[Un]Churned · 14 Jan 2026 · From the week of 12 January
18 Months, 7,000 Customers, 67% Support Resolution: Inside Intercom's FDE Strategy ft. Diego Ballona (Intercom)
These are notes on the conversation, checked against its transcript. The episode itself has the full discussion.
In brief
Diego Ballona, Senior Director of Engineering at Intercom, explains how the company's forward deployed engineering (FDE) team, made up of product managers, engineers and data scientists, works with Fin customers. He describes how the motion began about 18 months ago with five design partners, how the team is structured and divides work with customer success, and how outcome-based pricing aligns incentives. His central argument is that FDE accelerates time to value early in the customer journey and feeds product improvements, but that it may not suit every business and should not become an ongoing managed service.
For founders
- Start an FDE-style motion with a small set of design partners across segments and regions to test whether deeper engagement improves outcomes.
- Treat the AI agent deployment as a product, with a product manager accountable for several accounts.
- Use FDE early in the customer journey to accelerate time to value, with the aim that customers come to self-manage their agents.
- Engineers who sit with customers feel the pain first hand, which Diego said speeds up their iteration and ownership of the problem.
- Prototype a feature for one customer and hand it to a core product team to productize once demand is validated.
For revenue leaders
- Tie compensation, performance reviews and go-to-market incentives to resolutions and customer outcomes when pricing is outcome-based, as Intercom does for Fin.
- Keep CSMs accountable for the long-run success of the account and give FDE engagements a bounded mission.
- When a customer has no dedicated counterpart, set expectations up front.
- Diego said it is very easy to achieve great containment and deflection, but sustainable value is not easy.
- Diego said FDEs managing agents for customers on an ongoing basis does not work financially.
What was said 23, most useful first
The FDE test began with five design partners spanning customer segments and regions. Listen
Intercom chose five customers that mixed SMB, mid-market and enterprise and were located in different regions, to test whether deeper partnership drove better outcomes. Diego said there was customer value, but the team quickly learned product changes that would make Fin more self-serviceable and self-manageable within the first month.
“we chose 5 customers. They were a mix of all these segments that we spoke of”
Listen to the episode Strategy & market Link to this Report a problem
Intercom's FDE team used an LLM as a judge of conversation quality, and that tooling became a feature in the product. Listen
Diego said the team built sophisticated tooling for judging which conversations are high or low quality. That work led to the CX score feature, an AI CSAT that scores 100 percent of conversations, whereas survey-based CSAT could only score a small portion of contacts. He said it started as manual analysis by the FDE team.
“how to use LLMs as a judge for which conversations have quite high quality or low quality.”
A Slack channel support feature began as a rapid prototype for one customer and was productized within about three months. Listen
A customer had a large share of its support volume on Slack, which Fin did not support at the time. Engineers sat with the customer's head of support, and an engineer estimated a version could be built in two to three weeks. Intercom built it for that customer despite it not being on the roadmap, and because demand was validated, it was productized into a general availability product in about three months.
“we actually decided to invest in that because we validated demand.”
FDE fits the early customer journey, and that FDEs managing agents for customers on an ongoing basis does not work financially. Listen
He said FDE accelerates time to value and return on investment early on, while the real value of Fin lies in customers self-managing, building their own procedures and driving insights without depending on the vendor. He said models where FDEs manage agents on an ongoing basis have severe drawbacks and do not work financially.
“And they get involved in models where you have FDEs managing agents for the customer on an ongoing basis.”
Listen to the episode Strategy & market Link to this Report a problem
Outcome-based pricing aligns go-to-market and FDE teams around resolutions and customer outcomes, Diego said. Listen
Diego said Intercom charges for outcomes rather than conversations or seats. He said comp plans, performance reviews and incentives for go-to-market and FDE teams are tied to resolutions and customer outcomes, and that Fin's ARR is related to how much work is done for customers.
“from comp plans to performance reviews to all the incentives that we built around all of our like go to market teams”
Listen to the episode Pricing & packaging Link to this Report a problem
Fin customers average a 67% resolution rate across nearly 7,000 customers. Listen
Diego said Intercom has nearly 7,000 customers using Fin, and that the average outcome across all segments is a 67 percent resolution rate.
“the average outcome that our customers get it is 67% resolution rate.”
An FDE team needs three roles: a product manager, engineers and a data scientist. Listen
Diego said Intercom went through different ratios and compositions before settling on these three skill sets. The team works in a book of accounts, owns customer outcomes, and works closely with customer success, sales and the rest of go-to-market.
“we got to, to the realization that you need 3 roles and sets of skills.”
Listen to the episode Hiring & team building Link to this Report a problem
The FDE product manager is accountable for multiple accounts and treats the AI agent's deployment as a product. Listen
Diego said the PM works with an AE or ACSM to tease apart requirements, the problems to be solved and the outcomes the customer wants to drive. The PM treats the deployment and evaluation of the AI agent as a product.
“That person basically is accountable for multiple accounts.”
A large part of FDE engineers' work is configuring the agent rather than shipping custom features. Listen
Diego said a large part of the engineers' work is configuration, including fine-tuning prompts, running experiments, iterating on the agent's performance and giving customers insights so they can self-manage their agent. He said this is not shipping custom features.
“It's not like shipping custom features. It's like fine tuning prompts, running experiments.”
FDE engineers need product engineering skill plus go-to-market skills like engaging customers and communicating value. Listen
He said the bar for FDE engineers is the same as for product engineering, but they also need sales and go-to-market skills. Diego said engineers often fall in love with a solution rather than a problem, so being able to stay focused on the customer's problem is an important skill for FDEs.
“we sometimes engineers, we fall in love with solution, not with a problem.”
Listen to the episode Hiring & team building Link to this Report a problem
Data scientists in the FDE team own measurement, iteration and the direction of optimization for each customer's outcome. Listen
Given a sample of a customer's conversations, the data science team can identify the largest opportunities and where the most time is spent in customer service. Diego said they cover the quantitative metrics as well as the quality side, including how quality is measured.
“they own the measurements, they own the iteration, they own the direction of like optimization as well.”
Listen to the episode Metrics & finance Link to this Report a problem
Intercom included data scientists because customers care about outcomes beyond the vendor's own metric. Listen
Diego said no customer comes to Intercom only wanting to move the metric Intercom cares about, which is resolution rate. He said an analytical mind working with the customer helps the team understand what the customer cares about, and that he had not seen many companies do this.
“No customer ever comes to us and they just want to move the metrics that we care about, right, which is resolution rate.”
Listen to the episode Hiring & team building Link to this Report a problem
When a customer has no dedicated counterpart on their side, set expectations up front. Listen
Diego said Intercom learned over almost two years that when a customer does not have a dedicated person to work with, the team needs to set expectations at the start of the engagement.
“if they don't, you need to set expectations up front.”
Listen to the episode Retention & customer success Link to this Report a problem
Engineers sitting with customers feel the pain first hand, which Diego says speeds up iteration and ownership. Listen
Diego said this is different from aggregated research or metrics, which he called very valuable but slower to act on. He said the FDE motion has led Intercom to build a better product for all customers and better internal tooling for its go-to-market team.
“when you get engineers sitting with customers directly, they feel the pain first hand.”
Diego describes a split in which FDE builds early prototypes for specific customers and a core product team later productizes them. Listen
He said FDE works on early prototypes with specific customers. Once a product becomes a big part of the platform's infrastructure and needs more investment, a core product team takes ownership and productizes it for all customers.
“But at some point, it makes sense for a core product team to own the product because it becomes a big part of like our infrastructure of our platform.”
CSMs own the long-run success of the account, while FDE engagements are bounded and time-limited with a specific mission. Listen
Diego said the account is owned by the success team and the go-to-market team, while FDE comes in with a bounded engagement. He said FDE can shortcut product roadmap priorities and technical gaps, which would otherwise take a CSM and solutions architect multiple months of effort.
“the CSMS own the success of the customer in the account itself.”
Listen to the episode Retention & customer success Link to this Report a problem
FDE is not professional services rebranded, and that many companies running it are unsure what outcome it serves. Listen
He said many companies experimenting with an FDE motion think it belongs in selling or success and are not sure what outcome it serves. Diego said Intercom's FDE motion is a different motion from consulting or professional services.
“This is not like consulting or professional services like rebranded like it's a different motion.”
Listen to the episode Strategy & market Link to this Report a problem
Containment and deflection are easy to achieve with AI support, while sustainable value is not. Listen
He said it is very easy to achieve great containment and deflection, but it is not easy to achieve great sustainable value. Intercom's data scientists help identify where the most time is spent and where value is for the customer.
“it's very easy to achieve great containment and great deflection. It's not easy to achieve great value, sustainable value, right?”
Diego's advice is to validate whether FDE suits your business, because you may not need it. Listen
He said there is a lot of buzz, and every AI company is doing a version of this, but FDE might not be the right motion for a given business. Diego said Intercom found it useful because it built better products and gained customer insight it otherwise would not have had.
“Maybe you don't need it. Maybe you don't need it.”
Listen to the episode Leadership & culture Link to this Report a problem
Intercom's forward deployed engineering began as an unplanned response to mixed Fin results. Listen
Diego said the team was not set up to run an FDE motion when it started about 18 months ago. It noticed some Fin customers got excellent results and others got weaker ones, and wanted to understand why. The team found several components to the equation, including education, self-manageability and white-glove deployment.
“We didn't like set, set ourselves up to basically go and do an FDE motion.”
Listen to the episode Strategy & market Link to this Report a problem
Two years ago most customers had no one on their side to partner on the deployment, Diego said. Listen
When Josh suggested that 98 percent of customers lacked that person, Diego agreed that was about right two years ago. He said it has become more common since, as customers have added conversation designers, full-time content people and knowledge managers.
“I think two years ago was like 98% to your point.”
Listen to the episode Retention & customer success Link to this Report a problem
Good AI support outcomes depend on a strong knowledge base that captures business policies. Listen
He said customer service contains a lot of silent knowledge about the business, such as guidelines and policies, that has to be captured. He said customers now understand that without a great knowledge base there is no way to get great outcomes with AI.
“without a great knowledge base, there's no way you'll get great outcomes with AI”
Diego predicts customer-side roles such as conversation designers and knowledge managers will become more prevalent. Listen
He said customers increasingly have conversation designers, people working on content full time and knowledge managers, and called this trend really exciting. He expects these roles to become more common over time.
“These roles will become more and more prevalent over over time.”
Listen to the episode Retention & customer success Link to this Report a problem