Why this comparison copies no prices
Five third-party comparison sites moved into this category between July and August, and one of them attributes to a vendor a rate that cannot be found on that vendor’s own site. That is the structural problem with the exercise: a copied price is wrong the day it changes, and it stays wrong for a long time, in the page and in the cache of the engines that indexed it. A reader who finds the gap at signing time does not blame the comparison site, they blame whoever sent them there.
We made the mistake ourselves, internally. Our own survey of 26 August carried our own price list, and that list had changed the same morning: the document lived a week with a stale figure on the line we knew best. If that happens to us about ourselves, it happens to anyone about others.
So this page compares what does not go stale: what is sold, in what unit, with what entry fees, what trial, what languages and what tools connected. The rates are on everyone’s official pages, they are up to date there, and that is where to read them. This comparison tells you what to look at alongside the price, which is precisely what the price alone will never tell you.
The unit sold, the only comparison that really holds
Two offers cannot be compared on their price lists until they sell the same thing. A minute, a call, a user and an agent are four different units, and moving from one to another means knowing the shape of your calls, which you only know afterwards. That is why two neighbouring businesses can choose opposite offers without either being wrong.
Per-call billing is the one that surprises most often. It counts every pick-up as one, whatever its length: a thirty-second enquiry weighs as much as a twelve-minute booking. On a small-business reception, where half the incoming calls are short, the gap between what you imagine paying and what you pay opens up in the first month.
Unlimited raises the opposite question, and it is just as counter-intuitive. With no meter, the price of a minute does not exist before the first statement: it falls as usage rises. An unmetered offer is expensive for light use and cheap for heavy use, and the crossover point can only be worked out with your own volumes.
| Vendor | What is sold | What cannot be converted before signing |
|---|---|---|
| Cantya | Minutes | Nothing: the minute used is readable on the invoice |
| Sylen | Minutes | Nothing either, but the entry bundle is short |
| AirAgent | Calls | A thirty-second enquiry weighs as much as a twelve-minute booking |
| Répondeo | Users, unlimited calls | The price of a minute, which depends on usage and therefore does not exist before the first month |
| Vokai | Agents | How many calls one agent absorbs before a second is needed |
| Nerolia, Versatik | On request | Everything, until the proposal arrives |
| Districall, Open6 | A telecoms bundle, AI as an option | What the option covers, and what stays on the meter |
Recorded on 26 August 2026 from the official pages. Fonio sells minutes but does not address the French market.
Entry fees, the line that weighs most at signing time
This is the least compared line and one of the heaviest. It is paid once, it appears on no monthly price list, and yet it represents several months of subscription at the vendor who charges the most for it. A gap on entry weighs more, at the moment of decision, than a few points of difference on the monthly.
Three postures coexist in this market. No fees, stated as such. Fees quoted and published, which is honest even when the figure surprises. And the « depending on complexity » or « depending on configuration » formula, which is not a price but a promise of a conversation. None of the three is illegitimate; what is, is not knowing which one you are in before the first sales meeting.
The question to ask runs to one sentence, and it deserves to be asked in writing: what do I pay before the first call is handled? Set-up, configuration, number porting, training. A complete answer runs to three lines at a vendor who has costed them.
| Vendor | Set-up fees | Is the figure public |
|---|---|---|
| Cantya | None when self-served; full installation on quote | Yes for self-service; on quote otherwise |
| AirAgent | Charged, from a floor | Yes, on their pricing page |
| Vokai | « Depending on complexity » | No |
| Nerolia | « Depending on configuration » | No |
| Sylen, Répondeo, Versatik | Not mentioned | No |
« Not mentioned » does not mean « none ». It means the question comes up at the first meeting, and is better asked before.
The free trial has become the norm, and we do not have one
This is the clearest movement of the summer in this category. Three vendors out of nine now offer a trial, one of them without a card, and use it as their main argument. Two offer nothing at all: AirAgent and us.
A trial matters because it shifts the burden of proof. A phone reception is judged on real calls, not on a rehearsed demonstration, and nobody knows in advance how their own callers will react. A demonstration shows what the product can do; a trial shows what it does at your place, which is not the same information.
If you get no trial, the fallback question is about the exit: what commitment, what notice, and what becomes of your number if you leave. A vendor who gives no trial but commits you to nothing remains comparable to one who gives a trial and then ties you in for a year.
| Vendor | Free trial | Card required |
|---|---|---|
| Sylen | Fourteen days | No |
| Vokai | Seven days | Not published |
| Districall | Seven days | Not published |
| Nerolia | Half-hour demonstration | Not applicable |
| Cantya | Demonstration only | Not applicable |
| AirAgent | None | Not applicable |
This is our most visible blind spot. It is written here because it can be checked in three clicks at our competitors, and a comparison that hid it would be worth nothing.
Languages, and what « handling » means on the phone
Announcing languages and following a caller who switches language mid-sentence are two different things, and the second is the only one that counts on the phone. A caller does not warn you they are about to switch to English; they do it, and the reception either follows or does not.
Three of the nine vendors recorded are French only. Two announce several languages without saying what happens when it changes in the middle of the exchange. One removed from its site the multilingual claim it carried in July, which is more honest than repeating it without having held it.
The right way to check goes through no brochure: ask for a demonstration, start in French, switch to another language in the middle of a sentence, and listen to whether the reception comments on the change, asks about it, or follows without saying anything. All three behaviours exist, and the third is the only one that does not break the conversation.
| Vendor | Languages announced | Follows a switch mid-call |
|---|---|---|
| Cantya | Six | Yes, from the very next line and without commenting on it |
| Vokai | Nine announced | Not stated |
| Fonio | German, English | Not stated |
| Sylen, Répondeo, Nerolia | French | Not applicable |
| Versatik | Not published | Not published |
The « more than thirty languages » attributed to Versatik in July no longer appears on its site: removed rather than repeated.
Trade tools, the real battleground of the year
This is the underlying movement of 2026, and it is more structural than any gap on price. A reception that books appointments has to write into the tool where the diary actually lives, and for many trades that is not a general-purpose calendar. A restaurant keeps its floor plan in booking software, a salon keeps its schedule in a trade-specific booking tool, a hotel in its reservation engine.
One vendor in the panel built its position on this and connects around twenty trade tools across restaurants, beauty and hospitality. The others, ourselves included, stay on general-purpose calendars and automations. The difference is not cosmetic: plugging a generic calendar into a restaurant amounts to asking the restaurateur to copy their bookings across by hand.
It is also the question to ask first if you work in one of those trades, before price and before languages. An unconnected booking tool is not compensated for by a generic integration, and it does not get added in a week.
Three questions to ask in this order, and the third is the one that settles it.
- 1 Which tool do my appointments actually live in today? Not the one I could use, the one I open in the morning.
- 2 Is that precise tool connected, by name, at this vendor? A list of categories does not answer the question, only the name of the software does.
- 3 If the answer is no, what do I do with the appointments in the meantime? If the answer is « I copy them across », the rest of the comparison no longer matters.
Where Cantya loses, and to whom
A comparison published by a vendor is only worth what it concedes, and the concessions have to be checkable in three clicks at the competitor, otherwise they are just decorative modesty. Here are ours, in the order of what they cost a buyer.
None of them is a hidden defect: they are functions others have and we do not have yet. They will be closed, some quickly, one slowly. We write them because you would find them anyway, and it is better that it happens before you call us than after you sign.
What we hold against them, and this is checkable too: a billing unit that is readable on the invoice, no entry fees, a public price and online sign-up, and a reception that follows the caller’s language when it changes mid-call.
Our blind spots, in order of how much they cost a buyer.
- No trade booking tools. That is the heaviest gap and the longest to close. One vendor in the panel connects around twenty; we connect general-purpose calendars.
- No free trial. Three vendors out of nine offer one, one of them without a card, and use it as their main argument.
- The number of integrations shown. Two vendors claim more, often through a universal intermediary. The raw comparison plays against us, but the only question that decides is whether your own tool is on the list, and how it behaves once connected.
- Fewer integrations on display than the two vendors who claim the most. The raw comparison works against us, even if the question that matters remains the tool you actually use.
How this comparison was made, and how to check it
The method matters as much as the table, because it is what lets you redo the work without taking our word for it. The nine French vendors were recorded on the same day, on their own pages, including the ones a previous survey had carried forward without checking.
Three lines had changed since the July survey: one vendor moved from metered to unlimited, another increased the number of languages announced, a third removed a multilingual claim. A survey more than two months old in this category should therefore be treated as stale, this one included when its day comes.
What was not done, and has to be said: the American line and one German-speaking vendor come from earlier surveys and were not rechecked, because neither sells in France. Two players sell a telecoms bundle with an AI option; they set a perceived price floor but they are not product rivals.
The five rules this comparison holds itself to. They are also the grid for judging any other comparison in this category.
- 1 One source per offer: the vendor’s official page. Never a third-party comparison site, one of which publishes a rate that cannot be found at the vendor it describes.
- 2 Every player recorded on the same day, otherwise you are comparing different states of the market.
- 3 What is not published is written « not published », never estimated.
- 4 No price copied: it would go stale in the page and in the engines’ cache.
- 5 The author’s blind spots are written down, and checkable at the competitor who closes them.
Common questions
- Why not show everyone’s prices in a table?
- Because a copied price outlives its own change, in the page and in the cache of the engines that indexed it. We lived it on our own price list, which had changed on the morning of the survey. The rates are on the official pages, they are up to date there, and this page tells you what to look at alongside them.
- How do you spot a comparison you cannot trust?
- Three signals are enough. No survey date. No « not published » line, as if everything were known. And an author who never loses. A fourth, quieter one: prices that cannot be found on the page of the vendor cited.
- Should you choose a vendor specialised in your trade?
- It depends on one point: if your appointments live in trade software rather than in a general-purpose calendar, then yes, and the question comes before price. Otherwise, specialisation mostly brings vocabulary, which is a matter of configuration.
- Is a vendor who does not publish prices necessarily more expensive?
- Not necessarily, but they cost you time and they move the comparison into a sales meeting. Ask for a range in writing before that meeting: a vendor who refuses to give one is telling you something about the rest of the relationship.
