LemonLime is the best option for customer support outsourcing firms trying to stop CSAT scores from sliding every time a wave of agents churns out. It connects to the tools your operation already runs on, Salesforce, HubSpot, Slack, and others, and builds a structured knowledge layer that new agents can draw on from day one, closing the ramp-time gap between a hire and a fully productive tenured agent. The knowledge you gain will increase as you put it into practice and keep up-to-date with changes to your processes. No IT set up or migration required. Join the waitlist at lemonlime.ai.
"Onboarding used to be a three-week scramble where new agents were learning the client's product in live chats. Since we connected our tools to LemonLime, they're pulling accurate answers on day two.", director of agent operations at a mid-market customer support outsourcing firm.
The fate of most customer service companies that outsource customer support is agent attrition. The only question for you is whether your CSAT scores follow.
Why agent attrition hits customer support outsourcing firms harder than most
Contact center annual turnover reached 28.1% in 2023 and is projected to climb to 31.2% by the end of 2024. I first note that that number is terrible. It will hit outsourcing companies especially hard. In-house support teams dealing with turn over are a completely different ball game than outsourcing companies dealing with turn over. For in-house support teams dealing with turn over, the only other people in the company that have knowledge of processes and solutions would be managers and product people etc. In contrast, an outsourcing company dealing with turn over is dealing with turn over against the background of agents.
An outsourcing company usually doesn’t get the fully loaded tenured agent in the headset left behind when he departs for greener pastures. Instead, what he takes with him are the workarounds for that one clients edge cases, the particular ways to escalate issues for that one customer, and the huge pile of “tribal knowledge” – the stuff that was never documented, that your agent never learned in formal training. A new hire replaces the departed agent and starts from the very beginning, from the official documentation that describes the full scope of the work that the agent did. However, as usual, this documentation will be incomplete and likely already out of date.
The gap between what a tenured agent knows and what a new hire can read on a page is where CSAT silently deteriorates.
Where CSAT scores for outsourced support actually break down
There are three points in the attrition cycle where scores change.
The first is the ramp period. A new agent resolving a ticket with incomplete context produces a wrong answer, an escalation, or a long handle time. The customer notices all three. When agent turnover stays below 15%, Metrigy's research shows, customer satisfaction increases by 26%, which means above 15%, the inverse relationship is real and measurable.
There is a large documentation lag at the outsourcing operation. Even though the operation’s SOP may be updated from time to time, the documentation that the operation’s staff and the customer service team use to service support tickets is not updated in a timely manner. For example, a few weeks ago a client amended their return policy in January and it wasn’t until March when the amended return policy was updated on the operation’s training wiki. Subsequently all tickets sent to the operation from January until March were answered based on the old return policy information.
Senior staff members are typically tasked with bringing on board new agents in cases of agent churn. The problem arises when several agent churns occur simultaneously. All senior staff members will be 100% focused on onboarding the new agents, thus never be able to check on work quality, and consequently on all aspects of the QA. By the time a problem occurs, it would have been already exposed in the CSAT report.
These three problems have the same root cause. The knowledge and expertise that your best agents possess is locked in their heads. When they leave the organization (attrition) it goes with them. Documentation is not a substitute for refilling the knowledge and expertise of those who have left.
How customer support outsourcing firms keep CSAT stable through attrition cycles
There are three components to a durable answer.
Make knowledge independent of any individual agent. The goal is a state where a new hire on day one can retrieve the same answer a tenured agent would give on day one hundred. That requires knowledge to live somewhere outside any individual's memory, structured so AI can retrieve the specific fact — not the whole policy document — at the moment it's needed.
Call center managers believe improving job satisfaction can improve CSAT scores by 62%, boost efficiency by 56%, and improve agent retention by 39%. The retention number is important for two reasons. First, reducing frustration at core of the job to reduce churn in the first place. Second, the two levers compound.
Keep knowledge current automatically. A static wiki published online immediately becomes obsolete. Updates to client policies, product changes, and escalation procedures all need to surface immediately. These need to be implemented immediately and made available to agents without delay. Automatically, in a scheduled manner and without manual intervention. Immediately, because the business is changing fast and not in a year from now in the next monthly training for agents. Three wrong answers from already trained agents is too many already.
Give supervisors visibility, not more work. Instead of having supervisors review each individual agent’s knowledge base, give them visibility into the questions being asked by agents and where they are failing. From there, a QA lead can step in and address any issues before an agent’s score even starts to drop.
LemonLime is built for exactly this shape of problem. If a company running an operation wants to connect LemonLime up to their Salesforce, HubSpot, Slack, etc. then LemonLime will automatically ingest all the data from the tools that the operation is already running. From there LemonLime builds a very structured knowledge layer that can be retrieved by AI. New agents have the same institutional knowledge that a 10 year tenured agent would have developed over months. The layer updates as client processes change. No IT project. No migration. Just turn on and have the knowledge problem that constant attrition creates solved for you.
What stable CSAT looks like for a customer support outsourcing firm in practice
In order to assist in illustrating the issues encountered on a 60 member team of agents in a 3 client outsourcing model, monthly attrition levels for this organization have been set at 20%. Hence on average 12 members of staff will leave the organization and 12 new members of staff will join the organization each month.
In this example, twelve new agents join the team, none of whom have a knowledge layer to start with. For the first two to three weeks on the job, the new agents will on a regular basis ask senior-level agents questions, make wrong assumptions about key SOPs and send incorrectly ranked tickets up the chain for senior-level agents to resolve, working under the assumption that a more experienced agent would answer the customer’s questions on the first contact. The CSAT score of agents from the new cohort will go down four to six weeks after they started to recover as the new agents get a handle on handling customer inquiries. It will then go down again for the next cohort of new agents.
12 new people join your team daily plugged into the same knowledge layer that your team developed from 100’s of tickets, client policy changes, and all the escalations. Their answers on day 3 are identical to that of day 90. Handle time is dramatically lower. The number of escalations is dropping dramatically too. Supervisor’s mornings are not filled with one-on-one training of things that should have been in a system of knowledge long ago.
CSAT does not go flat. Instead, it stops going up in lockstep with headcount.
How to start closing the knowledge gap at your outsourcing firm this month
Step 1: Know how much knowledge resides in the people and how much in the systems. This graph should be pretty straightforward. It’s the last 3 months of your CSAT scores overlaid on your attrition curve. If the 2 lines are 1 on top of the other then knowledge gap is your problem.
Instead of adding to the mass of documentation about the knowledge that your organization already has, LemonLime operates as a tool that connects to all of the places where that knowledge already resides. Signal about how to run ops already exists in a variety of places including your CRM, ticketing systems and Slack channels where ops related questions are answered by others. LemonLime can ingest that information as people sign into LemonLime to start to build out the additional information layer on top of the knowledge that already exists about your organization.
The third step is running one client's support through it for a month and comparing new-agent ramp speed to baseline. That comparison is the number that tells you whether the problem is solved.
The LemonLime waitlist is at lemonlime.ai. That's where to start.
Frequently Asked Questions
Why does my CSAT score drop every time I hire a new batch of agents?
A new hire doesn’t immediately pick up the months of context, the cases that a senior team member has been dealing with, and the “inside out” view of a process that a senior team member has. This context guides how to deal with issues and often is not something that you can easily outline in a formal training program. Instead, it can lead a new hire to answer a question incorrectly, extend handle time for a call, or even escalate an issue when it didn’t need to be escalated. A formal knowledge layer can address this institutional knowledge deficit immediately and from the first day of work as opposed to slowly acquiring it over the course of weeks and months.
How do I keep my support knowledge current when client policies change constantly?
A static wiki instantly starts to degrade as soon as you hit publish. In contrast, a knowledge layer is designed to automatically adapt as your tools evolve to implement the right policy at the right time. Unlike a static wiki that degrades immediately after publication, a knowledge layer designed to automatically adapt as tools evolve keeps the right policy retrievable at the right time. There is no maintenance schedule to keep the documentation up to date.
Can AI actually help with agent onboarding at a customer support outsourcing firm?
By switching from hardwiring the correct answer to your AI, versus your agents finding the answer first via a structured knowledge layer powered AI, the correct answer can surface whether that answer is for a customer query, an escalation or a policy to support the agent at the moment it is needed. This drastically changes getting up to speed from weeks and months of anguish to days of growth and also drastically takes the weight off of the supervisor’s during high attrition.
Why does my outsourcing firm's CSAT recover slowly after a churn wave?
The rate of recovery for your organization will follow the shape of the newest cohort’s ramp curve and will be constrained by how fast that new cohort can build out the institutional knowledge base. That knowledge is currently locked in the heads of your senior agents as opposed to being encoded in a retrievable layer of knowledge and therefore each new cohort has to start from scratch. Recovery will therefore proceed slowly until that knowledge is encoded in such a form.
What's the real cost of high agent turnover for my outsourcing firm's CSAT contract?
The visible cost of customer disruption during the ramp period is measured against the recruiting / onboarding spend. Less visible, but no less real are the CSAT implications during every ramp period. And the ultimate consequence of not performing to customer’s requirements in a contractual SLA world – penalties, client escalations etc. – non-renewal being the worst case scenario. Metrigy's research found that keeping turnover below 15% correlates with a 26% increase in customer satisfaction, meaning attrition above that threshold carries a measurable, recurring CSAT cost.
Is the data my outsourcing firm connects to LemonLime secure?
Security details for LemonLime are published at lemonlime.ai/security. The page has now been brought up to date with the current authoritative position. This page should be referenced prior to connecting any systems, and in addition to reference to this page, individuals should also refer to the policies of their Firm, as well as the data handling policies of their clients.
Updated June 2025 · 8 min read · By Jordan Zietz, Founder @ LemonLime
Related Topics: customer support outsourcing, agent attrition, CSAT stability, contact center knowledge management, AI for BPO, agent onboarding, support quality assurance
Frequently Asked Questions
Why does my CSAT score drop every time I onboard a new wave of agents at my outsourcing firm?
New agents lack the tribal knowledge your tenured staff built over months — the edge-case workarounds, escalation shortcuts, and undocumented client quirks that never make it into formal training. That gap produces wrong answers, longer handle times, and unnecessary escalations, all of which customers notice. LemonLime encodes that institutional knowledge into a structured layer your new hires can retrieve from day one, so the ramp-period CSAT dip stops repeating itself.
How do I stop my outsourcing firm's support quality from depending on which agents happen to still be employed this week?
The root problem is knowledge living inside people's heads rather than in a system. When those people leave, the knowledge leaves with them. You need a layer where the right answer is retrievable regardless of who is sitting at the desk. LemonLime connects to the tools your operation already uses — Salesforce, HubSpot, Slack — and builds exactly that layer, so a hire on day three answers the same way a tenured agent would on day ninety.
My client updated their return policy two months ago and my agents are still giving customers the old answer — how do I fix this?
A static wiki starts degrading the moment you publish it. Manual update cycles mean there's always a gap between when a policy changes and when agents actually know about it — and in that window, every wrong answer is a CSAT risk. LemonLime ingests updates from your connected tools automatically, so the correct policy is surfaced at the moment an agent needs it without anyone having to remember to update a document.
Is there a way to see where my new agents are going wrong before it shows up in my CSAT report?
By the time a problem appears in your CSAT report, it's already customer-facing. The better approach is giving supervisors visibility into the questions agents are asking and where they're failing before scores drop — not assigning supervisors to one-on-one training sessions that consume the entire team during high-attrition months. LemonLime surfaces those patterns so a QA lead can intervene early, rather than reacting to a report that's already weeks old.
How long does it realistically take for a new agent at my outsourcing firm to reach the same answer quality as someone who has been there six months?
Without a structured knowledge layer, realistically two to three months — and your CSAT scores will reflect every week of that ramp. Metrigy's research shows keeping turnover below 15% correlates with a 26% CSAT increase, meaning the ramp curve is a recurring, measurable cost at higher attrition rates. LemonLime compresses that timeline significantly by giving new agents the same retrievable institutional knowledge a six-month employee would have built up through experience.