How a Management Layer Would Run a Pilot With Two Production Partners Side by Side
Two production partners. Both credible. Both with strong showreels and confident proposals. A platform that wants to commission AI native vertical drama at volume has to choose, and the usual way to choose is to commission one series from each and see what happens. Months later, two series have delivered. One performed better. Nobody can say whether that was the partner, the script, the genre, the launch week or luck.
This is the core problem with informal side by side pilots. They feel like a comparison, but the variables are not controlled, so the result is not evidence. This piece sets out how an external production management layer would structure a two partner pilot so that the platform ends it with a decision it can defend. It reflects the AXIS Management model as a working strategy position, built on the three layer structure of platform, management layer and production network, and on a pilot to portfolio route into volume commissioning.
1. Start From the Decision the Pilot Has to Support
A pilot is not an end in itself. It exists to answer a specific question that unlocks the next commissioning decision. Before any brief is written, the management layer would work with the platform to state that question in a single sentence. Which of these two partners should carry the first portfolio slate in this genre. Whether either partner can hold quality at the next price tier up. Whether the platform should split volume between both.
The question shapes everything after it. If the decision is about which partner should lead a romance slate, both pilot series should be romance. If the decision is about holding quality at a higher tier, both series should be briefed to that tier. A pilot designed without its decision in mind tends to produce interesting observations and no conclusion.
Agree what would change the decision
The management layer would also ask the platform what result would change its mind. If partner A is favoured going in, what would partner B need to show to win the first slate. Writing this down before the pilot starts protects the comparison from hindsight, where every result is read as confirming the original preference.
2. Control the Variables That Are Not Being Tested
The principle is the same one behind any fair comparison, including A/B testing in product and marketing work. Change one thing, hold everything else steady, and measure. In a production pilot, the thing being tested is the production partner. Everything else should be as close to identical as the platform can make it.
Same brief, or matched briefs
The strongest design gives both partners the same script and the same brief, and compares execution directly. That is not always practical, because a platform rarely wants two versions of one series on its app. The next best design uses two scripts matched on genre, episode count, episode length, number of lead characters, number of core locations and number of spectacle scenes. The management layer would define those matching criteria and check both scripts against them before sign off.
Same tier
Both series should be commissioned at the same point in the quality and price tier framework. Comparing a partner working at one tier with a partner working at another says more about the tier than the partner. If the platform wants to learn about tier as well, that becomes a separate question with its own design, rather than a confounding factor inside this one.
Same launch conditions
Release window, marketing support, placement on the app, paywall position and promotional spend all affect performance. The management layer would agree with the platform that both series launch under matched conditions, ideally in the same period, so that audience results can be compared without large external differences.
3. Define the Measures Before Either Partner Starts
A fair pilot measures each partner across the full production journey, not only the final retention curve. Audience performance matters most, but it is the noisiest signal and the one most influenced by things the partner does not control. The management layer would track three groups of measures.
Delivery measures
Did each milestone arrive when agreed. How many review rounds did each episode need. How many notes were addressed in the first revision. How often did the partner raise risks early rather than late. These are the measures that predict how a partner behaves at volume, when a slate of several series runs in parallel and late surprises become expensive.
Quality measures
Character consistency across episodes, continuity errors per episode found in review, number of shots returned for regeneration, adherence to the agreed style guide, and how the delivered series scores against the tier standard it was commissioned at. The management layer would apply the same review checklist to both series so these numbers can be compared directly.
Audience measures
Completion rate on episode one, drop off at the paywall episode, retention through the series, and conversion where relevant. These are reported by the platform and read alongside the production measures, not in isolation. A partner that delivers cleanly and on standard but underperforms with the audience tells a different story from a partner that delivers late and messily and still performs.
Set a minimum size for the pilot
A comparison needs enough material to be meaningful. One short series from each partner can show obvious differences, but it can also hide them, because a single strong or weak script can dominate the result. The management layer would agree with the platform a minimum pilot size in episodes and, where budget allows, consider two matched series per partner in the same genre and tier. The goal is not volume for its own sake. It is enough repetition that a pattern can be separated from a one off.
4. Run Both Partners Through the Same Management Process
In the three layer model, the platform commissions through the management layer, and the management layer works with partners in the production network. The value of that structure in a pilot is that both partners experience identical management. Same kickoff format. Same milestone schedule. Same review cadence. Same notes format. Same escalation path.
This is the property management analogy in practice. A property manager overseeing two buildings applies one maintenance standard, one inspection routine and one reporting format to both, so the owner can see which building is performing and why. A management layer overseeing two pilot series would do the same. Differences in outcome then point more reliably to differences in the partners themselves, rather than to differences in how they were briefed or reviewed.
Keep notes neutral and logged
Every note sent to either partner would be logged with its episode, timestamp, category and status. The management layer would check that the volume and severity of notes are applied consistently, so neither partner is held to a stricter standard. This log becomes part of the pilot evidence, showing not only how many problems each series had but how each partner responded to them.
5. Report Side by Side Throughout, Not Only at the End
A pilot that reports only at the end leaves the platform blind for months and invites second guessing when the result arrives. The management layer would report both series on the same page at each milestone. Where each series stands against schedule. How many review rounds each has used. Open risks for each. Quality findings for each.
This gives the platform early sight of patterns. If one partner needs three review rounds per episode and the other needs one, that is visible by the third episode, not the thirtieth. It also lets the platform intervene if something goes wrong that would contaminate the comparison, such as a script problem affecting one series and not the other, which the management layer would flag and resolve before it distorts the result.
Separate partner issues from brief issues
Some problems in a pilot come from the brief, not the partner. An unclear character description, a late script change from the platform side, a reference image that conflicts with the style guide. The management layer would tag each issue by origin, so the final comparison counts only what each partner was responsible for.
6. Turn the Results Into a Portfolio Decision
At the end of the pilot the management layer would produce a single comparative report built around the decision question agreed at the start. It would set out the delivery, quality and audience measures for both partners side by side, note any conditions that differed between them, and state what the evidence supports. The report answers the question the platform asked, not a broader one.
Three outcomes the report would usually point to
The first is a clear lead, where one partner outperforms on most measures and becomes the natural choice for the first portfolio slate in that genre and tier. The second is a split, where each partner shows strength in different areas, such as one stronger on delivery discipline and the other stronger on audience response, which can support allocating different series types to each. The third is an inconclusive result, where differences are small or confounded. In that case the right move is often a second, narrower pilot, rather than forcing a choice the evidence does not support.
Carrying the pilot into the portfolio
Whichever outcome applies, the reference material, style guides, character libraries and review data built during the pilot carry into the portfolio. The partner selected for the first slate starts with a known standard and a known review rhythm. The partner not selected may remain in the production network for future series where its strengths fit better. Nothing learned in the pilot is lost.
7. Understand How the Pilot Would Be Structured Commercially
In the AXIS Management model, the commercial structure rests on three mechanisms: a set fee for the management work, a management percentage on the production value under management, and production margin on the work delivered through the production network. In a two partner pilot, those mechanisms would apply across both series, so the platform sees one management relationship rather than two separate supplier contracts.
For the platform, the practical effect is a single point of commissioning for the pilot, one reporting line, and one comparative report at the end. The management layer carries the coordination work of running two partners in parallel, including the matching of briefs, the neutral review process and the side by side reporting, which would otherwise sit with the commissioning team.
The tier framework also shapes the commercial frame of the pilot. Because both series are commissioned at the same tier, the production value under management is comparable, and so is the effort each partner is asked to deliver. That keeps the cost side of the comparison as fair as the quality side. If the pilot then leads into a portfolio, the same mechanisms simply extend across a larger slate, so the platform is not renegotiating the structure of the relationship at the moment it wants to scale.
What the platform keeps throughout
Running the pilot through a management layer does not remove the platform from the decisions that matter. The platform still approves scripts, signs off the style guide for each series, owns the audience data and makes the final portfolio decision. What moves to the management layer is the coordination: briefing both partners, running the review process, keeping the notes log, and producing reports that let the commissioning team compare the two series without managing either one directly day to day.
Axis AI Studios Perspective
A side by side pilot is only as good as its design. Two series launched under different conditions, from different briefs, reviewed by different people, will always produce a result. It will rarely produce a decision. The value of a management layer in a pilot is the discipline it applies: a single decision question, matched briefs and tiers, defined measures, identical management for both partners, and reporting that separates what each partner controlled from what it did not.
AXIS Management is a working strategy position built on that discipline. The three layer structure of platform, management layer and production network, the quality and price tier framework, and a pilot to portfolio route into volume commissioning together describe how Axis AI Studios intends to help platforms commission AI native vertical drama at scale with less internal production overhead. A two partner pilot is one of the clearest places that model would earn its keep, because it turns a judgement call into evidence.
For platform teams thinking about how to compare production partners before committing to a slate, the conversation starts at business@axisaistudios.com.
FAQ
Why not simply commission one series from each partner and compare the results?
Because too many things differ at once. Different scripts, genres, launch windows and review processes all affect the outcome, so a better performing series does not prove a better partner. A structured pilot holds those factors as steady as possible so differences point more reliably to the partners themselves.
Does a two partner pilot need the same script for both series?
Not necessarily. The same script gives the cleanest comparison, but platforms rarely want two versions of one story live. Matched scripts with the same genre, length, character count, location count and tier are a practical alternative, provided the matching criteria are agreed and checked before production begins.
What happens to the partner that is not selected after the pilot?
That depends on the result. In a split outcome, both partners may carry different series types. Where one partner leads clearly, the other may remain in the production network for future series where its strengths fit. The reference material and review data from the pilot remain available either way.
Further Reading
For platform teams deciding when a pilot has done its job, the guide to what a platform would need to see in a pilot report covers the evidence that supports a move from pilot to portfolio.
For commissioning teams comparing partners before a pilot begins, the method for comparing two AI production bids covers how to normalise proposals that describe deliverables in different ways.
For readers who want the structural background, the explanation of how the three layer model splits responsibility covers what the platform, the management layer and the production network each own.

Let's set
the new standard together.
If you're working on something, we'd like to hear about it.
