2026 impact report: 237,000 beneficiaries, and 84,000 unique users
From July 2025 to May 2026, the SOLIS bot reached approximately 237,000 beneficiaries through 84,232 unique users, across eight deployments: five in Lebanon (CARE, LRI, TDH, Nabad, SI Lebanon) and three pilots (Ukraine, Cameroon, Haiti). To give the full magnitude of the activity, that represents 382,978 information-center messages, 75,558 broadcasts and 12,501 surveys completed over the period. The reach exceeds the objective set with Twilio, our technology partner, by 28%.
Reach
The Lebanon deployment, live since 2021, accounts for 99.4% of the total. That is maturity talking: the three pilots (Haiti, Cameroon, Ukraine) are only starting, and their per-beneficiary ratios do not yet reflect steady-state performance. We calculate beneficiaries from unique users, multiplied by average household size and adjusted by an overlap rate, rather than declaring the figure directly. That caution is a choice, not a constraint.
A division of labour between organisations
The numbers reveal that each partner made a different use of the tool its own. CARE carries the information-center volume, with 278,737 messages, nearly three-quarters of the total. Nabad leads on broadcasting, with 27,448 messages pushed, the highest volume in the network. SI Lebanon tops the combined surveys and broadcasts, while LRI and TDH show more targeted activity, matching their scope.
The share of organic contacts, the conversations beneficiaries start on their own without a broadcast prompting them, ranges from 54% at SI Lebanon to 97% at CARE and SI Haiti. A high rate signals that people return to the bot of their own accord: the best proxy for perceived usefulness we have.
Usage tracks the crisis
Lebanon is the only deployment with enough historical depth for a time-series read. Over the last eleven months, usage peaks coincide precisely with security incidents: 9,044 users in July 2025 during targeted strikes, 11,612 in December 2025 after a wave of assassinations, 12,007 in March 2026 at the height of the strikes and displacement, 11,616 in May 2026 during CARE's assessment campaign. Conversely, during the April 2026 ceasefire, traffic fell back to 4,404 users.
The read is unambiguous: the tool is used most heavily exactly when physical access to communities is most constrained. That is its reason for existing.
Operational value
Compared with manual channels, the bot generated the equivalent of €865,000 in staff time in Lebanon, roughly 22 full-time positions. We reach that figure by valuing each message shared at the human time it would have taken to share the same information by phone or in person, at an hourly rate between €15 and €25. The resulting value-to-cost ratio ranges from 2.6 at LRI to 50.3 at CARE, depending on usage intensity.
Cost per beneficiary sits between €0.12 and €0.22 on mature deployments, with the flat licence structure driving that figure lower with every additional beneficiary. By shifting part of the data collection to mobile, the system also avoided more than €76,000 in field travel costs.
Effects for beneficiaries
Beyond the savings for organisations, the bot changes daily life for the people it serves. Starting from a reduction documented by the GSMA, from 4.5 hours of waiting to a few minutes, and taking a conservative estimate of three hours saved per interaction, we estimate more than 29,000 days of waiting time spared for Lebanese beneficiaries over the period.
We also estimate that around 11,700 risky trips were replaced by a remote interaction, close to 23,400 hours of travel avoided. Applying to those trips an incident rate drawn from the humanitarian literature on gender-based violence, we approach the order of magnitude of 509 potential protection incidents avoided. That last figure is hard to document, and we present it as an order of magnitude, not a measurement.
Environmental efficiency
Shifting collection to mobile avoided roughly 376 kg of CO₂e, based on an average field round trip of 100 km. A second effect adds to it: the SOLIS bot's engine is deterministic, built on rules written by humans, not on a large language model. Its inference footprint is therefore negligible compared with a generative chatbot handling the same volume, an advantage that matters at a time when the carbon cost of AI has become a subject in its own right.
Our methodology
Every figure in this report rests on explicit assumptions, and we publish them. Beneficiaries are calculated as the product of unique users and average household size, 4 people in Lebanon and Ukraine, 5 in Cameroon and Haiti, adjusted by an overlap rate bounded between 10% and 50%, whose upper bound is calibrated on Nabad's data. Waiting time saved takes three hours per interaction, against the 4 h 26 documented by the GSMA, as a matter of caution. The HR valuation draws on a range of €15 to €25 an hour.
We own the limits too, because they are part of an honest figure. "Unique users" is an estimate, not a measurement: the bot is public, and our phone-number retention policy, capped at two months by default, rules out any individual tracking by design. Pilot figures are not representative of an established regime. Reach is presented as ranges, never as exact counts. Honest measurement is part of accountability, and we would rather offer a solid order of magnitude than a misleading precision.
The full report is available on request.
To help us improve our methodology, please feel free to contact us at: solisbot@solidarites.org