Three years ago, almost nobody reviewing an eSIM brand mentioned talking to a bot. Now it's routine. In our refreshed eSIM Customer Service Benchmark, the share of support reviews that mention AI, a chatbot or automated replies went from 0.6% in 2023 to 4.7% in 2024 and 6.4% so far in 2026.
The more useful finding is what doesn't follow from that. Brands with a lot of visible AI in support don't have happier customers, and brands with very little don't either. The presence of AI tells you almost nothing. What it can do tells you a lot.
Quick background: the benchmark covers the 20 largest eSIM brands on Trustpilot, 377,827 reviews. For this angle we read 9,883 confirmed support reviews, searched them for 17 AI-related terms and checked every hit by reading the review. Disclosure: I run Aissist, which sells AI support agents, so I have a horse in this race. That's partly why I wanted the data to be public.
Where AI is most visible
Across the industry, 5.4% of support reviews mention AI. It is most visible at BNESIM (18%, on a small sample of 82 reviews), Airalo (10%), Ubigi (9.2%), and Saily and eSimFLAG (7.6% each). It barely registers at GigSky (0.9%), eSIM.me (1.3%), Simify (1.5%) and aloSIM (1.8%).
Treat all of these as floors. AI only gets counted when a customer notices it and says so. A bot that quietly fixes your problem in 40 seconds tends to get reviewed as "support was quick," not "the AI was great." And Airalo and Holafly both hit Trustpilot's 400-result cap on the "AI" and "bot" searches, so their counts are understated.
More AI doesn't mean happier customers or unhappier customers
Put AI visibility next to how negative a brand's support reviews are, and there's no clean line:
Saily and Ubigi are the clearest pair. Similar AI visibility, and support reviews that are 16% negative at one and 60% at the other. MobiMatter shows the opposite shape: little visible AI, and 59% of its support reviews negative anyway. Whatever separates good support from bad in this category, "has a chatbot" isn't it.
What customers complain about when they complain about AI
In the ranking, Bot walls come fourth. The two biggest support failures across the industry are problems support couldn't fix (29 of every 100) and slow or missing replies (28). Those are the exact problems automation is supposed to solve. An AI that answers instantly fixes the second one. An AI that can't act on the account doesn't fix the first one. It just gets to "sorry, I can't help with that" faster.
The one thing that matters: can it resolve, and can it hand over?
My read of these reviews is that good and bad AI experiences come down to two questions.
Can it actually do something? Reissue the eSIM, check whether data is flowing, see that the plan expired early, apply a credit. An assistant that only answers FAQs is deflection, not resolution, and a traveler stranded without data notices the difference quickly.
When it can't, does a human take over, with context? The failure mode people write one-star reviews about isn't "I talked to a bot." It's "I talked to a bot and there was no way out." A handoff that carries the conversation over, so the customer doesn't repeat themselves, turns an AI miss into a normal support interaction.
If you're evaluating AI for eSIM support, those are the two questions to ask vendors. Include us.
The takeaway
AI is now a visible part of eSIM support and growing fast. On its own, it doesn't move customer sentiment in either direction. The brands getting it right use automation to be reachable instantly and to fix things, with a human one step away. The ones getting it wrong built a faster way to say no.
Full per-brand data and method: eSIM Customer Service Benchmark 2026
Top comments (0)