← Insights

Premium for humans: how hotel software quietly made support a paid upgrade in 2025

marc

If you renewed a hotel software contract in the last twelve months, your support tier probably got worse. You weren’t told. The line item in the renewal looks the same. The price might even be the same. But the thing you actually receive when something breaks at half past nine on a Friday night is no longer the same thing.

I’ve been tracking this across the six platforms we run or have recently reviewed, and the pattern is consistent enough to call by name. Vendors are replacing human support with AI triage by default. Talking to an actual person is becoming a paid upgrade, an enterprise-tier feature, or a hotline that quietly closes early. Several of the most-used platforms in European hotels did this in 2024 and 2025. None of them announced it as a change. They re-priced the human.

This is a contract amendment that nobody signed. It deserves a name and a conversation.

Key findings

  • Six platforms have visibly shifted support delivery in the last twelve months: Mews, Cloudbeds, Guestline, RoomRaccoon, Duve, HiJiffy. We have specific aggregator-corroborated complaint patterns for each.
  • Mews introduced an AI-first triage layer in 2025; users on aggregators report 'premium' tier (with human escalation) priced at roughly a 70% upcharge.
  • Cloudbeds support degraded sharply post-launch in operator reports; templated responses and tickets closed without resolution are the most common 2025 complaints.
  • Guestline's support quality reportedly declined after the Access Group acquisition closed in July 2023; the pattern is repeatedly cited in operator forums.
  • The economic logic is consistent: AI triage cuts support headcount cost; investor pressure on margin makes that cut attractive. The result transfers cost from vendor to hotel.

What changed, vendor by vendor

I won’t pretend I have an exhaustive industry survey. I have what I’ve seen across hotels I run, hotels colleagues run, and the patterns the team has observed in the reviews we’ve published. Six platforms account for most of the visible shift.

Mews. When Anna tested Mews at her Stockholm property, she noted templated responses to support tickets. By mid-2025, the pattern crystallised. The default support layer is now an AI-first triage. To get a human reliably, you upgrade to a premium tier that operators on aggregator platforms describe as roughly a 70 percent upcharge. Our Mews review holds up; the rating is justified by product quality. But the support degradation is real and ongoing. The acquisitions Mews announced in 2025, of Flexkeeping in September and DataChat in October, compound the question. More categories of your data flow through one vendor whose default support response is now a bot.

Cloudbeds. Sophie tracked the maths in our Cloudbeds review and flagged the 890 percent commission-reporting bug. That bug is corroborated. What I’ve watched separately is the pattern of how Cloudbeds handles it. Operators on aggregator platforms repeatedly describe templated responses, tickets closed without resolution, and difficulty reaching anyone with authority. Sophie’s review picks this up. A $245 million capital structure with SoftBank involvement explains, but doesn’t justify, what happens to support quality when growth rates plateau and operating expenses come under scrutiny.

Guestline. James reviewed Guestline after the Access Group acquired it in July 2023. Our Guestline review called the post-acquisition trajectory; operator complaints since then have backed it up. The most common refrain on Trustpilot and similar surfaces is some version of “it used to be the best, now you can’t reach anyone.” Access bought SHR in 2024 too, deepening the conglomerate’s hold on UK hospitality back-office. The support degradation isn’t an accident of integration. It’s the visible edge of a different operating model.

RoomRaccoon. Elena reviewed RoomRaccoon with appropriate small-property warmth, and the product genuinely works for the right operator profile. But the support-and-billing pattern surfacing across operator forums goes further than our RoomRaccoon review captured. Cancellation friction, billing aggression, 48-hour expiring offers, support phone numbers that don’t connect: these are systematic, not incidental. The bootstrapped European positioning is real. The operator-relationship posture is not what the marketing implies.

Duve. Elena also reviewed Duve, where the issue is more about scale-vs-fit than support malice. Duve raised $60 million in Series B in December 2025 explicitly to scale into global enterprise. Their support is increasingly calibrated for chains, not boutique independents. Our Duve review caught this. The follow-on observation: as the platform scales toward chains, the small-property hotelier waiting for a Saturday-night reply gets reframed from “customer” into “queue position.”

HiJiffy. Our HiJiffy review (James again) flagged AI quality issues that the May 2025 Aplysia 3 release addresses to a meaningful degree. The other half of the story is support response time. Hoteliers on aggregator platforms increasingly mention same-day claims that don’t quite hold up under real conditions. HiJiffy is not the worst offender on this list. They’re on it because the pattern reaches them too.

Why this is happening

I don’t think the vendors are villains here. I think the economic logic is structural, and once you understand it, the pattern becomes inevitable rather than malicious.

Three forces converge.

The first is investor pressure on margin. Every venture-backed hotel-tech company hit 2023 with a different cost-of-capital reality than the one they were funded under. The capital is more expensive; the return expectations sharpened; the path to either an IPO or an acquisition requires a tighter operating model. Support is one of the largest variable cost lines in B2B SaaS. It’s also the line where automation has just become economically credible, which leads to the second force.

The second is that large language models reached a point in 2024 where AI-first support stopped being a wild bet and started being a manageable risk. Three years ago, replacing humans with AI on the support desk produced visible failure within a week. Today it produces a quieter, slower, more deniable failure: the bot answers, the ticket closes, the hotelier moves on or escalates through another channel. The vendor has plausible deniability. The hotelier still resolves the issue, just not through the contract.

The third is consolidation. Mews acquired Flexkeeping and DataChat in 2025. Access has been rolling up UK hospitality software for years. Canary acquired OpenKey in February 2026. Optii has been part of MCR since 2021. When ownership concentrates, the operator-vendor relationship becomes structurally less symmetrical. Your account team’s incentives are less aligned with your individual property’s experience than they used to be.

None of this is conspiracy. It’s the predictable shape of a maturing industry. But “predictable” is not the same as “acceptable”, and the hotelier who renewed last quarter has every right to be upset.

Why this matters more for hotels than for other B2B SaaS

I’d argue this matters more in hospitality than it does in most B2B categories. Three reasons.

Hotel software problems happen in real time, with guests present. A CRM outage at a marketing agency is annoying. A PMS outage at a hotel is people standing at reception unable to check in. A guest-comms platform that misroutes a message during high season costs you the relationship with a paying guest who’s about to write a review you’ll have to read for years.

Hotel staff are not technical. The receptionist on a Saturday night isn’t going to debug an integration. The supervisor at 06:30 isn’t going to read an AI-triage chatbot’s clarifying questions and rewrite the ticket. They need a human who knows the system to pick up the phone within minutes. When the contract says “24/7 support” and the actual experience is “AI triage, then ticket queue, then maybe Monday”, the operational consequences land on the hotel, not the vendor.

Hotel-vendor switching costs are extreme. PMS migrations take weeks. Channel manager reconfigurations take days. Guest-comms platform changes break message history and template approvals. The threat of switching, which is supposed to discipline vendor behaviour in healthy markets, is structurally weakened here. Vendors know this. They know how hard it is for you to leave. The support tier deterioration reflects that knowledge.

What hoteliers should actually do

I have specific suggestions, drawn from contracts I’ve renegotiated in the last six months at my own properties.

Read the renewal clause about support. Not the marketing page. The contract amendment. The number of operators I’ve spoken to in 2026 who genuinely didn’t know their support tier had changed is unsettling. The information was there. It was in the small text. Get someone on your team to read every annual renewal contract carefully, and flag any language change about response times, escalation paths, or premium tiers.

Ask the vendor, in writing, to specify the response time for a Saturday-evening outage. Get the answer in writing. If the answer is “best efforts”, the answer is no. Compare what they write to what your operation actually needs.

Negotiate the premium support tier into the base contract. Many vendors are willing to bundle it for a smaller upcharge than the published rate, especially during multi-year renewals. Especially if you have multiple properties. Don’t accept the AI-triage default if your operation can’t run on it.

Build a fallback plan. Document the steps your team takes when the platform’s support is unreachable. Which colleague at which other property can be called for a workaround. Which integrations can be re-routed temporarily. This is a runbook. Most hotels don’t have it. The vendors expecting you to absorb the support shift are betting that you don’t.

Ask the prospective vendors, during procurement, what their support headcount looks like. Ratio of support staff to customers. Geographic distribution of those staff. Hours when humans are actually on duty in your timezone. These are answerable questions. The vendors who refuse to answer them are telling you something.

Read our reviews for the support sub-score. It’s there in the new score-breakdown format. We score support on what we actually experienced during testing. The score isn’t always high. That’s the honest signal.

The deeper question

The pattern I’ve described isn’t a phase. It’s not going to reverse when AI support quality reaches some threshold. The economics that drove the shift in 2025 are still in place in 2026, and they’ll still be in place in 2027.

What might change is hotelier behaviour. If enough operators push back during renewal cycles, vendors that have over-rotated to AI-first support will find their churn climbing and their customer-acquisition cost expanding to compensate. The market mechanism does work here, but only if the price signal reaches the vendor through real lost contracts, not through polite complaints.

I run my hotel software vendor relationships like every other operator should: as commercial relationships where my money funds a specific set of obligations. The obligations are getting reduced; my money is not. That’s a renegotiation conversation, not a complaint letter.

I haven’t switched off any of the platforms on this list. Most of them still earn their place. But the contracts under which they earn that place are not the contracts I signed three years ago, and the hoteliers I’m hearing from haven’t fully reckoned with that yet.

Common questions

Which hotel software vendors degraded support in 2025?
The six platforms we've observed showing visible support-tier degradation are Mews, Cloudbeds, Guestline, RoomRaccoon, Duve, and HiJiffy. The patterns vary (AI-first triage, templated responses, premium tier pricing, reduced human availability), but the direction is consistent: human support became more expensive, harder to access, or both.
Is AI-first support really worse than human support?
For routine queries (account access, basic configuration, documentation lookups), AI triage is often adequate and sometimes faster. For real-time operational problems with guests present, AI triage is meaningfully worse than reaching a human who knows the platform. The mistake is treating all support requests as equivalent.
Why did this happen in 2024 and 2025 specifically?
Three forces converged. Investor pressure on margins increased after capital costs rose in 2023. Large language models reached a point where AI-first support stopped failing visibly. Industry consolidation reduced the operator-vendor relationship symmetry. The combination made support automation economically attractive and structurally acceptable to vendors at the same time.
What should hoteliers do about it?
Read renewal contracts carefully for support-tier language changes. Negotiate premium support into the base contract during multi-year renewals. Build a documented fallback runbook for when vendor support is unreachable. During procurement, ask vendors directly about support headcount, geographic coverage, and timezone availability. Refusal to answer is itself an answer.
Is this pattern reversible?
Not by waiting. The economics that drove the shift in 2025 remain in place. What might change is hotelier behaviour. Vendors that over-rotate to AI-first support will see churn climb if operators push back during renewals. The market mechanism works, but only if the price signal reaches the vendor through actual lost contracts.

What I’d tell a colleague

If your contract is up for renewal in the next six months, treat the support tier as the most negotiable item on the page. It is the thing the vendor has changed silently in the last year. They will expect you to ask about price. They will not expect you to ask about response time. Ask anyway.

If your contract is locked in for another year or two, build the fallback runbook now. The vendor’s support degradation is not your operational excuse when something breaks during a busy weekend. Your guests don’t care about your vendor’s investor pressure.

And read the support sub-scores in our individual reviews. We test this dimension specifically because the vendors no longer advertise it honestly. The score we publish reflects what we experienced when we needed a human, not what the marketing page promised.

Marc, with input from Anna, Sophie, Elena, James, and Thomas