There is no typical number, and the ranges you will find quoted online are close to meaningless, because the category is not billed on one meter. Two teams with identical budgets can get quotes an order of magnitude apart for the same capability if one has a large static archive and the other has a small one that grows every week. Before you can estimate anything, you need to know which meter you would be billed on, since that decides the shape of the bill far more than the rate does.
The meters that exist
Volume processed is the most common: you pay by hours of footage ingested and indexed. Library size under index is a variant, where you pay for how much material stays searchable rather than how much arrived this month. Seats bill by how many people can run a search. Query or API volume bills by use, which behaves very differently depending on whether five people search occasionally or a system queries continuously. On top of any of these there is often a platform fee that exists regardless of usage.
Most quotes combine two or three of these. The one that matters is whichever tracks the thing you have a lot of, and that is the one to model.
Numbers to bring before you ask anyone
Four figures make an estimate possible. How many hours are already in the archive. How many hours arrive per month. How many people will genuinely run searches, which is nearly always fewer than the number who say they want access. And how much of the archive actually needs to be searchable, because most teams discover that a large fraction of what they store has no realistic chance of being reused.
That last figure is the one that moves budgets. Indexing everything is the default assumption and it is rarely the right one. Deciding what stays out is a cheaper decision than any negotiation, and the same reasoning applies to comparing options on total cost rather than on headline rate.
The costs that are not on the quote
Getting material in. If a system requires footage to be uploaded or migrated, you are paying for transfer, for a second copy of a large archive, and for someone's time supervising it. Vivu does require an upload, so this line applies to it too: footage goes into a Vivu project and is indexed once there, which makes the amount you choose to upload the part of this cost you control. Ask every vendor on your list about it early, because it can be larger than the license.
Reprocessing. Models improve, and when they do, someone has to decide whether the archive gets re-indexed. Ask what triggers that and who pays for it.
Integration. A search tool that does not reach the place people already work becomes a tab nobody opens, and the work of connecting it is real even when the software is straightforward.
The processing bill underneath. If the system transcribes what it ingests, there is a per-hour cost in there whether it is itemized or not, and the per-hour arithmetic of transcription is worth doing separately so you can see what the rest of the platform is actually charging for.
When you should not be buying this
If your archive is small enough that one person knows where everything is, you do not have a search problem. If two people request footage a month, the manual route is cheaper than any license, and it will stay cheaper for a while. If your material is well organized and turnover is low, the file names are still doing their job.
The purchase starts making sense when the people who knew the archive have left, when the volume passes what any individual holds in their head, or when the requests come from outside the team and arrive faster than someone can answer them. That inflection is about the organization, not the size of the library, and it is why comparing yourself to a benchmark team's spend tells you very little. What the tools in this category actually replace, and therefore what you are buying, is laid out in the breakdown of what each type of tool does.
How to read your own situation
Work out your four numbers, then ask each vendor which meter they bill on and what happens when the number that describes you doubles. If the price scales with something you plan to grow, you are buying a liability alongside the software. If it scales with something stable, the quote you receive now is roughly the quote you will be paying in two years.
If nobody in the organization can produce those four numbers, that is the useful finding. A team that does not know how many hours it holds or how often footage is requested is not yet in a position to evaluate a quote, and the fix for that is a week of counting, not a procurement process.
FAQ
How do I estimate the cost before talking to a vendor?
Build the estimate from your own numbers, since the meters vary too much for a published figure to transfer. Count the hours in the archive, the hours arriving per month, the number of people who will really search, and the share of the archive that needs to be searchable at all. With those four figures you can evaluate any quote structure quickly, and you can tell immediately which of them a given quote is sensitive to.
What should I ask about how new footage gets billed?
Ask what happens when material arrives, specifically whether it is billed at the moment it is indexed, on an ongoing basis for staying searchable, or both. Then ask the same question about material you delete or archive. A team that produces a lot each month and keeps everything forever has a very different long-run bill from one that produces the same volume and retires it after a year, and this is the question that separates them.
Is a proof of concept worth paying for?
Often, if it runs on your own footage rather than on a prepared demo set. The thing you are testing is not whether search works in general, it is whether it works on your material: your shooting conditions, your naming conventions, your terminology, the parts of the archive with bad audio. A trial on someone else's sample library tells you almost nothing about how it will perform on yours.
Does the price include the storage the footage lives in?
Not usually, and it is worth checking whether it duplicates storage you already pay for. Some approaches require material to be copied into the vendor's environment, which means paying twice for the same footage plus the transfer. Others index material where it already sits. This is a structural difference rather than a pricing detail, and at archive scale it can be a larger line than the license itself.