Expansion revenue costs a fraction of new-logo revenue to acquire, converts at several times the rate, and compounds. It is also the thing most SaaS companies leave to chance, assigning it to account managers as an aspiration rather than building it as a system.
Six plays cover nearly all of it. Most companies should run two or three well rather than all six badly.
Play 1: Seat expansion
The simplest mechanism. The customer adds people, you bill for them.
What makes this work is not sales effort but product design. Products that spread inside an organisation have shared objects, invitations, comments and notifications that pull in colleagues. Products that do not spread are single-player tools, and no account manager can fix that with a quarterly call.
The operational piece is removing friction: self-serve seat addition, no quote required, prorated billing handled automatically. Every approval step between wanting a seat and having one costs you expansion.
Trigger: licensed seats at 85% utilisation, or a new domain appearing in invitation attempts.
Play 2: Tier upgrade
Moving a customer from one package to a higher one. This depends entirely on packaging: if your tiers are not meaningfully different, there is nothing to upgrade to.
The feature that drives most upgrades in B2B is rarely glamorous — single sign-on, audit logs, advanced permissions, SLAs. These are the things that become necessary as the customer grows or as IT gets involved, which makes the upgrade feel like a natural step rather than a sale.
Keep at least one such capability above the tier most customers land on. Our notes on pricing page structure cover how to do that without looking punitive.
Trigger: a user attempts a gated feature, or the account crosses a size threshold where compliance requirements typically appear.
Play 3: Cross-sell a second product
Harder than it looks, and frequently attempted too early. A second product only expands revenue if it serves the same buyer with an adjacent problem. If it serves a different department, you are running a new-logo sale inside an existing account, with all the same cost.
The test is whether your current champion can approve it. If they cannot, treat it as net-new business and resource it accordingly.
Trigger: behaviour in the first product that indicates the second problem — exporting data to a tool you also sell, for instance.
Play 4: Usage growth
Where pricing is tied to consumption, expansion happens without a conversation. This produces the strongest net revenue retention figures in software.
The work here is not selling but enabling: helping the customer find more use cases, more volume, more workloads to move onto the platform. Adoption work and revenue work become the same activity, which is the structural advantage of the model.
The trade-offs — forecasting difficulty, downside exposure, bill shock — are covered in our piece on usage-based pricing.
Trigger: consumption approaching the committed amount, or a plateau that suggests an unaddressed blocker.
Play 5: Price uplifts
The most neglected play in SaaS. Many companies have not increased prices for existing customers in years, which means every year of inflation and every feature shipped has been given away.
Two mechanisms. A contractual annual uplift — commonly 3–7% — written into new agreements, which compounds silently and requires no selling. And periodic list price increases applied to the existing base with notice, which is harder and needs a value story.
The common fear is churn. In practice, a well-communicated single-digit increase on a product customers depend on produces very little cancellation, and the revenue is almost pure margin.
Trigger: renewal date, with the uplift already in the contract so it is not a negotiation.
Play 6: Multi-year commitments
Not expansion in the strict sense, but it increases contracted value and removes a renewal risk. A two or three-year term in exchange for a modest discount and a locked-in annual uplift is often a good trade for both sides.
This works best with customers who have already proven value and whose usage is stable. Offering it to a struggling account is how you lock in a bad relationship for three years.
Trigger: second renewal, strong health score, demonstrated outcome.
Building the system behind the plays
Three pieces of infrastructure make expansion repeatable rather than opportunistic.
- Triggers routed to a human or an in-product prompt. Not a monthly report somebody reads. The value of a trigger decays within days.
- A self-serve path for anything small. Seats, usage top-ups and single-tier upgrades should never require a conversation. Reserve people for the large or complex moves.
- Clear ownership and compensation. Decide whether account managers, CSMs or a dedicated expansion team own each play, and pay them for it. Expansion that nobody is paid for does not happen reliably.
What usually blocks expansion
When expansion is flat, the cause is almost always structural rather than effort-related.
No headroom in the packaging. Everyone is already on the top tier. This is a pricing project, not a sales one.
Value does not deepen. The product is as useful in month twelve as in month one, so there is no honest reason to pay more.
The product does not spread. It stays with its original buyer because nothing in it invites a colleague.
Friction in buying. Adding a seat requires a quote, a purchase order and two weeks.
Each of these has an owner outside the account team, which is why expansion targets handed to customer success in isolation so rarely move.
Forecasting expansion
Expansion revenue is more predictable than new business, and most companies forecast it far less rigorously, which leaves accuracy on the table.
Build the forecast bottom-up from the installed base rather than as a percentage assumption. For each account: current contract value, renewal date, seat utilisation, usage trend, and whether a specific expansion trigger has fired. That produces three buckets — expansion already identified, expansion likely from utilisation, and flat — and the first two can be forecast with real confidence.
The discipline worth adding is to forecast contraction alongside it. Downgrades and seat reductions are almost never forecast, and they are the reason expansion forecasts routinely overshoot by a wide margin. Accounts with utilisation below 50% at renewal are contraction candidates, and treating them as such a quarter early gives you time to fix the adoption problem instead of negotiating the reduction.
Expansion in a downturn
When customers are cutting budgets, the standard plays stop working and two different ones take over.
Consolidation. If your product can replace two others the customer is paying for, that is an expansion sale framed as a cost reduction — the only kind that gets approved in a freeze. This requires knowing what else is in their stack, which is a question worth asking in every review whether or not you are in a downturn.
Term extension in exchange for price certainty. A customer facing budget uncertainty may value a locked three-year rate more than a small discount. You get committed revenue; they get predictability.
What does not work in a downturn is the feature-led upgrade. Nobody is approving net-new spend for a capability that is nice to have, however good the demo.
Frequently asked questions
When should expansion start?
After the customer has achieved their first success, not before. Expansion attempted during a struggling implementation damages the relationship and rarely closes.
Who should own expansion revenue?
Whoever has the commercial relationship and is compensated for it. Splitting it between CSMs with no quota and AEs focused on new logos usually means nobody does it.
Does discounting expansion make sense?
Volume discounts at genuine thresholds are reasonable. Ad hoc discounts to close an upgrade teach customers to wait for one, and they reset the price permanently.
How much of new revenue should come from expansion?
At scale, 30–50% is a common and healthy range in mid-market and enterprise SaaS. Much lower suggests untapped potential; much higher can mean new-logo acquisition has stalled.
