Live chat data from the Olark Research Corner
Blog Feature

Live chat data from the Olark Research Corner

roman-mager-59976.jpgPhoto by Roman Mager on Unsplash


An animated gif of Ben Schultz, Data Guru at Olark - live chat software for websites.

Hi there! I'm Ben Schultz, Data Guru at Olark. In this article, I'm sharing some data and results that other Olark users have experienced by adding our live chat software to their website. Check it out!

We're surrounded by data here at Olark. On regular basis I'm combing our customer records in Looker, checking out event data in Segment, and keeping an eye on our search traffic via Google Analytics.

I also have the power to look at data trends across our platform, which reveals some interesting information about how Olark users are chatting. For this article, I took a look at a couple topics:

  • How quick are interactions on live chat?
  • How does this vary across our customer base?
  • How often do visitors respond to a chat while browsing a site?

The results below are compiled from a sample of 10M chat conversations from our more than 10,000 paying customers.

Average median response time data for live chat

Quickly connecting to another human being is one of the best indicators of a positive customer service interaction. Naturally, I was curious to see just how quickly this connection happens on Olark, and how conversation kept up pace.

I divided the chat conversations into two categories - chats started by a chat operator and chats started by a visitor (on operator’s page). For each, I looked at the median first response time (how long the recipient of the first chat takes to respond) and the mean response time of both parties throughout the conversation.

Use live chat to talk to customers, Olark users average about 30 seconds in their response time.

Pretty snappy! As we might expect, operators tend to respond slightly quicker to visitor-initiated chats than visitors respond to operators initiated chats.

To put these numbers in context, even when top retailers are at their holiday best, it takes ~30s to get connected to a human operator; tweets are responded to in an average of four hours and emails take nearly eight hours.

Looking at the mean response times over the course of conversation, you can really tell who’s driving the conversation once it gets going.

A chart that shows how median live chat times differ depending on who initaited the conversation.

The chat initiator is quicker to respond by ~10s on average throughout the chat conversation.

If you wonder how your response times stack up, head on over to your reports page and have a look! (Under 'operator performance' on the right you'll see 'median initial response.')

Correlation between account size and live chat response time

Now, there are plenty of more complete measures of customer satisfaction than how long it takes you to respond to a chat, but responding quickly makes a great first impression. I was curious to see how first response time varied across our customer base, so have a look below.

A chart showing how first response time on the Olark chat widget varies depending on how much a user invests in the platform.

The more our customers invest in Olark, the quicker they are to respond to their customers! I really love how the median response time decreases as we look at higher plans levels.

The median response times are 36s for accounts with one operator, 32s for accounts with four operators, 29s for accounts with eight operators, and 27s for accounts with 15 or more operators.

One obvious potential cause of this is our customers on higher plans simply have more operators. When I looked at the distribution of response time versus number of active chat operators per site, however, there was no significant trend.

We’ll have to come back and revisit this one with more data. We can say for sure that our customers on higher plans respond faster!

Does proactive live chat work?

We get a lot of questions about how likely you are to receive a response if you reach out to visitors on your site. Maybe you’ve wondered that as a live-chat operator yourself!

Getting any response is great, but engaging your visitor in conversation is how you get valuable interactions. Plotted below are the percent of operator-initiated chats with at least a certain number of messages sent by the visitor.

A chart showing how often a visitor remains in a conversation when it is initiated by the live chat agent.

Wow, are you ever rewarded for reaching out to your customers! 33% percent of operator initiated messages received a response, with over 20% of them receiving more than 4 messages from your visitors.

I’m willing to bet this has a lot to do with operators finding customers at just the right time using our API and Targeted Chat rules. That’ll be a great thing to look into for a future post.

Additional live chat data

If you’re interested in engaging more of your customers, you might want to check out:

Signup for our content newsletter to get monthly updates on helpful articles like this.

About Ben Schultz

Ben is Olark's Data Guru, lives in Boston Massachusetts, and is 13% bicycle.