When everyone can do it, what will customers pay you for?

If you had to call a financial services firm because someone in your family had just died, would you want to talk to a person or a machine?

I would have assumed a person. So did Nikola Mrkšić, co-founder of PolyAI. When his team analysed the call volumes at a major financial client, bereavement calls were near the top. Nikola’s immediate instinct was simple: route them straight to an experienced human. But legacy technical constraints made that impossible.

So they built an automated agent instead. It spoke calmly, listened without rushing, and walked callers through the physical claim process step by step.

Nikola told me that customer satisfaction scores on those bereavement calls were noticeably higher than on the human-handled lines. He refused to believe it. He went back and had the team re-run the numbers. The data held.

Because when someone has just lost their partner, they don’t want a junior call-centre operator stumbling through awkward condolences and misplacing the paperwork. They want quiet, patient clarity. They want to know what to do next.

The customer just wanted the agony taken out of the administration.

What are your customers actually trying to get done? What do they want to feel while dealing with you? Which parts of your delivery create genuine relief, and which parts do you only assume they value because you have always done them that way?

Customer value is the difference your service makes to the person paying for it: the result they get, the effort it takes them and how confident they feel along the way. Not your price list, and not how clever your internal delivery looks on a spreadsheet.

If a client is stuck waiting for an answer, getting it in five seconds matters. But buying the same generative software as every competitor on your street won’t tell you which answer they need, or what they will do with it once they have it.

Step 1. Why do customers really choose you?

Ask recent buyers to explain why they chose you before you decide what needs changing. Your specialist expertise may have won the mandate. Or perhaps you were simply the only supplier who understood their mess and did what you promised on time. Those are completely different reasons to buy, and they demand completely different improvements.

A few years ago I visited Webmart for an interview. Before I set off, someone rang to ask what I drank. Tea, white, no sugar. When I arrived in Bicester, I found a parking bay with my name on it and a red carpet rolled out to the front door. I rang the bell and my tea was waiting.

I wasn’t even a customer. I told Alex M H Smith about that visit on the podcast because years later, I still remember the attention to detail.

I didn’t buy any printing that day. But the visit left an indelible impression before anyone pitched me a product. Whether your customers would value that kind of welcome is something to ask them, not an excuse to buy a red carpet.

Speak to five customers who bought from you recently. Five is a practical start, not a statistical survey. Ask them:

• What happened inside your business that triggered the search for help?
• What nearly stopped you from choosing us?
• Why did you pick us in the end?
• What did you compare us against?
• What was the moment you felt certain you had made the right call?

If they tell you you were professional, ask what specific behaviour created that impression. If they say they trusted you, ask for the exact moment that trust landed.

Write down their exact words. Look for patterns, but pay close attention to the disagreements. A nervous first-time buyer looking for reassurance values something completely different from an experienced procurement head who wants raw turnaround speed.

Step 2. Which customer problem should you solve better?

Pick one friction point where you can make an immediate, noticeable difference. It might be an onboarding delay, a confusing handover between sales and delivery, or radio silence while work happens. Make the change concrete enough that a client could tell you whether it happened without checking notes.

When Ivo Dimitrov described the AI-first accounting product he was building at Finom, one insight hit home. A good accountant can answer your question, but you first have to know what to ask.

Ivo wanted their product to raise issues the founder hadn’t thought to ask about, from missed tax credits to what to do next. When we spoke, it was in closed beta in Germany, so this was an ambition being tested rather than a finished revolution. Flagging an iceberg before the client spots it is irreplaceable customer value.

Do they have to chase your team for an update? Spot an emerging risk themselves? Work out which account manager to call?

If your customer interviews reveal that your first month feels chaotic, don’t build a new software platform. Test a simple, one-page roadmap showing what happens every week, who owns each milestone and what you need from them. Give them one direct phone number and commit to an exact weekly check-in time. If your clients already find your onboarding effortless, pick another headache to cure.

Step 3. Where can technology make room for that improvement?

You can automate part of the customer interface directly, or strip routine admin off your senior people so they have breathing space for high-value client work. Either way, decide in advance where that saved time goes.

Agentic AI means software that completes tasks on its own, not just drafts text or summarises a conversation. Gartner predicts it will resolve 80% of common customer service issues without human help by 2029.

Jotform is an online form builder with 35 million users. Aytekin Tank described how it introduced AI support alongside its human team. Simple, repetitive questions received immediate automated answers, freeing human agents for complex problems.

The total volume of support questions shot up. Because when customers expect an hour’s wait for an email reply, they don’t bother asking. They struggle in silence, find an ugly workaround, or churn. Once they could get an instant answer, they asked everything.

Aytekin also moved people off the support queue to audit the conversations between customers and the AI. That makes crude cost savings difficult to measure by ticket count alone. Factor that into your own calculations before promising your board an overnight headcount cut.

Instead of an account manager burning three hours copying status updates across four separate systems, an automated script can draft the summary. The account manager can spend twenty minutes reviewing it, spotting an operational snag, and picking up the phone to talk to the client.

Lock that saved time in the diary. If you don’t deliberately allocate the freed hours to customer relationships, your team will simply fill the gap with more internal email.

Step 4. How do you test whether customers experience something better?

Test your improvement on a tight cohort of ten clients before rolling it out company-wide. Agree in advance what success looks like and what failure looks like. Teams always fall in love with processes they have just invented; give your customers the permission and the space to tell you where it still hurts.

If you have redesigned onboarding, track how often new clients chase your team for updates. Ask whether they know who owns their project and what happens next Tuesday. Compare those numbers with your previous baseline, without pretending a small test settles every question for all time.

Pick one or two straightforward metrics that mirror the customer’s reality.

Alongside the quantitative checks, ask open questions: what felt easier, and what was still clunky?

You might find that your sleek new client portal went completely ignored, but having one direct WhatsApp contact made the client feel like royalty. Or that a faster automated deliverable created extra confusion because nobody explained how to interpret the results.

Step 5. How do you build on what you learn?

Turn a proven improvement into standard operating procedure. Assign clear ownership, bake it into training, and inspect the reality every month. If an experiment flops, kill it quickly before you burn more capital on an internal vanity project.

Aytekin Tank shared the real secret behind Jotform’s AI support gains. In the early weeks, their AI resolved only about 25% of customer queries. Over three months of rigorous iteration, resolution climbed to roughly 75%, with human staff handling the tough remainder.

Aytekin’s team reviewed the conversations between customers and the AI. They didn’t blame the AI model; they fixed the gaps in their own product and rewrote the user guides the AI was learning from.

Start with your own support inbox or the raw feedback from your initial test. Pick the friction point customers stumble over most often, implement a fix, and measure the difference. Involve the frontline people delivering the service in that weekly review. They know whether the fix is working, and they know what it takes to sustain it.

Once you have hard evidence that your service is noticeably better, use it boldly in your sales conversations. A binding promise of a dedicated point of contact or an agreed weekly progress call gives prospective buyers a tangible reason to choose you over a cheaper, looser competitor.

Start with one conversation

Book five customer conversations, pick one persistent friction point, assign someone to own the fix, and agree on how you will measure whether the client noticed.

If you are one of our coaching clients, bring those interview notes to our next quarterly session. We will build a concrete plan around them: what your buyers genuinely value, where your team is dropping the ball, and which customer improvement will unlock your next stage of scale.

Start by giving your next five customers an unforgettable reason to pay you.

Frequently asked questions

What if customers want different things?

Look for common operational situations before trying to blend conflicting feedback. A brand-new client usually craves reassurance and proactive contact, while a ten-year veteran simply wants rapid turnaround with zero fuss. Pick one distinct customer cohort and fix their specific problem first.

Never apply a blanket policy that solves a new client’s anxiety by drowning your established accounts in unwanted meetings. Segment your interviews by company size, deal value or how long they have worked with you, and read each group separately. You will usually find the loudest requests come from one segment while the quiet majority is perfectly content. Improve for the segment you most want more of, and tell the others plainly what you are doing and why. Most customers accept a service that is not built exclusively for them as long as you are straight about it.

Should I also talk to people who didn’t buy?

Yes, whenever you can get them on the phone. Lost prospects tell you why you were rejected or why they chose to do nothing at all. Keep those two conversations separate. Treat lost prospects as an opportunity to listen to real market objections, not as an excuse to reheat an old sales pitch.

Buyers will justify the money they spent, so their praise needs decoding before you act on it. Someone who walked away owes you nothing and will usually tell you the unvarnished truth: too slow, too expensive, too similar to the incumbent. Chase two or three lost prospects from the last six months, ask what they chose instead, and write down the exact words they use. Those words belong in your next positioning conversation, not in a drawer.

Should we lower our prices if technology makes delivery cheaper?

Not automatically. Your cost to deliver, your client’s perception of value, and their willingness to pay are three distinct things. If software cuts your production hours in half, you have just expanded your gross margin. Lowering your price simply gives away that margin without proving anyone will buy more.

Test whether clients want the same output cheaper, or whether they would gladly pay your existing fee for a more strategic, higher-touch outcome. If a tool saves your team ten hours a week, the temptation is to cut your fee and look generous. Resist it until you know where those ten hours actually went. If they went into client-facing work, your service just got better and the price should hold. If they vanished into internal admin, you have found your next improvement project, not a discount.

How do I stop a small test becoming another permanent project?

Set a hard stop date, assign a single owner, and define the decision criteria upfront. Before the pilot begins, write down the numbers that will cause you to roll it out, tweak it, or scrap it. If a three-week trial produces muddy results, refuse to let it drag on by default.

Kill it cleanly, capture the lessons, and move your resources onto the next experiment. Name the owner in writing, hand them the decision criteria, and have them present the result to you in fifteen minutes with a single recommendation. Pilots drag on when nobody owns the kill decision, so make that person explicit rather than collective. A firm that kills weak experiments quickly will run more experiments, and running more experiments is how you find the improvement your customers actually notice.

What should you do next?

---

When software makes delivery almost free, continuing to compete on capability alone is a fast route to margin collapse. The businesses that protect their pricing are the ones that take the time to discover what their customers actually struggle with, and build an experience around it that software cannot replicate.

Pick five customers. Ask them what they genuinely value and what they wish were simpler. Fix that one friction point, give an owner the authority to make it stick, and measure whether your clients notice the difference.

That is how you turn customer experience from a vague aspiration into an uncopyable competitive moat.

Four ways to take this further

1. Book a call. If your service is starting to feel commoditised and pricing pressure is creeping in, Dominic can help you diagnose whether the gap is positioning, customer insight, or team execution. No obligation, no pitch. You will know quickly whether this is the right kind of help.
2. Grab the book. Mind Your F**king Business gives founder-CEOs a practical way to stop being the bottleneck and build a company that scales with clarity, accountability, and strong gross margins.
3. Listen to Aytekin Tank on Scale to Win. Hear how Jotform scaled to 700 employees and 35 million users by using AI to solve real customer friction rather than chasing technology for its own sake.
4. Subscribe to the newsletter. Every Tuesday, Dominic shares practical advice on founder leadership, strategy execution, and breaking through the Founder Ceiling.

Your move. The technology is already commoditised. The experience you build around it is entirely up to you.