When we set AIVA's pricing, the default was obvious: pick a few tiers, put a monthly number on each, add some seat limits, ship a pricing page that looks like everyone else's. We got about three weeks into designing that page before we stopped and asked a more basic question: what is a monthly fee actually for?
What a monthly fee is really pricing
A subscription fee isn't really a price for usage. It's a price for access — the right to use the product, whether you use it a little or a lot that month. The vendor likes this because it makes revenue predictable: the same invoice in a slow month and a busy month, easy to forecast, easy to build a sales team around.
The customer's experience is the mirror image. In a slow month, you're overpaying for capacity you didn't touch. In a busy month — the one where the product is actually earning its keep — you're not paying any more for the extra value you got. Subscriptions convert your business's natural variation into either waste or a ceiling, depending on which side of average you land on.
For a call-and-chat-handling product like AIVA, that mismatch is sharper than usual. A phone line doesn't ring on a fixed schedule. A clinic gets forty calls a day in flu season and half that in a quiet month. A salon's phone barely rings on a Tuesday and doesn't stop on the Saturday before wedding season. Whatever number we put on a monthly plan would be wrong for most months of most businesses' years.
The seat problem
There was a second issue, specific to what AIVA is. Standard SaaS pricing scales with seats — the number of humans logging in. AIVA doesn't have that unit. Nobody at your business logs into AIVA to answer a call; AIVA answers the call. Pricing by seat would mean pricing by something that has nothing to do with what the product does. We'd have been inventing a number to make the pricing page look familiar, not because it meant anything.
A seat count measures how many people use a tool. It doesn't measure how much work the tool did. For AIVA, only the second number means anything.
What we built instead
AIVA is pure pay-as-you-go. No monthly fee, no per-seat charge, one shared balance that covers every channel — voice at ₹4 a minute, web chat at ₹2 a conversation, SMS at ₹1 a message. You top the balance up in packs, and it draws down only when AIVA actually does something: answers a call, handles a chat, sends a message. This isn't a novel structure we invented — it's how usage-based telephony infrastructure already prices itself; Twilio's own per-minute and per-message rates work the same way underneath a lot of the software layered on top of it. We just held the same logic at the product layer instead of wrapping it in a flat subscription once it reached the customer. A quiet week costs you almost nothing. A week where your phone won't stop ringing costs you proportionally more — but it's also the week AIVA is generating the most value for you, so that's the correct week for the bill to be bigger.
How the balance and recharge actually work
Mechanically, it's closer to a prepaid mobile connection than a SaaS subscription. A new account starts with ₹500 in free credit, and from there, any top-up — ₹500, ₹2,000, ₹5,000, or a custom amount — sits as a balance that draws down per call, per chat, and per message as they happen. There's no invoice waiting at the end of the month for usage that already happened; the balance is visible in the dashboard in near-real time, and a business can watch it move the same way it would watch a prepaid data balance.
The one risk this model introduces that a postpaid subscription doesn't — running out of balance mid-month and having calls go unanswered — is exactly why auto-recharge exists: a threshold a business sets itself, below which the balance tops up automatically from a saved payment method, so a genuinely busy week doesn't turn into a dropped call just because nobody checked the dashboard in time. It's opt-in, not a default a business has to discover the hard way.
Where the two models actually cross over
"Usage-based is fairer" is an easy thing to assert and a more useful thing to actually work out. So say a hypothetical flat-fee competitor charged a fixed ₹12,000 a month, all-inclusive, no matter how much you used it — a made-up number, purely to make the comparison concrete. At AIVA's blended rates, a business would need to run up a genuinely large volume before its usage-based bill caught up to that flat ₹12,000 — a clinic doing 1,000-1,200 calls a month at two to three minutes each, plus a normal amount of chat and SMS, is still comfortably under it most months.
Now run the same hypothetical fee against a slow month. Say a wedding venue in the off-season runs maybe 60 calls a month at three minutes each — call it ₹720 worth of actual voice usage on AIVA's rates. Under the flat-fee model, that business still pays the full ₹12,000 for roughly ₹720 of real activity. That's not a rounding difference. That's the entire structural argument for usage pricing, expressed in one number: the flat-fee customer is subsidizing capacity nobody used that month, and the usage customer pays for exactly what happened, in the slow month and the busy one alike.
A second worked example: a salon's Tuesday versus Saturday
The clinic and wedding-venue examples above both compare month to month. It's worth zooming into a shorter cycle too, since a lot of small businesses feel their real volume swing week to week rather than month to month. A single-chair salon might field 15 calls on a quiet Tuesday, each running about a minute and a half — call it ₹90 of voice usage for the day. That same salon's Saturday, the day every salon's busiest, might see 60 calls at the same average length — around ₹360 for the day.
Under AIVA's pricing, that difference just happens automatically: Tuesday costs a quarter of what Saturday costs, because Tuesday did a quarter of the work. Under a flat monthly plan, that same salon pays an identical seventh of its monthly fee whether the calendar shows a Tuesday or a Saturday, which means every quiet Tuesday is quietly overpriced to make the plan's math work on the Saturdays. Weekly swings like this rarely get modeled explicitly when a business picks a plan — they just show up later as a nagging sense that the flat fee never quite matched what was actually happening on the phone.
The honest edge case — where a flat rate would actually win
It would be dishonest to pretend usage-based pricing wins at every volume, so let's say plainly where it doesn't. A business running enormous, highly consistent volume every single month — a large multi-location chain answering tens of thousands of minutes a month, every month, with almost no seasonal variation — could, in principle, negotiate a flat rate lower than what standard per-unit pricing would charge for that same volume. That's not a flaw in the logic; it's the same logic working in the other direction. At high enough guaranteed volume, a vendor can afford to price access rather than usage, because the usage is no longer in question.
That's exactly the case custom pricing and invoicing exists for. We don't pretend every business should stay on the standard per-unit rate forever — past a certain volume, talking to us about a negotiated rate is the more honest answer than sticking to ₹4 a minute out of principle. The difference between us and a subscription-first vendor isn't that we never do flat pricing. It's that we don't default to it for businesses it doesn't fit, which is most businesses, most of the time.
Why most SaaS still defaults to subscriptions anyway
It's worth being fair to the rest of the software industry here, since defaulting to subscriptions isn't purely a lazy choice on other vendors' part — it solves a real problem for the vendor, even when it solves the wrong problem for the customer. A subscription creates predictable monthly recurring revenue, which is the metric investors and acquirers actually price a software company on. A vendor with steady MRR can forecast headcount, raise money more easily, and report growth in a single clean number every board meeting. Usage-based revenue is real revenue too, but it's lumpier and harder to project a quarter out, which makes it a less comfortable number for a vendor optimizing for how their own business gets valued.
That's a legitimate business reason to prefer subscriptions — it's just a reason that serves the vendor's fundraising and reporting needs, not necessarily the customer's actual usage pattern. We're bootstrapped, with no outside investors to report an MRR number to, which removes the main reason a vendor would default to subscription pricing even when it's the wrong shape for a given customer's month. That's less a virtue than a structural fact about who we answer to, but it's the honest reason usage-based pricing was ever an option for us in a way it might genuinely not be for a venture-backed competitor under pressure to show smooth, predictable growth every quarter.
Doesn't usage-based pricing make budgeting harder?
This is the most common pushback from a business owner used to a fixed monthly line item, and it deserves a straight answer rather than a dismissal. A flat fee is genuinely easier to drop into a budget spreadsheet once and forget about. Usage-based pricing asks a business to actually look at a number that moves, which is more attention than a fixed cost demands.
The honest response is that the variability is usually smaller and more predictable than it sounds before you've seen it — most small businesses' call volume doesn't swing wildly week to week outside of an identifiable season or event, which means a few months of real data in the dashboard usually gives a business a reliable range to budget against, not a genuinely unpredictable number. And the alternative isn't actually more predictable, it just hides the variability inside a number that's wrong in both directions — a flat fee that's "predictable" every month is still overcharging in the slow ones and potentially undercharging relative to value in the busy ones. Predictable and accurate aren't the same thing, and it's worth being honest that usage-based pricing trades a little of the first for a lot more of the second.
Testing this before committing to anything
Because there's no plan to sign up for, there's also nothing to commit to before you know whether the model works for you. Every new account starts with ₹500 in free credit, no card required — enough for a genuine sample of real calls, chats, and messages, not a feature-capped demo. You can watch exactly how your own business's usage pattern behaves against these rates before a single rupee changes hands. Start a free account and the first real invoice you'll ever see is built from your own numbers, not from a rate card you had to guess against in advance.
What this changes about the relationship
The honest reason this pricing model works for us long-term is that it removes a lever we'd otherwise be tempted to pull. Under a subscription, our growth comes from adding accounts and raising tiers, whether or not the product is doing more for any single customer that month. Under usage pricing, our revenue only grows when AIVA is handling more real conversations for people who are already customers. If AIVA gets worse — slower, less accurate, escalating things it shouldn't — usage drops and so does what we're paid. That's a healthier incentive than a subscription renewal date.
It also changes what a sales conversation with us looks like, if you ever have one. We have no reason to push you onto a bigger tier than you need, because there is no tier — only your actual usage, which we'd rather see go up because you're getting more calls answered well, not because we talked you into overprovisioning.
If you're moving off a subscription tool
If you're currently running a subscription-priced answering service or IVR and thinking about switching, the useful exercise isn't comparing our per-minute rate to your current plan's sticker price — it's comparing it to what you actually paid last month, including whatever sits under the base plan. We've written a longer breakdown of what usually hides under an answering service's advertised rate — overage minutes, setup fees, after-hours surcharges — that's worth reading before you run that comparison, because the sticker price is rarely the number that matters. Businesses that make the switch are almost always surprised by which number turns out to be bigger, and it's usually not the one they expected going in.
We didn't set out to build a pricing philosophy. We set out to avoid billing people for a number that didn't mean anything. The rest followed from there — down to the pricing page itself, which still doesn't have a single tier on it.