LivePerson: Is ARPC Rising Because the Customer Is Better, or Because the Base Is Smaller
ARPC rose to $680 thousand in 2025, but retention fell to 78% and the contracted base kept shrinking. At LivePerson, the higher average customer still looks more like the result of a smaller base than proof of a healthier one.
The main article argued that LivePerson bought time through cost cuts, but did not yet repair the trust, debt and revenue problems underneath the story. This follow-up isolates one of the more important questions below the surface: does the rise in ARPC in 2025 really point to a better customer base, or mainly to a smaller one that now consists of relatively heavier accounts.
The evidence leans toward the second explanation. ARPC rose to $680 thousand from $625 thousand in 2024, but retention fell to 78% from 82%, still far below the company's long-range target of 105% to 115%. At the same time, management describes slower-than-expected renewals and weaker new bookings, driven in part by customer uncertainty around LivePerson's financial stability. Add lower deferred revenue, a shrinking contract acquisition cost asset, and hosted revenue that is still falling sharply, and it becomes difficult to read ARPC as a clean sign of a better customer.
LivePerson asks investors to view ARPC and retention independently from revenue, deferred revenue and remaining performance obligations. That is too convenient. In a software company, customer quality is not measured only by the average account size. It also has to show up in whether the contracted revenue base stops shrinking.
| Metric | 2024 | 2025 | Read-through |
|---|---|---|---|
| ARPC | $625 thousand | $680 thousand | 8.8% increase in average customer size |
| Retention | 82% | 78% | Revenue from the installed base still shrank even after upsells |
| Hosted services revenue | $261.7 million | $207.6 million | The recurring engine is still contracting |
| Total deferred revenue | $58.3 million | $54.4 million | Less contracted revenue is sitting on the balance sheet |
| Contract acquisition costs | $33.6 million | $24.0 million | The sales asset is running down instead of rebuilding |
ARPC is up, but the existing customer is still spending less
The important point is not ARPC on its own, but the gap between ARPC and retention. ARPC measures average recurring revenue per enterprise and mid-market customer over the trailing twelve months. Retention measures the trailing-twelve-month change in total revenue from existing customers after upsells, downsells and attrition. So when LivePerson reports a higher ARPC but only 78% retention, it is effectively saying that the customers who remain are larger on average, while the existing base as a whole is still generating less aggregate revenue than it did a year earlier.
That is exactly why ARPC is not enough to declare better customer quality. If the base were genuinely improving in a broad way, retention would at least be stabilizing. Instead, it fell another 4 percentage points and remained 27 to 37 points below management's own long-range target range.
Management's explanation points in the same direction. The company describes slower-than-expected renewals and new business bookings, linking that weakness both to customer concerns about financial stability and to longer enterprise buying cycles, including for higher-value AI solutions. That is not the language of a customer base getting healthier. It is the language of a sales motion that remains harder than it should be, even if some of the accounts that stayed are bigger.
The contracted base does not confirm improvement
The second layer of evidence comes from the contracted base. At the end of 2025, remaining performance obligations stood at $175.6 million, with about 98% expected to be recognized within 24 months. That gives a floor of visibility, but not much more. The definition excludes amounts that are cancelable by the customer, optional renewal periods, and obligations that are billed and recognized as performed. So RPO should be read as limited contracted visibility, not as proof that the engine is already rebuilding.
Deferred revenue tells a more cautious story. Total deferred revenue fell from $82.0 million at the end of 2023 to $58.3 million at the end of 2024 and then to $54.4 million at the end of 2025. In 2025, the company also recognized $57.9 million of revenue that had been sitting in beginning deferred revenue, yet it still ended the year with a lower balance. Put simply, the bucket kept draining.
The contract acquisition cost asset is no more encouraging. LivePerson explains that this asset consists of incremental sales commissions paid to obtain contracts and that it is amortized over four years. The balance fell from $33.6 million to $24.0 million. That is not just routine amortization. Using the opening balance, ending balance and the disclosed $17.3 million of amortization expense, the implied 2025 replenishment was only about $7.7 million of new contract acquisition costs, versus roughly $14.5 million in 2024. In other words, the new layer coming onto the balance sheet is no longer refilling the older asset at anything close to the prior pace.
That may be the single most important point in this follow-up. ARPC can rise even if smaller or weaker customers fall out of the base. But if deferred revenue is declining at the same time and the contract acquisition cost asset is not rebuilding, the more defensible read is of a narrower contracted base, not of a healthier sales engine.
Revenue disaggregation does not show broad improvement either
Revenue mix gives ARPC no help. Hosted services revenue, the heart of the recurring model, fell 20.7% to $207.6 million. Professional services revenue fell 28.8% to $36.1 million. The decline is showing up in both layers, not only in services.
| Revenue source | 2024 | 2025 | Change |
|---|---|---|---|
| Hosted services | $261.7 million | $207.6 million | -20.7% |
| Professional services | $50.8 million | $36.1 million | -28.8% |
| Total revenue | $312.5 million | $243.7 million | -22.0% |
If customer quality were improving across the base, the recurring hosted engine should at least be moving toward stabilization. Instead it is still shrinking sharply. That strengthens the view that ARPC is capturing a shift in who remains, rather than a system-wide improvement in demand or wallet expansion.
No customer is above 10%, but that still does not prove broad improvement
There is one point that does matter in LivePerson's favor: no single customer accounted for 10% or more of revenue in 2025, 2024 or 2023. That makes the simplest version of the story, one outsized whale driving the entire ARPC increase, less likely.
But it does not settle the issue. The filing gives no distribution below the 10% threshold, so it is impossible to know whether the ARPC increase reflects broad-based improvement or a smaller surviving group of larger customers. As an inference from the full evidence set, rather than as an explicitly disclosed fact, the second reading still looks more plausible.
Conclusion
The 2025 ARPC number says that customers still on the platform are larger on average. It does not prove that the customer is better. To make that case, LivePerson would have needed at least one of three things: stabilizing retention, a flattening in hosted revenue, or a rebuild in deferred revenue and the contract acquisition cost asset. The filing offers none of those confirmations.
So this follow-up pulls the question back to the right place. The key issue is not whether the average account got bigger. It is whether the contracted base stops shrinking. Until LivePerson can show real improvement in retention and in the pace of renewals and bookings, ARPC still looks more like the result of a smaller base than evidence of a stronger one.
Disclosure: Deep TASE analyses are general informational, research, and commentary content only. They do not constitute investment advice, investment marketing, a recommendation, or an offer to buy, sell, or hold any security, and are not tailored to any reader's personal circumstances.
The author, site owner, or related parties may hold, buy, sell, or otherwise trade securities or financial instruments related to the companies discussed, before or after publication, without prior notice and without any obligation to update the analysis. Publication of an analysis should not be read as a statement that any position does or does not exist.
The analysis may contain errors, omissions, or information that changes after publication. Readers should review official filings and primary sources before making decisions.