The HubSpot Deal Loss Agent Playbook

Turn lost deals into a playbook

HubSpot's Deal Loss Agent is the part of Breeze that reads your losses so your team can stop repeating them. Point it at a timeframe, and it reviews the deals you marked closed lost and the records around them, then generates a report that names the common loss patterns and recommends what to do about them. You can ask it a specific question or let it surface general recommendations, and you can share the report or download it as a PDF. It is currently in Beta and it runs in Breeze Studio.

The distinction that matters is that this is analysis, not a dropdown. Most teams capture a loss reason on every closed-lost deal and then never look at it again. The Deal Loss Agent does the part nobody has time for: it reads across all of those deals, plus the calls, emails, and notes attached to them, finds the themes, and turns them into an argument about where you are leaking revenue and how to fix it.

This playbook is the complete picture: how it works, what it reads, the two ways to prompt it, the report it produces, setup and permissions, how to configure and automate it, an honest read on where it fits and where it falls short, and how INSIDEA deploys it for revenue teams. Every product detail here is checked against HubSpot's own documentation, because the feature is in Beta and moving quickly.

The problem it removes: loss reasons nobody reads

Closed-lost is where sales intelligence goes to die. A rep marks a deal lost, picks a reason from a dropdown, and moves on to the next opportunity. The reason gets stored and never read. Multiply that across a year of losses and you are sitting on the single richest dataset about why you do not win, and almost no one has the time to analyze it.

When teams do try, it is a heavy lift. Real win-loss analysis means pulling the closed-lost deals, reading the notes and call summaries, clustering the reasons, separating the noise from the signal, and writing it up. It is the kind of project that gets scheduled, slips, and quietly dies. So the patterns that are costing you deals, a pricing objection you could answer, a competitor you keep losing to, a segment that never converts, stay invisible.

That is the tax the Deal Loss Agent removes. It does the reading and the clustering across every lost deal in the window, and it writes the analysis, so the intelligence in your closed-lost pile finally gets used. The point is not to replace sales judgment about what to change. It is to put the evidence in front of the people who can change it.

How it works, from closed-lost to report

The agent runs a clear loop. You set the scope, it reviews the losses and the records around them, it finds the patterns, and it produces a report. Understanding the loop matters because each step maps to a choice you make when you configure it.

You define a timeframe to search across, for example a quarter. The agent reviews the deals marked closed lost in that window and their associated records, the calls, emails, notes, and documents tied to each one. It clusters what it finds into common loss patterns, and it generates a report with those patterns and concrete action recommendations. You steer the run in one of two ways: enter a specific question you want answered, or leave it open and let the agent return general recommendations.

What it reads, and the two ways to prompt it

The agent's raw material is your own loss history: the deals marked closed lost across the timeframe you set, and the records associated with them. That association is what makes it more than a chart. It is not just reading the loss-reason field; it is reading the calls, the emails, the notes, and, where you enable it, the text inside documents attached to the deal.

You also give it context. When you configure the agent you can add knowledge, such as your ideal customer profile, so it evaluates losses against who you actually want to win, and you can add tools that extend what it can do, like retrieving document text from a file property. Then, at run time, you choose how to prompt it.

Most teams use both: an open run to see what the losses are shouting about, then specific questions to pressure-test a suspicion. Because you can save and revisit runs, the analysis becomes a conversation with your own data rather than a one-time export.

The report and what you do with it

The output is a report, not a data dump, built to be read and shared. It names the common loss patterns and pairs them with recommended actions, and HubSpot gives you several ways to work with it depending on who needs to see it.

The ability to see the agent's process matters more than it sounds. Win-loss findings only change behavior if people trust them, and being able to expand the analysis and see which deals and signals drove a conclusion is what turns a report into a decision the team will actually act on.

Setup and what you need

The agent lives in the Breeze Marketplace and is set up through Breeze Studio. It is broadly available across editions, which is part of what makes it approachable, but it has real prerequisites a Super Admin has to handle first. It is in Beta, so confirm the current requirements on HubSpot's own pages before you build a process on it.

One practical note that decides how useful the first run is: the agent reads your closed-lost deals and the records around them, so the quality of the analysis tracks the quality of that data. Consistent loss reasons and real notes on lost deals produce sharp patterns. Empty fields and blank notes produce vague ones. The setup that matters most is not technical; it is the loss-capture discipline underneath.

Configure and automate it

Because it runs in Breeze Studio, the Deal Loss Agent is configurable rather than fixed, and the configuration is where you make it fit your business. A Super Admin sets it up once, and the whole team runs against the same, tuned analysis.

You set the timeframe it searches across, so the analysis matches your review rhythm. You add knowledge, most importantly your ideal customer profile, so losses are judged against the customers you actually want. You add tools that extend what it can reach, such as pulling text from documents attached to a deal. You manage access, deciding who can edit the agent versus simply run it. And you test it with a question before you publish, so the first real run is not a guess. On Professional and Enterprise, you can go one step further and put it on a schedule.

Configure it once, to your ICP and your review cadence, and win-loss analysis stops being a project you keep meaning to run and becomes a report that is always waiting for you.

Scheduling is the quiet superpower. A trigger that runs the analysis at the close of every quarter, or every month, turns win-loss from an occasional scramble into a standing input that leadership can count on, without anyone having to remember to pull it.

The use-case library

The agent is not only for a formal quarterly review. The teams who get the most from it point it at every place a loss pattern hides. Here is the practical library.

Where it fits with the other Breeze agents

The Deal Loss Agent is one of several Breeze agents, and knowing its lane keeps expectations right. This is the retrospective agent: it looks backward at what already happened and turns it into learning. It does not source new pipeline or answer a customer; it explains why the pipeline you had did not close.

Together they form a loop. The Deal Loss Agent tells you why you are losing, that insight sharpens your ICP and your plays, the Data Agent and Company Research Agent make sure reps act on better information, and the Prospecting and Customer Agents carry out the motion. Closing the loop, feeding what you learn from losses back into how you sell, is where a team stops repeating the same expensive mistakes.

Pricing and where it falls short

On cost, the honest summary is that the agent comes with a qualifying subscription rather than a separate per-use fee you reason about. It is available across Marketing, Sales, Service, Data, and Content Hub on Starter and above, with scheduled automation reserved for Professional and Enterprise. Because it is in Beta and HubSpot's AI features run on a credit model that evolves, confirm the current credit usage and any limits on HubSpot's pricing page before you lean on it heavily.

None of these are reasons to avoid it. They are reasons to fix your loss-capture discipline first, scope each run with intent, and treat the report as evidence for a decision rather than the decision itself. Do that and the losses you were already paying for start paying you back in learning.

How INSIDEA runs it

The agent is powerful, but its value is capped by the data underneath it and the discipline around it. The teams who turn it into fewer losses are not the ones who ran it once; they are the ones who fixed the foundation and built the cadence. This is the sequence we use with clients.

Start with loss capture, because the agent can only find patterns your team actually recorded. Standardize the closed-lost reasons, and make real notes on lost deals a habit, before you expect sharp analysis. Add your ideal customer profile as knowledge so losses are judged against the customers you want to win. Scope the first run to a clear window and read the general patterns, then drill in with specific questions. Put it on a schedule so win-loss becomes a standing input at every quarter close, not an afterthought. Then close the loop: route the findings into pricing, process, and enablement, and feed the sharper ICP back into how you target and how you sell.

The agent reads your losses so you stop repeating them. Your job is to capture losses accurately, scope the question well, and act on what the report tells you.

This is exactly the work INSIDEA does every day. As an Elite HubSpot Partner, we have built HubSpot RevOps and sales foundations for more than 1,500 businesses across 25+ countries, and the pattern that holds up is always the same: clean data, sharp questions, findings wired into action, an honest loop from loss to change. If you want the Deal Loss Agent set up to actually reduce losses rather than just describe them, start with a strategy call and we will map your loss-capture, your review cadence, and where the findings should drive change first.