All insights

In-house AI ad production or an agency: how to decide (2026)

A decision table for building AI ad production in-house or buying it from a studio: cost per approved ad, approval work, rights and AI labels.

Kevin Dosanjh
Co-founder / CMO, AI Vidia · Updated

Build AI ad production in-house when one person can own the look, approve product facts and run the tools every week. Buy it from a studio when you need a steady volume of finished ads faster than you can hire. Compare cost per approved ad, never tool prices, and keep product approval inside your team either way.

Overhead view of a light oak desk with a printed product fact sheet, a phone showing a vertical ad of an unlabelled amber bottle on a stone block, a short checklist with a pencil, an orange sticky note and a second stack of blank paper
Image generated with AI: a product fact sheet, a 9:16 ad frame and a review checklist, which both routes need before an ad is approved.
On this page9 sections
  1. 01What stays with your team either way
  2. 02In-house or studio: the decision table
  3. 03When in-house is the better choice
  4. 04When a studio is the better choice
  5. 05What each route costs per approved ad
  6. 06A split model: keep approval, buy production
  7. 07Who labels AI ads and who owns the files
  8. 08Test before you hire or sign
  9. 09A checklist before you decide

Choose between in-house AI ad production and an outside production studio or agency by looking at the work you actually have. Product knowledge, creative direction, production and approval do not have to sit in one place, and where you split them decides the cost.

What stays with your team either way

Your team knows the product, customers and business boundaries. That knowledge must be available regardless of who produces the material. Identify who can confirm product facts, prioritise messages and approve the finished work.

In-house production can provide close access to this knowledge. An agency can provide external capacity. Neither resolves an unclear assignment or missing decision owner. Start with the working process rather than assumptions about which organisational model is more modern.

In-house or studio: the decision table

Read across each row. If most of your answers land in the first column, build the capability. If most land in the second, buy production and keep approval.

Where in-house production and a studio each fit, by decision criterion
CriterionIn-house fits whenA studio fits whenAsk either way
ApprovalOne named person signs off product facts, claims and the AI label every week.You can name an approver but not a producer.Who stops an inaccurate ad, and how fast?
InputsYour team owns current product photos, packaging files and approved claims.You can send product photos and sign a written product fact sheet before production starts.Who keeps the reference files current?
Volume and rhythmDemand is steady and the same few products repeat.You need a fixed number of finished ads every week, more than your team can make.How many ads actually went live last month?
RevisionsYour producer has time for feedback rounds on every file.You want one agreed revision round per video, written into the scope.What counts as a revision, and what is a new concept?
Creative judgmentSomeone in-house can hold one look across dozens of videos a month.You want concepts, scripts and variant plans from people who do it daily.Who decides which idea gets made?
Real footageUse, taste, fit or a real customer has to be filmed.AI scenes can carry the story and your existing footage can be cut in.Which shots must be filmed for the claim to be true?
Rights and labelsYour legal team reviews each AI tool's commercial terms.You want file transfer and the labelling plan in one written agreement.Who applies the AI label at launch?
Operational fitTools, prompts and file naming already sit inside your workflow.Finished files must arrive named for your media buyer's testing convention.Who uploads, and who reads the results?
A printed fact sheet on a clipboard with an orange tab, a pencil, and a phone standing upright that shows a vertical ad of a plain white mug on a wooden table
Image generated with AI: a product fact sheet beside a 9:16 ad frame. Whoever produces the ad, your team signs off the facts the frame must match.

Do you have consistent products and a familiar visual direction, or does every assignment require new scenes, people and techniques? Is demand continuous or concentrated in a few large campaigns? These answers affect which capacity is worth building.

List routine work and common exceptions. Urgent requests, new markets and changed product facts can consume substantial time. A staffing plan based only on the ideal week will miss that work.

When in-house is the better choice

In-house wins when the work repeats and the product knowledge is hard to hand over. A brand with a small, stable range and a designer who already ships ads every week should invest in better tools before it buys outside production.

  • Your ads need filmed proof: a person using, tasting or wearing the product.
  • Product facts change daily, and a same-day correction matters more than volume.
  • Compliance requires prompts, source files and the whole pipeline to stay inside the company.
  • A creative lead already has spare hours each week to direct and review.

If two or more apply, build in-house first. The AI Vidia versus in-house comparison lists the same trade-offs for our own service.

When a studio is the better choice

  • You need more finished ads each month than your team can brief, make and check.
  • Nobody in-house owns AI production, and hiring would take longer than a test.
  • You want a fixed weekly delivery with one revision round per video, written into the agreement.

On AI Vidia's monthly plans for AI video ads, the first review files arrive on day 3 and finished files are ready for your ad account by day 7, counting the brief as day 0. The one-time First Batch runs on the same clock. Every clock starts once we have your product photos and your signed product fact sheet.

The Grow plan is EUR 4,500 a month for 30 finished videos up to 30 seconds, delivered weekly in 9:16, and usually fits brands spending EUR 20,000 to 50,000 a month on paid social. That price does not show that an internal role can go, and it promises no particular return.

If an agency of record or a procurement review sits between you and production, the page for brand teams shows how a paid pilot runs and what your procurement team can check first.

Agree the boundary with whoever runs your advertising. Producing ads does not include campaign management by default. AI Vidia publishes its split in a who-does-what table: either your team or AI Vidia launches the ads and runs the tests, agreed with each client.

Need new ads?
Describe your product and what you want to test.
Send your brief

What each route costs per approved ad

Price lists compare the wrong thing. The useful number is the total cost of a route divided by the ads your team approved for launch.

Use one formula for both routes: external fees plus your team's hours at your hourly cost, divided by the ads you approve. Count the hours spent on outputs you reject. They are part of the cost.

  • Internal time for concepts, briefing, production and approval.
  • Tools, training and technical maintenance.
  • External production and separately agreed tasks.
  • Revisions, file handling and handover.
  • Periods with more or less demand than planned.

For scale on the in-house side, the US Bureau of Labor Statistics put the median annual wage for graphic designers at $62,960 in May 2025, before benefits, tools and the time it takes to direct the work. Use your own salary data; the median only shows the order of magnitude.

On the studio side, AI Vidia's monthly plans work out to roughly EUR 167 per finished video on Start down to EUR 120 or less on Studio, including concepts, variant planning and 9:16 files. The pricing page lists the full scope and what no plan includes, such as media spend.

ToolCost per usable ad: in-house or studioPut your in-house route in Route A and a studio quote in Route B. Count only the files your team approved.

Example numbers. Replace them with yours.

The example compares a do-it-yourself setup (Route A) with a managed studio (Route B). Every number in it is made up.

Currency

Labels only. Nothing is converted.

One rate for both routes: a cash cost or an internal rate.

Route B as a published AI Vidia plan (optional)

Copies only the published fee and file count into Route B, and sets the currency to EUR.

External fee or tool cost

Studio, freelancers, a share of a tool subscription, music. Count each cost once.

Your team's hours

Briefing, checking, fixes and uploads. Hours a vendor bills belong in the fee.

EUR 1,200 of team time
EUR 250 of team time
Files produced

Every file made for this brief, rejected ones too.

Files approved

Signed off by whoever owns product accuracy.

50% of produced
80% of produced
Files launched

Approved files that went live. One held for a later test is not wasted.

60% of approved
63% of approved

Cost per approved file, example numbers

Route AEUR 160 
Route BEUR 156.252% lower

Route B costs EUR 3.75 less per approved file (2% lower).

These routes are within 10% of each other. A few files or hours would flip the result.

What one approved file is made of
Route AEUR 160
Route BEUR 156.25
Files at each stage
Route A30 / 15 / 9
Route B30 / 24 / 15

Where to start

Want a second pair of eyes on your numbers?

Copy the result, then paste it into the brief form so we start from your numbers.

What would flip it

Route A needs at least 16 of 30 files approved to match Route B's cost per approved file. Approved now: 15.

Route B needs at least 24 of 30 files approved to match Route A's cost per approved file. Approved now: 24.

All the numbers
Example numbers, not AI Vidia prices or results
All the numbersRoute ARoute B (example)
Total costEUR 2,400EUR 3,750
Per file producedEUR 80EUR 125
Per file approvedEUR 160EUR 156.25
Per file launchedEUR 266.67EUR 250
Share of produced that is approved50%80%
Share of approved that is launched60%63%
How it is worked out

Total cost = external fees + your team's hours x hourly cost. Cost per approved file = total cost / approved files.

This calculator counts production cost per usable file. It does not score creative quality, predict results or replace a quote.

A launched ad is not a winning ad: whether it worked needs a test with a defined result.

Your numbers stay in this browser unless you copy a link or the result.

A link holds your numbers after the # in the address, a part browsers do not send to a server, but anything that records the full page address could see it, so share it only with people you trust.

We count that the tool was used, never what you typed.

Take the calculator's made-up example. Route A costs EUR 1,200 in fees plus 24 team hours at EUR 50, so EUR 2,400, and 15 of its 30 files are approved: EUR 160 per approved ad. Route B costs EUR 3,500 plus 5 hours, so EUR 3,750, with 24 of 30 approved: about EUR 156. One more approved file on Route A, 16 instead of 15, brings it to EUR 150 and makes it the cheaper of the two.

Small changes in approval rate decide the answer. Run it with your own numbers before anyone calls one route cheaper. The cost per usable ad guide explains each field.

A split model: keep approval, buy production

Imagine a furniture brand with an internal designer and a product manager. They can provide accurate dimensions, images and brand direction, but have limited time to edit many video variants. They could keep concept prioritisation and product approval internally while buying the defined production externally.

If nobody has time to prepare a usable brief, more external producers may not help. First assign someone to consolidate inputs and make decisions. This example illustrates responsibility, not a measured financial advantage.

Also assign maintenance of reference files and approved descriptions. Quality can deteriorate if both parties assume the other is updating the source material.

Who labels AI ads and who owns the files

Outsourcing production does not settle who labels the ad. Under the EU AI Act, deployers must disclose deep fake image, audio or video content, and the European Commission says these transparency rules apply from 2 August 2026. The Commission adds that a company remains a deployer when contractors operate the system on its behalf and under its control. Whether the deployer on a typical ad job is your brand, the studio or both is set out in our guide to who has to label an AI ad under the EU AI Act.

Meta labels ad images created or edited with third-party AI tools with AI info. When a studio makes the ad and your team publishes it, write down who discloses on each platform; our guide to AI labels on Meta, Google, TikTok and EU ads covers each one.

Ownership needs the same care. The US Copyright Office concluded in January 2025 that prompts alone do not give a user enough control to be the author of AI output, while human selection, arrangement and editing can still be protected. Ask any producer, internal or external, which files and style systems pass to you and how the human work on them is documented. What the agreement should say about usage rights is covered in our guide to usage rights for AI ads. On AI Vidia's plans, all our rights in the finished ads transfer to you on delivery; AI presenters are licensed as they appear in your ads, and our prompts, workflows and project files stay with us. How we handle labels, product accuracy and ownership is set out on our AI governance page.

AI presenters need one more check. The FTC's final rule bans reviews and testimonials that misrepresent that they come from someone who does not exist or who never used the product. An AI presenter is AI-generated, not a customer. Whoever makes the ad, agree who approves each script, and keep every product claim to your signed product fact sheet.

Test before you hire or sign

Choose a representative assignment with the product details that usually cause difficulty. Complete the brief, first concept, consolidated feedback and delivery. Record your own workload as well as the appearance of the finished result.

Check whether errors are understood and corrected, versions can be found and files suit their intended use. A beautiful sample without a usable process is weak evidence for a long engagement. One sample also cannot establish how every future assignment will run.

Agree which approved files, references and descriptions your team should retain. If an external producer leaves, the next person should be able to understand the product and latest decisions without starting over.

Test this during the trial. Ask someone who did not make the video to locate the right product, script and export. Their questions reveal documentation gaps that can be addressed before expanding capacity or committing to a longer engagement.

If the table leaves you undecided, test the question with money on the line. AI Vidia's First Batch costs EUR 999: one concept made as 4 videos and 4 images, which you can run against your current best ad in your own ad account on one metric agreed up front. Log your team's hours during the test and put them into Route B afterwards.

A checklist before you decide

  1. Name the person who approves product facts, claims and the AI label.
  2. Count the ads that actually went live last month, by product.
  3. List the shots that must be filmed for a claim to be true.
  4. Write down your team's hours per finished ad today.
  5. Price both routes per approved ad, with the same hourly rate for your team.
  6. Agree what one revision round covers.
  7. Decide who uploads files, launches ads and reads results.
  8. Put file transfer, reference files and the labelling plan in writing.

Start with item two. Open last month's export from your ad account, count the ads that went live and enter that number and your team's hours as Route A in the calculator. It takes under half an hour and gives you the figure any studio quote has to beat. If roles are still unclear, use the team structure guide.

Frequently asked questions

01Is in-house AI ad production cheaper than hiring an agency?
In-house costs less when your team approves most of what it makes and already has the hours; a studio costs less when those hours are the bottleneck. Add external fees to your team's hours at your hourly cost, divide by the ads you approved for launch, and compare the two numbers. The calculator above does the arithmetic with your own figures.
02What should stay with our team if we hire a studio?
Product facts, claim approval and the final sign-off stay with your team. A studio can write concepts and scripts, produce the files and check them against your references, but only your team can confirm that a price, an ingredient or a fit claim is true. AI Vidia asks every client to sign off a written product fact sheet and the look before anything is generated.
03When does a split between in-house and a studio make sense?
A split works when your team has the product knowledge and approval time but too few production hours. Keep concept priority and sign-off inside, buy defined production outside, and write down who briefs, who approves, who launches and who reads the results. Without a named approver neither route works, and adding producers only adds files.
04Who is responsible for labelling AI-generated ads?
Agree it in writing before the first brief, because outsourcing production does not settle who labels. In the EU, the AI Act requires deployers to disclose deep fake image, audio or video content, and the European Commission says this applies from 2 August 2026. The studio or agency that decides how AI is used in the ad is the deployer. A brand that only commissions the ad, without deciding how AI is used, is not, but becomes one once it makes that decision itself, for example by choosing an AI presenter. Name who decides the label for each platform and market and who applies it at launch.
05How do we test a studio before signing a monthly retainer?
Run one paid test on a representative product before any monthly commitment. Give the studio your current best ad as the control, agree one metric up front and log your team's hours as well as the result. AI Vidia's First Batch gives you the material for EUR 999: one concept made as 4 videos and 4 images.
06Is a studio or an in-house team cheaper for a global brand in several markets?
Neither route is cheaper by default, because each extra language adds a review step to both: a native speaker has to check every ad in that language, whoever produces it. AI Vidia reviews English and Danish in-house; other languages go to your own reviewer before anything ships. Our fixed plans run to 60 videos a month; Studio, from 100 a month, and several markets are quoted on a call. Put the review hours into both routes in the calculator above.

How we made this article

Regulatory, platform and wage facts link their primary sources, read on 13 October 2026. Plan prices come from AI Vidia's published pricing page. The cost example is arithmetic on the calculator's made-up numbers, not a benchmark or a client result.

We research and draft with AI assistance and check figures against the sources below.

Sources

  1. 01European Commission: Transparency obligations under Article 50 of the AI Act (FAQ), read 13 October 2026
  2. 02US Copyright Office: Copyright and Artificial Intelligence, Part 2: Copyrightability, January 2025
  3. 03US Bureau of Labor Statistics: Occupational Outlook Handbook, Graphic Designers (pay data for May 2025, page modified 27 August 2026)
  4. 04Meta Help Center: How AI-generated images in ads are identified and labeled on Meta
  5. 05Federal Trade Commission: Federal Trade Commission announces final rule banning fake reviews and testimonials, 14 August 2024

Next step

Get a proposal for your next ads.

Tell us about your product, audience and ad production needs. We review your brief and reply with a proposed next step.

Send your brief

Read next