The Deflection Illusion: A Four-Week Audit of Calls, Chats and Closed Sales

Most UK SMEs buy live chat with a simple hope: fewer phone calls. It sounds sensible. If buyers can ask a quick question in the widget, surely the phone stops ringing? In practice, chat often shifts the conversation rather than removing it. Sometimes it creates a second queue. The only way to know which is happening is to test the phone number, not the chat feature list.

Why fewer calls is not a result on its own

A drop in call volume can mean good deflection. It can also mean your phone line is broken, your opening hours changed, or buyers gave up. Meanwhile, a flat call volume does not prove chat is useless. It might be catching new enquiries that would never have called. Judging chat by call count alone is like judging a salesperson by how often they use the phone. You need outcomes.

Set the baseline before you change anything

Run the audit over four clear weeks. Week one is baseline. Pull your call logs from the phone system and, if possible, the CRM. Group calls into rough reasons: order status, delivery, product question, pricing, returns, account issue, other. Do the same for chat transcripts. If you have no reason codes, read a sample of 30 calls and 30 chats. You are not building a data warehouse; you are looking for patterns.

Also record sales outcomes. For each enquiry, note whether it became a quote, order, booking or nothing. Use a simple reference such as email address, order number or CRM ID. This is the part most teams skip, and it is the part that tells you whether chat is adding value or just adding noise.

Weeks two and three: tag every conversation

Now tag chats and calls by reason and outcome. Ask advisors to choose a reason from a short list at the end of each chat. Keep it to five or six options. Add one field for outcome: unresolved, resolved, quote sent, sale, wrong department. For calls, use the same categories. If your phone system supports call tracking, add a second number to the website? Not necessary for this test, and it can confuse buyers. Better to keep the number stable and compare like with like.

At the end of each week, count three things: total calls, total chats, and total sales from both. Then count duplicates. A duplicate is the same buyer asking the same question in both channels within 48 hours. You will spot them in transcripts and call notes. It happens more than teams expect, especially when chat is staffed only part of the day.

Week four: run the numbers

Calculate four figures. First, call volume by reason. Second, chat volume by reason. Third, duplicate rate: duplicate enquiries divided by total enquiries. Fourth, conversion by channel: sales divided by enquiries. If you want a fifth, track first response time and handle time, but do not let those distract from the main question.

Here is the benchmark that matters. Genuine deflection looks like calls falling in the exact categories your chat handles, while chat volume rises in those same categories. The duplicate rate should stay under roughly 10 to 15 per cent. Conversion should hold steady or improve. Total contacts per sale should not rise. If calls are flat, chat is rising and duplicates are climbing, you do not have deflection. You have another inbox.

The awkward middle ground

Many SMEs land in the middle. Calls fall a little, chat rises a lot, and conversion is murky. That usually means the widget is catching pre-sale questions that never became calls, which is good, but your team is also answering the same delivery query twice because your help pages are weak. Fix the content before you blame the channel. A clear delivery page and an order-tracking link in the chat window can remove a chunk of repeat contacts without any new software.

Watch for deflection that is too successful. If calls drop sharply but chat conversion is poor, you may be pushing high-intent buyers into a slower channel. A buyer who wants to talk about a £10,000 installation may not want to type. Give them a clear route to a call, not a hidden one. Deflection should mean the right conversations move to the right place, not that every conversation becomes a chat.

What to do with the result

If chat is deflecting well, keep staffing it and stop apologising for the phone. If it is duplicating, tighten your routing. Add an out-of-hours message that sets expectations. Put common answers in the widget before the first question. Train advisors to close the loop: if a chat ends unresolved, offer a call back rather than asking the buyer to ring in. If the test shows chat is generating extra sales that never would have called, count that as a win, even if the phone still rings.

The phone number test is not about proving a point. It is about seeing the whole sales desk. Calls and chats are not rivals. They are two doors into the same room. Measure who comes through each door, what they ask, and whether they buy. Then you can decide whether your widget is deflecting calls or simply adding another queue with a nicer font.