How to Compare Two AI Production Bids When the Deliverables Are Described Differently

Two quotes arrive for the same series. One is priced per episode and lists a revision allowance. The other is priced per finished minute and describes revisions as reasonable and included. One promises delivery of graded masters with separated audio stems. The other promises finished episodes. The difference between the two numbers is thirty percent, and there is no honest way to tell whether that gap represents a better price or a smaller scope, because the two documents are not describing the same thing in the same language.

This is the normal condition of AI production procurement rather than an edge case. The category is young enough that no standard deliverable vocabulary exists, so every production company describes its work in the terms that suit its own workflow. A commissioning team that compares the bottom line numbers is not comparing bids. It is comparing two different guesses about what the series will need, one of which is probably optimistic and neither of which is wrong on its own terms. The work is to rebuild both documents onto a shared spine first, and only then to look at price.

Why Two Bids for the Same Series Are Rarely Comparable

A bid is a description of a scope, and scope in AI production has more degrees of freedom than commissioning teams expect. Episode count is fixed and everything downstream of it is negotiable: how long an episode runs, how many shots make up a minute, what condition a shot has to be in before it counts as delivered, how many passes of change are included, what happens to the character assets afterwards, and what the production company owes you when something goes wrong. Two companies can both bid honestly on the same brief and differ by a factor on any one of those.

The gap widens when the brief itself is loose. A request for proposal that specifies the story and the episode count but not the delivery specification invites each bidder to supply their own assumptions, and those assumptions are exactly where the price difference lives. The cheaper bid is frequently cheaper because it assumed fewer revision cycles or a lighter finishing standard, not because the company is more efficient. The expensive bid is sometimes expensive because it read the brief more carefully and priced the work the brief actually implies.

None of this is detectable from the totals. It is detectable from a structured read of both documents against the same set of questions, which is what the rest of this piece sets out. The method takes a commissioning team about two hours per bid and it is the difference between a decision and a coin flip.

1. Normalise the Unit of Delivery

Start by converting both bids to the same denominator. If one is priced per episode and the other per finished minute, you need the episode runtime from both to reconcile them, and you should ask for it in writing rather than inferring it. An episode is not a fixed quantity in vertical drama. A ninety second episode and a two minute twenty episode differ by more than fifty percent in content volume, and a bid priced per episode against an assumed shorter runtime will look cheap until the runtime is specified.

Then go one level below runtime, to shot count. Shot count is the real unit of production work, because it drives generation volume, review volume and edit volume together. Ask both bidders how many distinct shots they have assumed per episode and what happens to the price if the delivered edit needs more. A company that can answer that question precisely has modelled the work. A company that treats shot count as an implementation detail has priced a feeling.

Write both bids out on a per shot basis. The number will look unfamiliar and it is not the number you will negotiate on, but it is the only figure that makes two differently structured quotes sit next to each other honestly. In most head to head comparisons, the per shot conversion alone closes half of the apparent price gap.

2. Normalise the Revision Boundary

Revisions are where the largest hidden cost differences sit, and the language around them is almost always vague on purpose. Reasonable revisions included is not a commercial term. It is a promise that will be interpreted generously in month one and narrowly in month three, and the interpretation that matters is the one that applies when you are unhappy with episode 41 and the schedule is tight.

Ask both bidders for three specific things. How many revision passes are included per episode. What defines a pass, meaning whether a pass is a round of consolidated notes or a single change request. And what specifically falls outside the allowance, which is usually a change of direction rather than a correction of execution. The distinction between a correction and a redirection is the one that generates disputes, and a bidder who has a clear written line there has run productions before.

Then price the excess. A bid with four included passes at a stated overage rate is comparable to a bid with two included passes at a lower headline price, and often turns out to be cheaper once you assume a realistic note volume for a first commission. Assume more revision than you expect to need. First commissions run heavier on notes than subsequent ones, because the house style is still being established.

3. Normalise What Counts as Finished

Finished is the least standardised word in any AI production bid. For one company it means colour graded episodes with mixed audio, burned in subtitles in the commissioning language and platform ready encodes. For another it means assembled episodes with a temp mix, on the reasonable assumption that the platform will finish to its own specification. Both are defensible positions and the difference between them is a real budget line that somebody has to carry.

Build a delivery specification of your own and send it to both bidders rather than trying to reverse engineer theirs. It does not need to be long. Container and codec, resolution and frame rate, audio configuration and whether stems are separated, subtitle format and languages, thumbnail and key art, and whether the episode arrives platform encoded or as a mezzanine master. Ask both bidders to confirm inclusion or exclusion line by line against that list.

The exercise has a second benefit beyond comparability. It forces your own team to decide what you actually need before the production is underway, which is the cheapest moment to decide it. Delivery specification changes agreed after generation has begun are expensive in every production model, and they are the most common source of unbudgeted cost on a first commission.

4. Normalise Who Holds the Assets

Ask both bidders what you receive besides the episodes. Character reference sets, trained models or adapters where they were used, the prompt library, the style guide, the continuity bible, the project files. These are the assets that determine whether a second season is a continuation or a restart, and whether you can move the work to another partner without losing visual continuity. Their commercial value is significant and it is frequently not mentioned in either bid.

Two bids can differ substantially in price while differing more substantially in what they leave you holding afterwards. A bid that delivers episodes and retains everything else is a licensing arrangement described as a production arrangement. A bid that transfers the full asset set is selling you an ongoing capability, and it is reasonable for that to cost more. Neither is wrong, but you should know which one you are buying before you compare the numbers.

This is the point where total cost of ownership thinking earns its place. The relevant figure is not the cost of this series. It is the cost of this series plus what the next one will cost given what you hold when this one finishes. A commissioning team planning a single title can reasonably ignore that. A team planning a slate cannot.

5. Normalise the Failure Terms

Every bid describes what happens when things go well. Compare what each one says about the other case. What is the remedy if delivered episodes do not meet the agreed specification. What is the remedy if delivery is late. What happens if the series is delivered and the commissioning team judges it unusable, and who decides that. What notice either party can give and what is owed at that point.

Most AI production bids are thin here, which is a characteristic of a young category rather than bad faith. Thinness is still a risk you are taking on, and the right response is to ask for the terms to be written rather than to assume the relationship will hold. A bidder who responds to that request with specifics is telling you something useful about how they operate. A bidder who responds with reassurance is telling you something too.

Pay particular attention to the acceptance mechanism, meaning who signs off that an episode is delivered, on what criteria and within what window. An unspecified acceptance process is the single most common cause of payment disputes in production contracts of any kind, because it lets both parties hold a different and sincere belief about whether the work is done.

6. Rebuild Both Bids as a Total Cost of Engagement

With the five normalisations applied, rebuild each bid as a single figure that includes the base price, a realistic revision overage allowance, any delivery specification items the bidder excluded that you will have to buy elsewhere, and any asset items you will need to recreate later because they are not transferring. That figure is the comparable number. It is usually meaningfully different from the headline, and the ranking of the two bids reverses often enough that the exercise is worth doing every time.

Then add the items that have a cost but not a price. How much of your own team time each arrangement will consume, which is largely a function of how much reporting the production company does without being asked. How much continuity risk you are carrying if key assets do not transfer. How much of the specification you are still going to have to write yourself after signature. These belong in the decision even though they will not appear on an invoice.

Document the rebuild and keep it. When the series is delivered, comparing the rebuilt estimate against what actually happened is the fastest way to become a better commissioner, and it gives the next procurement round a baseline that came from your own production rather than from a vendor deck.

What the Normalised Comparison Usually Reveals

Three patterns come up repeatedly. The first is that the cheaper bid is cheaper because it assumed a lighter finishing standard, and the gap closes to near nothing once both are priced to the same delivery specification. The second is that the more expensive bid included a revision structure generous enough to be worth more than the difference, particularly on a first commission where note volume is unpredictable.

The third is the one that matters most for a platform or media company thinking past a single title. Bids differ more in what they leave behind than in what they deliver, and the difference compounds. A team that commissions three series from a partner who retains the asset set has three series and no production capability. A team that commissions the same three from a partner who transfers it has three series and the ability to commission the fourth from anyone, including themselves. That difference rarely appears in the comparison until somebody deliberately puts it there.

Axis AI Studios Perspective

Axis AI Studios writes bids against a delivery specification rather than against a headline price, because a quote that cannot be compared line by line is not a quote a commissioning team can defend internally. Every Axis proposal states episode runtime, assumed shot count per episode, the revision structure with the correction and redirection boundary written out, the full delivery specification item by item, and what transfers to the client at the end of the engagement.

That last point is a position rather than a convention. The character references, prompt library, style guide and continuity bible built during a series are production assets, and a client who commissioned them should hold them. It is what makes a second season a continuation of the first, and it is what keeps the commissioning relationship a choice rather than a dependency.

If you are holding two bids that do not describe the same work in the same language and you need a third read against a normalised specification, we will write one. Send the brief and both proposals to business@axisaistudios.com.

FAQ

Is a per episode price or a per minute price better for a commissioning team?

Neither is inherently better, but a per episode price is only meaningful alongside a stated runtime and a stated shot count assumption. Without those, per episode pricing lets a bidder compete on an assumption the commissioning team cannot see. Ask for all three figures regardless of how the bid is structured, then convert both bids to a per shot basis to compare them.

What is a reasonable revision allowance on a first commission?

More than you expect to need. First commissions carry heavier note volume because the visual register, character interpretation and house preferences are all still being established against real output rather than against a brief. The specific number matters less than having the correction and redirection boundary written down and an overage rate stated in advance, so a heavy notes month is a budget line rather than a negotiation.

Should the delivery specification come from the commissioning team or the production partner?

From the commissioning team, even if it is short and imperfect. A specification written by the buyer makes bids comparable and forces the internal decisions that are cheapest to make before production starts. A production partner will usually improve it, and a good one will tell you which lines are unnecessary, but the initiating document should be yours.

Further Reading

For the questions that belong in the evaluation before price enters the conversation at all, the production partner due diligence checklist covers what to verify about a company before comparing its numbers.

For turning the normalised comparison into a repeatable process across a slate, the supplier scorecard guide covers how to score partners on criteria that hold from one commission to the next.

For writing the failure terms the bids usually leave thin, the production SLA guide covers the provisions a commissioning agreement needs before signature.