoring evidence-backed or black-box?

Some tools return an AI-generated score without showing what produced it. That's a problem the moment you need to defend a QA decision in a coaching conversation, a compliance audit, or to a regulator directly. A score should link back to the exact transcript exchange behind it.

Can you build scorecards without technical configuration?

Defining scorecard criteria in plain English, rather than through keyword rules or technical logic, determines how quickly your QA team can build, test, and refine scoring frameworks without depending on vendor support. It also determines how well scoring reflects the nuance of real conversations.

Does the platform distinguish between selective scoring and always-on coverage?

As covered above with AI Call Scoring and Auto QA at Scale, this isn't just semantics. For regulated contact centres, the distinction has direct compliance implications, and it's worth confirming which one you're actually being sold.

Is human oversight built in?

Auto QA should support your QA team. Reviewers need to be able to question outputs, challenge scores, and recalibrate criteria as things change. The governance stays with the people who understand the operation.

What does implementation actually look like?

How success outcomes are defined, how Topic Discovery informs scorecard build, what training and hypercare look like; these are worth asking of any vendor you're evaluating. Implementation is where the gap between a platform that delivers and one that doesn't usually opens up.

Ready to see it in action?

The difference between a speech analytics rollout that delivers and one that disappoints rarely comes down to the technology itself; it usually comes down to how it's implemented, how success gets measured, and how quickly a team turns what it's shown into something it acts on.

MaxContact combines Conversation Analytics, AI Call Scoring, Auto QA at Scale, Topic Discovery, training, and ongoing Customer Success support in a structured implementation process built around that. Whatever you're trying to fix, whether it’s QA workload, call coverage, compliance monitoring, or getting more out of the conversations you're already having, we'll work with you to define what success looks like and build toward it.

Book a demo to see how Conversation Analytics and Auto QA work in practice.

‍