A client asks you to generate AI B-roll for a corporate edit. You open Runway. The subscription is $76/month. What do you quote?
You have no idea. Neither does anyone else in your position. The tool vendors publish subscription pricing. They do not publish what it actually costs to deliver a usable B-roll package to a client. That number is different, and if you quote the wrong one, you lose money on the project.
This is happening across the industry right now. Freelance editors and agency producers are getting asked to price AI-generated B-roll, and the only reference points are subscription pages that bear no relationship to actual production cost.
The tool costs $76/month. What you quote the client should be based on what the project actually costs you, not what the subscription page says.
The Subscription Price Is Not the Production Cost
Here is what the subscription pages advertise:
| Tool | Advertised Price | What You Get |
|---|---|---|
| Runway Gen-3 Turbo | $76/month | 625 seconds of generation |
| Kling Creator | $7.99/pack | 13-19 generations depending on resolution |
| Veo 3 | $0.15-0.40/second | Per-second API pricing |
If you look at these numbers and think “AI B-roll is almost free,” you will quote too low, burn through your hours, and eat the difference.
The subscription gets you access to the tool. It does not account for the four cost components that make up the real production price: iteration, QA, post-production, and learning curve.
Where the Real Cost Lives
Iteration (the 75% discard rate)
Users report that three out of four AI video generations are unusable for client-facing output. The reasons: character inconsistency, physics errors, lighting that does not match the edit, content filter rejections, or output that simply does not fit the brief.
Each failed generation costs the same as a successful one. The API charges per second of generation, not per second of usable output. If you need 30 seconds of clean B-roll and your discard rate is 75%, you are generating 120 seconds to get 30. The tool does not refund the other 90.
One user reported burning $1,000 in eight days learning to prompt Veo 3. That is not an outlier. That is the iteration tax at work on a new tool with no established production workflow.
QA (the review time nobody quotes)
Every generation must be reviewed against the brief. Does the lighting match the rest of the edit? Is the motion natural or does it have the AI drift artifacts? Does the content clear brand safety? For a 30-second B-roll package with 6-8 shots, that is 6-8 review decisions per generation round. At three rounds of iteration, that is 18-24 review decisions before you have a usable set.
Your hourly rate applies to every one of those decisions. If you charge $75/hour and spend 2 hours on QA, that is $150 in review time for a 30-second package. The tool subscription did not include that.
Post-Production (the finishing work)
AI-generated B-roll is raw footage. It needs color grading to match the rest of the edit, sound design if applicable, and assembly into the timeline. If the client is using the B-roll in a branded piece, you may need to do visual effects cleanup on AI artifacts.
This work is the same whether the source footage came from a camera or a generation model. The difference is that traditional B-roll from a shoot is usually lit and graded consistently. AI-generated footage varies shot to shot, which can increase the grading time.
Learning Curve (the first-project tax)
If this is your first AI B-roll project, or your first with a specific tool, expect 8-20 hours of learning time before you are generating usable output efficiently. That is education cost. It amortizes across future projects, but it exists, and it hits hardest on the first client engagement.
The Real Cost Formula
Here is the formula that produces an honest quote:
True project cost =
(tool cost for total generation seconds)
+ (iteration hours x your hourly rate)
+ (QA hours x your hourly rate)
+ (post-production hours x your hourly rate)
True per-second cost = total project cost / deliverable secondsFor a 30-second B-roll package, experienced operator:
| Cost Component | Calculation | Amount |
|---|---|---|
| Raw generation (120s at $0.12-0.22/s, accounting for 75% discard) | 120 x $0.17 avg | $20 |
| Iteration time (1.5 hours at $75/hr) | Prompt refinement, re-generation | $112 |
| QA and review (1.5 hours at $75/hr) | Shot selection, brief compliance | $112 |
| Post-production (1-2 hours at $75/hr) | Color, assembly, cleanup | $112 |
| Total | $356 | |
| Per deliverable second | $356 / 30s | $11.87/s |
For a first-time operator, add 8-20 hours of learning curve. That moves the total to $950-1,850 for the same 30-second package.
The advertised cost of 30 seconds of AI B-roll is $7-15. The real cost for an experienced operator is $350-500. For a first-time operator, it can exceed $1,000.
What To Quote the Client
The client does not need to see your iteration math. They need a project price that covers your actual cost and margin. Here is how to structure the quote:
Option A: Per-project pricing. Quote a flat fee for the deliverable. “30 seconds of AI-generated B-roll, 6-8 shots, color-matched to your edit, delivered in 5 business days: $500-800.” This is the cleanest structure for both sides. The client knows the cost. You absorb the iteration variance and price it into the fee.
Option B: Hourly plus tool cost. Quote your hourly rate for production time, plus tool costs as a pass-through. This is transparent but harder for clients to budget against because the total depends on iteration.
Option C: Tiered packages. Offer packages by deliverable length and complexity. 15 seconds / 30 seconds / 60 seconds, with defined shot counts and revision rounds at each tier.
The structure matters less than the honesty. If you quote $100 for 30 seconds of B-roll because the tool subscription costs $76/month, you will lose money. If the client’s budget cannot absorb a realistic quote, scope the project differently: fewer shots, lower complexity, or a clear conversation about what AI B-roll costs in practice versus what the marketing pages suggest.
The Decision Framework
Before quoting any AI B-roll project, answer these four questions:
- What is the discard tolerance? Internal use tolerates higher discard. Client-facing brand work requires more iteration. Price accordingly.
- Is this your first project with this tool? If yes, factor in learning curve. Do not eat that cost and pretend it does not exist.
- Does the output need to match existing footage? Matching lighting, color, and motion style to a traditional edit increases post-production time.
- How many revision rounds does the client expect? Each round multiplies QA and iteration. Define this in the scope.
If you are pricing AI B-roll for the first time, the single most important thing is to track your actual hours on the first project. Compare your tracked time against your quote. Adjust for the next project. The gap between advertised tool cost and real production cost closes with experience, but it never closes to zero.
The tool is a material cost. Your time is the production cost. Quote the production cost, not the material cost.
For production teams that want predictable project pricing without managing the iteration math, the capabilities page shows how managed production handles cost variance so the client gets a fixed deliverable price regardless of how many generations it takes.