Most contact centre CSAT programmes measure the wrong moment. They survey after the call's over, average the scores, slap a target on the wallboard, and wonder why the number drifts sideways for a decade. We've sat through enough QBRs to know how this story ends.
The teams that actually move customer satisfaction don't chase a number. They fix the four or five points in the journey where customers decide whether they trust the brand. One of those points is the payment moment — and almost nobody isolates it in their CSAT data, which is why we keep finding it as the silent killer of otherwise healthy scores.
Here are ten things that work, ranked roughly by how much lift we've seen them deliver in UK and US contact centres. Some are obvious. A couple aren't. The last one is the reason we built DTMF masking in the first place.
1. Cut average hold time before you touch anything else#
The single strongest predictor of a bad CSAT score is time on hold. Not call length, not transfers, not agent tone — hold time. The CCMA's 2025 UK Contact Centre Benchmark put the median hold time at 47 seconds and the 90th percentile at over four minutes. The 90th percentile customers are the ones writing the one-star reviews.
Before you spend money on AI coaching or speech analytics, audit your hold queue. Half the holds in most centres are agents looking up information they shouldn't have to look up. The fix isn't faster agents. It's better knowledge base structure, fewer disambiguation steps in your IVR, and a desktop that doesn't require six tabs to confirm an address.
2. Stop forcing customers to repeat themselves#
If the customer's already entered their account number in the IVR, the agent shouldn't ask for it again. We see this every week and it's the cheapest CSAT lift in the business. Pass the IVR data into the agent screen. Pass the channel history too — if they emailed yesterday, the agent should know.
Forrester's 2024 customer experience research found that repeat-context calls scored 32% lower on CSAT than calls where the agent already had context. That's not a coaching problem. That's a screen-pop problem.
3. Train agents on the recovery, not the script#
Scripts make new agents look competent. They don't make experienced agents great. The CSAT difference between a six-month agent and a two-year agent isn't script fluency — it's how they handle the moment a call goes sideways.
We've watched agents save calls that started with a furious customer just by saying "I can see why that's frustrating, let me sort it." No script. No de-escalation framework. Just a person taking ownership. Train for that, measure it, promote on it.
4. Fix your post-call work, not your call length#
Average handle time targets are a tax on CSAT. Agents who feel timed rush. Rushed agents miss the thing that would have prevented the callback. The callback is the CSAT killer, not the extra 40 seconds.
If you must measure something, measure first contact resolution. Our friends at the CCMA report FCR sits at around 71% across UK centres, with the top quartile pushing 84%. The gap between those two numbers is where most of your CSAT loss lives. There's more on this in our piece on improving call centre customer satisfaction if you want the operational deep-dive.
5. Survey at the right moment with the right question#
Most CSAT surveys go out hours after the call, when the customer's moved on. The few people who respond are either ecstatic or furious. The middle 60% — the customers whose experience actually tells you something — never reply.
Send the survey within five minutes of the call ending. Ask one question, not seven. "Was your issue resolved?" beats "Rate your agent's professionalism from 1 to 10" every time. The first question tells you something you can act on. The second tells you nothing.
6. Make callbacks the default, not the exception#
If wait time exceeds 60 seconds, offer a callback. Most queue platforms can do this today. Most centres don't enable it because they're afraid of the call-back commitment.
Customers who take a callback rate their experience higher than customers who waited the same length of time in queue. They feel respected. Their time wasn't wasted. The data on this is unambiguous — we've seen 8-12 point CSAT lifts from callback adoption alone.
7. Give agents the authority to fix things#
Nothing tanks CSAT faster than "I'll have to check with my supervisor." Every escalation is a CSAT hit, even when it ends well. The customer wanted resolution, not a queue ticket.
Pick five common issues. Give frontline agents the authority to resolve them — refunds up to a threshold, fee waivers, courtesy credits, expedited shipping. Audit monthly. The fraud cost is dwarfed by the retention gain. We've seen this work in insurance, retail, travel and utilities. It's not industry-specific.
8. Separate the channels but keep the context#
Voice, chat, email, and WhatsApp all have different CSAT dynamics. Don't average them together — you'll mask the failing channel. Chat CSAT tends to run 8-12 points higher than voice, but for a different reason than people think: chat customers self-select for non-urgent issues. Voice gets the hard problems.
Measure each channel separately. Stop comparing them. The right question is whether voice CSAT for complex issues is better this quarter than last, not whether chat is beating voice.
9. Reduce the things that aren't worth asking about#
Every authentication question, every "can I have your postcode for verification," every "sorry, I need to read this disclaimer" — these are CSAT taxes. Most are there because someone in compliance asked for them eight years ago and nobody's removed them since.
Audit your call openers. Audit your verification flow. Audit the wording. There's a version of every required disclosure that's 40% shorter and still legally sound. Customers don't want efficient. They want respectful — and efficient feels respectful when the alternative is wading through boilerplate.
10. Fix the payment moment#
Here's the one most CSAT programmes miss. Look at your CSAT data and split it by whether the call included a payment. If you don't have that split, build it tomorrow. We bet your payment-included calls score 6-15 points lower than your non-payment calls.
The reason is usually one of three things. Either the agent put the customer on hold to take card details "securely" (40+ seconds of dead air, CSAT crater). Or the customer was bounced to an IVR mid-call to enter card details (transfer, friction, drop-off). Or the agent stayed on the line and the customer read their card number out loud while feeling watched (no recovery from that feeling).
The fix is DTMF masking with channel separation. The agent stays on the call. The customer types the card on their keypad. The tones are masked from the agent and from any call recording. The agent sees the transaction confirm in real time. No hold. No transfer. No awkwardness. Average call CSAT lifts 5-10 points within a quarter, every time we see it deployed properly.
Most contact centre leaders don't think of this as a customer satisfaction lever. It's filed under "PCI compliance" and ignored by the CX team. That's a mistake. The payment is the moment of highest customer attention in the entire call. If it feels awkward, the whole call feels awkward retrospectively.
What to stop measuring#
While you're at it, stop measuring agent CSAT in isolation. It's a leadership metric, not an agent metric. When you put it on the agent's scorecard, you create incentives to game the survey rather than improve the call. Measure team CSAT. Coach individuals on the recordings that tell a story, not on the score itself.
And stop reporting NPS quarterly. NPS at the call level is too noisy to coach with. Save NPS for the relationship survey, twice a year. Use CSAT, FCR, and customer effort score at the interaction level. That's a metrics stack you can actually act on.
What we'd do first#
If you ran a 300-seat contact centre and could pick three things to do this quarter, we'd pick: hold-time audit, callback enablement, and payment moment fix. Those three between them tend to deliver 70% of the available CSAT lift, and none of them require a transformation programme.
The payment fix is the one most teams skip because it sounds like an IT project. It isn't. A modern secure telephone payments setup deploys in days, not months, and the CSAT data starts moving the same week. If you'd rather see how we do it for similar teams, our contact centres page has a few customer outcomes worth a read, and the CSAT glossary entry covers the measurement basics.
Final thought#
CSAT isn't a number. It's a proxy for whether customers think you respect them. Most of the ten levers above come back to that one idea. Respect their time. Respect their context. Respect the awkwardness of paying over the phone. Get those right and the score follows.
