Improve your self-service portal, reduce calls

Most large enterprises would prefer customers to solve routine tasks — paying a bill, changing a subscription, checking a balance — through a self-service portal. It is faster for the customer and far cheaper than a phone call. Every call handled by an agent carries a cost in money and time.

The problem is that when the portal is confusing or incomplete, customers simply call instead. And most companies struggle to work out why. They know call volumes are high; they do not know which portal shortcomings are driving them.

To fix a self-service portal you have to know the exact reasons customers pick up the phone. That information is sitting in your calls — you just need a way to read it.

Why you don't know why customers call

Today, most companies understand call reasons through agent notes and category tags. When a customer calls, the agent selects a predefined category and writes a short note.

That data is rarely good enough for a product owner to act on. Categories are broad, notes are inconsistent, and neither captures the specific friction that sent the customer to the phone. The result: leaders have no clear picture of why customers avoid the portal, or what would change that.

How to discover what frustrates customers

Speech-to-text turns every call into searchable text. Once transcripts exist, you can analyse them by topic and see, in the customers' own words, why they preferred to call rather than self-serve.

If a product owner could fix a specific issue on the portal, that issue shows up as a cluster of calls in the analysis. You are no longer guessing which change matters most — you can see the volume attached to each one, and prioritise accordingly. This is the same principle behind efficiency with AI: analyse every conversation, find the biggest driver, fix the root cause.

Small fixes, real reductions in call volume

Clients in banking, telecom and insurance use this approach to work through their call transcripts and find concrete, fixable issues in their portals. The changes are often small; the effect is not.

• A menu button was badly placed on one client's portal, so customers could not find it and called instead. Repositioning it reduced overall call volume by around 2%.
• Payment invoices were unclear, driving a large volume of billing calls. Minor wording changes on the invoices led to up to 15% fewer billing calls.
• Loyalty membership information was missing from the portal. Adding it cut related call volume by around 5%.

It is tempting to dismiss these as too small to be worth acting on. The examples show the opposite: minor changes, informed by real conversation data, produce meaningful reductions in call volume — which flow straight to cost.

The payoff beyond cost

A clearer, more usable portal means fewer calls, lower costs and, importantly, happier customers who got what they needed without waiting on hold. It also frees agents. With routine questions deflected to self-service, agents have more time for the customers who genuinely need a human — and more room to focus on retention and upselling.

Frequently asked questions

Why do customers call instead of using self-service?

Usually because the portal is unclear, incomplete or hard to navigate for the task they came to do. Analysing your call transcripts reveals the specific friction points in the customers' own words.

How does conversation analysis reduce call volume?

It groups calls by topic so you can see exactly which portal shortcomings drive contacts. Fixing the highest-volume issues removes the reason customers were calling, deflecting those contacts to self-service.

Are the fixes usually big projects?

Often not. Many high-impact changes are small — repositioning a button, rewording an invoice, adding missing information — but they only become obvious once you can see the call volume attached to each issue.

What data does this rely on?

Your own recorded calls, transcribed with speech-to-text and analysed by topic. It uses every call, not a small manual sample.

Where to go next

• The bigger picture: Efficiency with AI in customer service
• Deflect more contacts: Knowledge base agent
• How transcription works: Speech-to-text
• See it on your data: Request a demo

Want to know why your customers still call instead of using self-service? We will analyse it on your own conversations. Book a demo.