There is a fast way to tell whether a company's expansion revenue is actually under control. I ask to see the upsell pipeline. If they open the same board as new business, with the new logos and the renewals and the cross-sell deals all stacked in the same stages, I already know the forecast is fiction. Expansion is the cheapest revenue most B2B teams have, and it is almost always the worst-instrumented. The issue is not that these teams do not care about the base. It is that they are running a fundamentally different motion through a pipeline that was designed for something else.
Why this is worth fixing now
Expansion has quietly become the main event. ChartMogul's retention work puts expansion at up to 40% of growth for companies past $15M ARR, up from 30% in early 2021, and for any company holding net revenue retention above 100%, expansion accounts for more than half of all revenue. SaaStr has made the same point from the other side: 73% of Salesforce's new bookings come from its existing customers. When more than half your growth comes from the base, you cannot keep running the base as an afterthought on the new-business board.
New business and expansion are not the same motion
A new-business deal is linear. One opportunity, one close, a clean set of stages that a rep walks a prospect through once. Expansion does not behave that way. The same account throws off many opportunities over years, the timing is driven by events on the customer's side rather than yours, and the deal is really a standing relationship with a dollar amount attached. Force both motions through one pipeline and three things break at once: you cannot separate a new logo from an upsell in the forecast, your stages mean different things to different reps, and the board fills up with deals that have no active buying journey behind them. I am the first person to tell a team not to overcomplicate their CRM. But one pipeline for two motions is the wrong kind of simple.
Open a deal the day you close the customer
Here is the idea. The moment a new-business deal closes, open an expansion deal on that account, and let it sit in an open holding stage. Every account in your book then has exactly one open expansion deal, attached to that customer's current spend.
Open is not your pipeline. It is where potential lives. Your reps only work stages two through five, from acquiring interest to closed. Open exists so that the day-to-day activity an account manager does against the current book gets logged against a real number, without cluttering the working forecast. The alternative most teams reach for is to have reps create an upsell deal by hand only when something concrete appears. That feels leaner, and it costs you two things: you lose the systems-level view of net-new money coming into each account, and you lose all the activity tracking on the accounts you are nurturing but not yet actively selling.
Manual in, automated forward
The debate that always comes up here is whether stage movement should be automated or left to the rep. The answer is both, at different points in the pipeline.
Entry is manual. The account manager moves a deal out of open and into acquiring interest when a real buying journey starts: they are pitching a new product, a new team, an expanded use case. That manual move is the signal that a human decided this opportunity is live. It carries intent that no automation can infer from behaviour alone.
Progression is automated. Once a deal is in the working pipeline, let events carry it forward. An expansion event over a threshold you set, say $20k of new commitment, advances the deal automatically. Copy the activity forward from the closed new-business deal with a lookback window so the expansion deal is not born empty. The rule is simple: humans supply intent at the top of the pipeline, the system supplies the mechanics downstream.
Two views, not two systems
You do not need two objects or two dashboards for this. You need two views of the one expansion pipeline.
The first view is the working pipeline: acquiring interest through funding and closed. That is your real expansion forecast, and it should be small. The second view is a list of every account whose expansion deal is still sitting in open. Those are the customers you are not actively working. Read correctly, that list is a prospecting queue for your own base: these are the accounts where I do not have a live deal, so maybe I should go start one. You get potential and active in one place, without asking a rep to reconcile two reports or hunt across two objects.
The attribution question you will not fully solve
The moment you automate expansion-deal creation and progression, you inherit a hard question: who actually drove this number? Was it the account manager, was it product-led because the customer simply added seats, or was it automation firing on an event nobody touched? You will not answer that perfectly, and you should be honest with yourself about it. What you can do is measure sales activity on the deal, count the calls, emails, and meetings in the window, and classify each closed expansion as product-led, automation-led, or rep-led. That is enough to protect comp and to know whether your account-management motion is genuinely moving expansion or just taking credit for revenue that was going to land anyway. Treat it as a heuristic, not a proof.
While you are in there, add one hygiene mechanic: if a deal's close date passes and nobody has updated it, it dies back to open or closes lost automatically. Leave that out and stale expansion deals will sit in the pipeline forever, and your forecast rots from the inside. Making the close date actually mean something is the cheapest forecast-accuracy fix there is.
A pattern from the field
We recently worked with a Series B B2B SaaS team in the DACH region with a strong account-management motion and, on paper, healthy expansion. The problem showed up the moment we opened the CRM. New logos, renewals, and upsells were all living in one pipeline. The board carried hundreds of deals, most with no active buying journey, created on a campaign basis rather than because a customer was actually moving. Nobody could answer a simple question: of next quarter's number, how much is new and how much is the base? We rebuilt it as a separate expansion pipeline with an open holding stage, manual entry into the working stages, and an event threshold that auto-advanced qualified deals. Within a few weeks the working pipeline held a fraction of the deals, and the team had an expansion forecast they actually believed for the first time.
What I would do
- Split the pipeline. New business gets its own board, expansion gets its own. Stop reporting the two as a single number.
- Open a deal per account at close. Every won customer gets one expansion deal in an open holding stage, attached to their current spend.
- Define the entry signal. Write down exactly what moves a deal from open into the working pipeline, a pitched product, a new team, a renewal-plus-upsell conversation, and make that move manual and deliberate.
- Automate downstream, not upstream. Let events over a threshold advance deals that are already live. Never let automation decide on its own that something is a real opportunity.
- Build the two views. Working pipeline for the forecast, open-deal list as a prospecting queue for the base.
- Instrument attribution with activity. Count activity on the deal to classify product-led versus rep-led expansion, and treat the answer as a heuristic.
- Make close dates fatal. A passed close date with no update closes or resets the deal. No permanent residents in the pipeline.
Expansion is the highest-margin revenue you will ever book, and most teams run it on instrumentation they would never tolerate for new business. You do not need a new tool. You need a pipeline that matches the motion: separate from new business, open by default, worked deliberately, and honest about what is real. If your upsell and your new logos are still living on the same board, that is the first thing I would fix, and it is one of the things our revenue operations work does first. If you want a hand, get in touch. For the adjacent problem of where renewal revenue leaks before it ever reaches a pipeline at all, read the companion piece on the renewal data gap.
Sources
- ChartMogul. “The SaaS Retention Report: The New Normal for SaaS.” August 2024. chartmogul.com
- SaaStr. “What’s a Good Net Retention Rate in SaaS?” 2025. saastr.com
- Andreessen Horowitz. “Retention Is All You Need.” September 2025. a16z.com