The First 90 Days of Commissioning AI Vertical Drama: What Goes Wrong and How to Prevent It

The first 90 days of a vertical drama commission produce either a series that is on track for platform acquisition or a series that requires expensive revision at delivery. The outcome is almost entirely determined by five decisions made before episode one enters generation — and all five are commissioning decisions, not production decisions.

The production partner executes against what the commissioning party decides. A correctly specified commission with a validated premise, an approved arc map, a precise visual register, and a delivery specification confirmed with the platform produces a series the production partner can execute correctly. An under-specified commission produces a series built on assumptions — and assumptions discovered in week seven cost significantly more to correct than the brief development session that would have prevented them.

This post covers the five decisions that go wrong in the first 90 days, what causes each one, what it costs when it happens, and the specific action that prevents it.

Decision 1: Commissioning Without a Validated Premise

The most expensive first-90-days error is committing the full production budget to a premise that has not been commercially validated. A 70-episode series at AI-native standard professional quality costs $60,000 to $100,000. The commissioning business that signs the production agreement without first running a three-episode concept test is making a $60,000 to $100,000 bet on creative conviction rather than commercial evidence.

What causes it: the concept test feels like a delay. The commissioning party wants the series, not a test. The production partner, depending on their incentives, may not push back on this.

What it costs: if the premise fails at the paywall — hook rate below 45%, continuation rate below 55%, paywall intent below 30% — the full production budget has been spent discovering what a $15,000 to $20,000 concept test would have told the commissioning party in week four.

How to prevent it: run the concept test before the production agreement for the full series is signed. The three-episode test takes three to four weeks and $15,000 to $20,000. The full series commission follows only if the concept test clears the go thresholds. The concept test's production infrastructure — character reference pack, arc map for the three-episode window — carries directly into the full series production without rebuild.

Decision 2: Signing the Production Agreement Without Reading the IP Provisions

The production agreement's IP provisions determine who owns the characters, the story world, and the sequel rights after delivery. Most first-time commissioners sign production agreements drafted by the production partner and focus on the price and the delivery date. The IP provisions in the production partner's standard agreement are drafted to protect the production partner's interests.

What causes it: the commissioning party treats the production agreement as a logistics document rather than as a franchise infrastructure document. Legal review feels like overhead on a creative project.

What it costs: the commissioning business that does not own the IP from the primary series cannot commission the sequel without negotiating adaptation rights from whoever does own it. The sequel premium — the higher licensing fee commanded by a series with documented performance data — is not capturable by a commissioning party that does not own the IP.

How to prevent it: confirm three specific provisions before signing. The commissioning party owns all IP produced under the commission including characters, story world, scripts, and generated visual content. The exclusivity window with the primary platform is capped at 12 months maximum. AI tool usage documentation is included in the delivery package for chain of title purposes. If the production partner's standard agreement does not contain these provisions, request them before execution.

Decision 3: Under-Specifying the Visual Register

The visual register — colour temperature, lighting direction, environment category, character wardrobe arc — is the specification that determines whether the series' visual identity is consistent across 70 episodes and matches the platform's acquisition quality standard. A visual register described in mood terms ("premium, cinematic, dark") produces a series whose visual identity reflects the generation operator's aesthetic interpretation. A visual register specified in generation-parameter terms produces a series whose visual identity is reproducible from the brief.

What causes it: the commissioning party knows what they want visually but does not know how to translate that into the technical specification language that generation operators use. The brief development session skips this because the conversation is more comfortable at the creative description level.

What it costs: a visual register discovered at the generation sample review — when the commissioning party sees episode five and says "this is not the look I imagined" — requires a pre-production rebuild of the character reference pack and the style guide before generation continues. At 15 episodes already generated, the cost is $3,000 to $8,000 in regeneration credits plus the two-to-three-week timeline extension.

How to prevent it: the brief development session must produce a visual register specification at the generation parameter level. Not a mood board. Not a creative description. The specific colour temperature, the lighting direction, the contrast ratio, the environment category, and the character wardrobe state for each arc position. This is a three-to-four-hour conversation with the production partner that prevents a $5,000 to $10,000 correction later.

Decision 4: Not Confirming the Delivery Technical Specifications With the Platform Before Production Begins

The delivery technical specifications — codec, resolution, aspect ratio, audio loudness standard, subtitle format, metadata structure — vary by platform and change over time. A production that delivers to a platform's specifications from six months ago may deliver a package the platform's current technical team rejects.

What causes it: the production partner confirms delivery specifications at the start of the production based on their most recent delivery to that platform. If the platform has updated its specifications in the intervening period, the production partner may not be aware.

What it costs: discovering a specification mismatch at delivery — subtitle format incorrect, audio loudness at the wrong LUFS target, metadata structure missing a required field — delays platform acceptance by two to four weeks while the delivery package is corrected. In a production already at the eight-week mark, this extension is commercially and reputationally costly.

How to prevent it: the commissioning party contacts the platform's technical delivery team directly in week one of the production and confirms the current delivery specifications. The confirmed specifications are documented in the production brief as the delivery requirement. The production partner executes against the confirmed specification rather than against their most recent delivery memory.

Decision 5: Missing the Arc Map Approval Gate

The arc map is the production's structural blueprint. It specifies the paywall episode position, the midpoint reversal, the antagonist arc's specific injustices, and the resolution sequence. It is produced in pre-production week two and presented to the commissioning party for approval before scripting begins.

The arc map approval gate is the single most commercially important approval event in the entire 90-day production window. An arc map problem discovered at the script review — when 30 scripts have been commissioned against it — costs script revision across every affected episode. An arc map problem discovered at the arc map review costs a two-to-three-hour conversation.

What causes it: the commissioning party approves the arc map without sufficient engagement. The document looks complete. The approval happens quickly because the commissioning party wants the production to proceed. The arc map's structural problems — paywall at the wrong position, antagonist arc without specific injustices, resolution that discharges tension too early — are not visible on a quick read. They become visible when the scripts built on the arc map fail the structural review.

What it costs: arc map revision after scripting has begun requires script revision proportionate to the number of episodes affected. A paywall position correction discovered at episode 15's script review requires revision of episodes 5 through 20 — 15 scripts — before the correct paywall structure can be established.

How to prevent it: allocate three to four hours to the arc map review. Not a quick read. A structured evaluation against four specific questions. Does the paywall episode position produce the maximum emotional investment the series has built to that point? Does the antagonist's specific injustice at the paywall moment create urgent, righteous anger rather than diffuse dissatisfaction? Does the midpoint reversal change the rules of the situation convincingly rather than just escalating the existing situation? Does the resolution discharge specific emotional debts rather than resolving the situation generally? If any answer is uncertain, resolve it at the arc map stage.

The 90-Day Prevention Checklist

The five errors above are all preventable with decisions made before week three of the production. The prevention checklist:

Week one: confirm the delivery technical specifications directly with the platform's technical delivery team. Document the confirmed specifications in the production brief.

Week one to two: review the production agreement's IP provisions before signing. Confirm character ownership, world ownership, exclusivity cap, and AI documentation provisions.

Week two: run the brief development session to produce the visual register specification at generation parameter level, not at creative description level.

Week two: evaluate the arc map against the four structural questions above. Allocate three to four hours. Do not approve without confident answers to all four.

Before week one: confirm whether the concept test is preceding the full series commitment. If not, understand what commercial evidence the full series commitment is based on.

Five decisions. All in the first 90 days. All preventable before they cost money.

Axis AI Studios Perspective

The first 90 days of a vertical drama commission are where the series is either set up correctly or set up for expensive correction. At Axis AI Studios, the pre-production process is explicitly designed to prevent all five errors described in this post. The brief development session produces the visual register at specification level. The arc map review is a structured gate with defined evaluation criteria. The delivery specification is confirmed with the platform before production begins. The IP provisions are proposed in the commissioning party's favour, not in the production partner's favour.

For businesses commissioning AI-native vertical drama for the first time who want the first 90 days to go correctly, reach out at business@axisaistudios.com.

FAQ

How Long Does the Complete Pre-Production Process Take Before Generation Begins?

The pre-production process — brief development session, arc map development, character reference pack build, and commissioning party approvals at each gate — takes two to three weeks. Generation begins in week three, running partially in parallel with the scripting process. The two to three weeks of pre-production are not delay. They are the investment that prevents the revision costs and timeline extensions that under-specified pre-production produces later.

What Happens if the Concept Test Fails?

A failed concept test is not a failed investment. It is a $15,000 to $20,000 data point that identifies specifically which element of the production underperformed — the cold open, the arc map, or the paywall position — before $60,000 to $100,000 is committed to a full series built on the same underperforming element. The correct response to a failed concept test is targeted revision of the failing element, a revised concept test, and a full series commission from the revised and validated premise.

Can the Arc Map Be Changed After Scripting Has Begun?

Yes, but at increasing cost the further into scripting the change occurs. An arc map change before any scripts are commissioned costs nothing except the time of the brief development session. An arc map change after episode 15's scripts are commissioned costs revision of all scripts that depended on the changed structural element. An arc map change after generation has begun costs revision of both scripts and generation outputs. The correct intervention point is always the arc map review in week two.

Further Reading

For the concept test that prevents Decision 1, the guide to the concept test as a business decision covers the go thresholds, distribution methodology, and the go/stop decision framework.

For the production brief that prevents Decision 3, the guide to what a good AI-native production brief looks like vs a bad one covers concrete examples of under-specified and correctly specified briefs across every section.

For the arc map structure that the Decision 5 review evaluates, the 70-episode arc mapped beat by beat guide covers the structural skeleton including paywall position, midpoint reversal, and antagonist arc specification.