Mihshuv Yashir: How Much of the BPO Backlog Surge Is Actually Firm
Mihshuv Yashir's BPO backlog jumped to NIS 1.392 billion, but nearly 38% of it is non-binding, and even part of the rest relies on work without a formal order and on contracts that can be terminated on prior notice. That does not cancel the growth engine, but it does shift the 2026 test from the backlog headline to actual revenue conversion and margin.
How firm the backlog really is
The main article argued that BPO is the most visible growth engine inside ONE. This follow-up isolates the question the headline number does not answer on its own: how much of the surge in backlog really equals revenue visibility, and how much still depends on customer behavior, ordering habits, and contracts that can be reopened.
The short answer is that the backlog is real in terms of activity depth, but softer than the headline suggests. As of December 31, 2025, BPO backlog stood at NIS 1.392 billion, up from NIS 443.6 million a year earlier, a jump of about 214%. That is a large number, but inside it already sit NIS 527.6 million of non-binding orders, about 37.9% of the total.
The more important point is that the softness does not stop at the line explicitly labeled non-binding. In the backlog note itself, the group says that a significant portion of its contractual engagements in the segment allow the customer to terminate for any reason with several months' prior notice. At the same time, it says the backlog also includes revenue for which no formal order has yet been received, but which the group still treats as binding based on work already started, a final tender award, or prior experience with framework agreements that are assumed to renew.
So there are really three layers of firmness here: non-binding orders; orders treated as binding even though they do not yet have a formal order behind them; and contracts that do exist but still give the customer an exit right on prior notice. That is not the same thing as a backlog made up only of signed, fixed, non-cancellable orders.
The chart also shows another important point: the softer layer is concentrated mainly in the near term. Out of the 2026 backlog, which stands at NIS 578.3 million, about NIS 320.6 million is non-binding, roughly 55.4% of the coming year. By contrast, the 2029-and-later portion is mostly classified as binding. That distinction matters because the market usually gives the most weight to the next 12 months, precisely where contractual firmness is weaker.
| Backlog layer | What the filing says | Why it matters |
|---|---|---|
| Non-binding orders | NIS 527.6 million in total, including NIS 320.6 million tied to 2026 | This is workload visibility, not full contractual lock-in |
| Orders treated as binding without a formal order | Work has already started, there is a final tender win, or framework agreements are assumed to renew | Even part of the "binding" bucket still relies on judgment and prior behavior |
| Contracts with customer exit rights | A significant portion of contracts can be terminated on several months' notice | Even apparently signed revenue can be less durable than the headline backlog suggests |
Why this is not just an accounting nuance
One could argue that this is still a conservative way to describe backlog, and that execution history is what really matters. Here the filing does provide an important balancing point: during 2025 there were no material cancellations or changes in practice versus the 2025 backlog disclosed at the end of 2024, and near the report-approval date the changes in backlog were described as immaterial.
That means the backlog is not fictitious. The company is not showing a history of large gaps between reported backlog and actual execution. But that still does not make the backlog firm in the full economic sense. What it does mean is that the company knows how to manage its customer base and contract structure so that even a relatively soft layer still rolls into real work. That is a meaningful difference. The risk here is not necessarily a sudden collapse. It is an overly aggressive reading of visibility quality.
In other words, this backlog says a lot about ONE's penetration with customers, its position in tenders, and its ability to hold a large volume of work. It says less about how contractually locked every shekel inside that number really is. Anyone reading it as if it were a hard industrial order book may miss that distinction.
Volume does not automatically mean economics
That brings up the second test, and it is even more important: is the surge in volume already translating into firmer economics. The 2025 picture says not yet, at least not fully. BPO segment revenue rose from NIS 313.7 million to NIS 499.3 million, an increase of about 59.2%. Segment profit rose from NIS 32.5 million to NIS 41.4 million, an increase of only about 27.2%. Segment margin fell from 10.37% to 8.29%, a decline of 2.08 percentage points.
That is the core of the continuation thesis. If backlog were not only larger but also economically firmer, one would expect at least part of that advantage to show up in profitability. Instead, 2025 looks like a year in which work volume accelerated much faster than the ability to turn it into segment profit. So the right question is not whether demand exists. It is at what economic cost that demand is being held.
The scale shift matters here as well. On April 17, 2025, ONE completed the acquisition of ONE Line, a company that employs about 2,400 workers and operates outsourced customer-contact and service-center activities. The segment note states that from the first consolidation date, ONE Line is reviewed as part of the BPO and technology-support segment. At the same time, the number of employees in that segment rose from 1,794 at the end of 2024 to 4,334 at the end of 2025. In other words, almost the entire jump in BPO headcount is already explained by the addition of ONE Line.
That is critical for how 2025 should be read. The surge in backlog and the surge in headcount are not clean organic numbers. They sit on a broader activity perimeter, an acquired contact-center platform, and a heavier service layer. What the filing does not provide is a separate breakdown of ONE Line's exact contribution to segment revenue, profit, and margin. So it is not possible to say precisely what caused the margin compression. It is possible to say that the 2025 numbers blend growth, acquisition, and mix change, and that economics have not yet expanded at the same pace as backlog and staffing.
What the market should measure now
That leads directly to the follow-up conclusion. BPO backlog is not a number that should be dismissed, because the company showed no history in 2025 of material cancellations against this backlog. But it is also not a number that should be read naively as locked-in future revenue. There is a large non-binding layer inside it, part of the binding layer includes work without a formal order, and above both sits a contract base that can be terminated on prior notice.
That is why 2026 will be a test of backlog quality, not just backlog size. If the larger backlog keeps rolling into revenue without unusual leakage, and if segment margin stabilizes or improves at the same time, then 2025 will look like a transition year in which ONE built a larger platform and simply needed time to absorb the scale jump. If, by contrast, revenue keeps rising while margin stays under pressure, or if visible gaps appear between backlog and execution, then the market will start reading the backlog surge less as contractual strength and more as volume that was bought at a high economic cost.
The right read today is therefore not that the backlog is weak, and not that the backlog closes the case. The right read is that ONE has built a much larger BPO layer, but the contractual quality and the economic quality of that layer still need to be proven in the next reports.
Bottom line
The BPO backlog surge is real in terms of scale. It is also backed by a meaningful perimeter change, after the ONE Line acquisition and more than a doubling of segment headcount. But contractual firmness and segment economics are not the same thing. Nearly 38% of the backlog is non-binding, more than half of 2026 backlog sits in that layer, and even inside the part classified as binding there is work without a formal order and contracts that can be terminated.
The reasonable counter-thesis is that this caution may be excessive: 2025 did not show material cancellations, so even a relatively soft backlog may still be a good indicator of execution. That is a serious argument. But as long as revenue is growing faster than profit, and the filing does not separately break out ONE Line's contribution or contract quality within the backlog, the more careful reading still makes more sense.
So the next test is very clear: the conversion rate of 2026 backlog into revenue, segment margin after the first full year following the ONE Line integration, and the company's ability to show that the larger BPO layer produces not only more work, but also visibility and profitability the market can rely on.