A Playbook for Improving Unemployment Insurance Delivery

Customer Service

Customer Service

First Contact Resolution

First contact resolution is when a claimant calling with a question or concern has that issue resolved on their first contact, rather than having to reach back out again.

First contact resolution rates are a standard measurement in private sector customer service, but we didn’t find any state that was tracking it for unemployment. In fact, some states shared with us that they were unable to even count call volume until the pandemic led them to adopt new call center technology like AWS Connect.

We recommend a demonstration project to measure first contact resolution.

Some examples of what this project could focus on include:

  • How to resolve one claimant across all channels (e.g., chat, email, and phone)
  • The ideal call reason categories for measuring contact reason with enough granularity to work through and automate or eliminate the top reasons
  • What the top reasons for requiring multiple contacts are
  • What the ideal mix of channels is (e.g., email, chat, phone, IVR, SMS)
  • The necessary staffing ratio across channels (e.g., chat vs. phone)
  • What first contact resolution rate is achievable

Contact Methods

States need multiple customer service channels to properly service unemployment claimants. They also need to be able to reconcile the same individual across all of the contact channels.

Staff contact centers appropriately.

If the only request an agent can answer is a password reset, a phone center with one million live agents who answer in under 30 seconds is completely useless.1 Many states rushed to hire hundreds or thousands of call center agents, which ultimately just further frustrated claimants at significant cost to the agency.

Identify common questions that could be handled through a different channel.

It’s important to measure why people are calling in and ensure those reasons can be addressed via specific channels. Make it clear to claimants what they can and can’t accomplish through a given channel, and expand self-service offerings so that live agents can focus on individuals who can’t access or use self-service.

“At the beginning of the pandemic, the PUA Call Center was outsourced. Originally, DETR contracted with Alorica to run an informational call center with 100% contract staff. Many claimants complained that the call centers could only pick up a call and say “Yes, I see your claim in the system and there appear to be no issues; I will elevate your claim to an adjudicator.” Nevada Strike Team

Story from the field

North Carolina provides claim status updates through an automated phone system. If the claim status is that the claimant needs to take additional action, that claimant is immediately routed to a live agent, skipping wait time.

Develop a contact center taxonomy.

A demonstration project could be developing an optimal contact center taxonomy and routing tree methodology for capturing granular call reasons and maximizing self-service.

Chat bots can help reduce call center volume.

Many states introduced chatbots during the pandemic to help reduce call center volume for basic questions, provide claimants with claim status updates, and other tasks.2 Chatbots have proven useful in other benefits areas like WIC,3 but only if the self-service answers are effective.

“One thing that many states have been implementing that has helped to address long backlogs in a claimant-friendly manner is to set up callback systems and establish online chat technology to answer basic questions and help people avoid common mistakes. States can also establish triage protocols as a part of their business practices so they can better allocate resources. That way, calls coming in about password resets or claim status can be directed to staff specialized to handle simpler questions, freeing up adjudicative staff time.”4— National Employment Law Project

Stories from the field


Texas launched a chatbot named Larry. “From conception to deployment _‘“Larry_’” was up and running in four days with help from their private sector partners AWS and Accenture and has answered 4.8 million questions for 1.2 million people.”5

North Carolina

North Carolina added a webchat by analyzing the most commonly-searched terms on the website to add new content. They measure the success of their chatbot relative to their call center volume, adding new chatbot topics regularly to help deflect the need for live agents. Since deploying chat on AWS Connect, more than 100,000 people have successfully self-served.

Allowing people to schedule calls can reduce call volume.

Many states reported that allowing individuals to schedule appointment times with a call center agent drastically reduced the number of call trunk lines in use, and dramatically increased claimant satisfaction because they didn’t have to wait on hold or redial all day.6 Wisconsin found that the greater the call hold times, the greater the number of redials — a trunk line death spiral.

Aggregate claimants across channels to reduce workload.

Measuring the number of pending customer service requests by unique claimant can incentivize states to resolve identities across platforms. Without resolution, the same person can count tens if not hundreds of times in the customer service “backlog.” Further, once you successfully resolve that claimant’s issues, you can close out all of their related messages across all channels, reducing overall workload.

Story from the field

Tennessee reports success using ZenDesk to identify the same claimant across all channels, which also provides the claimant with superior support (since a customer support professional can see all of their historical requests and claim updates).

Ask claimants for their preferred contact method.

Some states continue to have policies where claimants must opt IN to electronic mail. Given today’s usage patterns, this should flip: states should ask constituents up front to state their preferred contact method, and follow it. While some may select physical mail, it’s overall the slowest method, and should never be the default.

Offer several ways for claimants to contact your program.

Provide an expansive number of contact channels, like:

  • Automated phone system (IVR)
  • Live agent phone system
  • Text messaging
  • Chatbot
  • Electronic messaging
  • Lobby management technology for in-person offices

Story from the field

In Connecticut, the Consumer Contact Center expanded CTDOL’s staff and modernized its platforms, enabling virtual and live chat features, a call scheduling option, as well as phone, text, and email communications. Since July, the Contact Center has handled nearly 700,000 cases and currently handles up to 3,000 calls per day.

Help Documentation

Improving help documentation can have a huge positive impact on state unemployment systems. Poor quality help documentation was cited repeatedly7 as a barrier and frustration for claimants.

High-quality help documentation offers many benefits, like:

  • Helping claimants apply and recertify faster
  • Reducing load on customer service
  • Reducing overall errors

To be most helpful, documentation should meet these criteria.

High-quality help documentation is:

  • Written in plain language
  • Easily discovered via search — including both external search engines like Google, and also within an employment website’s search feature
  • Contextual — the explanation for a potentially confusing question should be right there next to the confusing question
  • Data-driven — write content to address top questions8 from customer service and website searches

The format of documentation can vary.

Help documentation can take multiple forms: contextual help bubbles, user guide videos, frequently asked questions, how-to articles, automated chatbot answers, etc.

Ad hoc support groups9 where claimants helped one another navigate the unemployment process sprung up in nearly every state. Some states partnered with these groups, while others didn’t.

We recommend that every state develop relationships with its unemployment navigator communities.10

Whether they’re a Facebook group or a local Legal Aid office, these navigators are a crucial source of information about obstacles that states can resolve, as well as a powerful form of customer support that can reach a wide population.

Stories from the field


California hired an Unemployment Stakeholder Advocate to represent the needs of claimants in agency operations. Read the position description.


Through the pandemic, Michigan UIA deepened its relationship with community partner agencies, such as large nonprofits and workforce development centers, by connecting navigators directly to UIA caseworkers. Navigators shared that this connection was a very useful tool to help claimants get their questions answered more quickly than if they’d reached out to UIA on their own.

Community partner agencies expressed deep respect for the staff at UIA who navigated a tumultuous year with outdated technology. These relationships are an example of an effective form of communication from UIA to its users and provides a model that might be used more broadly.11


In Nevada:

“The Chair of the Strike Force was in frequent communication with a founding member of the Nevada Pandemic Unemployment Facebook group. This individual shared valuable information with the Chair who passed these insights from a claimant feedback loop to ESD. This type of communication should occur with DETR and community stakeholders just as other state agencies do with their stakeholders. Perhaps that was not possible when the crisis was at its highest peak as there was no one available to do it. However, this is an essential role that should be created in the future.” Nevada

Go to the next section: Procurement



Supported by

The Families & Workers Fund