BlogBuild on

Don't rebuild. Build onhow creative teams turn every ad into the starting point for the next

8 min read

Coming soonThe Meta Ad Library on the canvas is coming soon.

In November, Liz DeAngelis of Brainlabs explained why she wasn't ready to let Meta generate her clients' creative and keep adjusting it. A brand in that position, she said, "might know that performance improved without knowing what actually improved it", and whatever Meta learned would stay inside Meta, out of reach of the TikTok plan, the YouTube cut and the next brief. "I'm just not ready to outsource the learning."

Her objection reaches well past Meta. Every campaign teaches a marketing team something about its product and the people who buy it: which opening held, which shot of the product reads on a phone, which wording legal will sign. Most of that ends up in a file name, a comment thread or the head of whoever made the ad. The case for building on ads instead of rebuilding them is the case for keeping that knowledge where the team can read it, change it and take it to the next brief.

Every decision your team makes stays in the workflow

A finished ad is hundreds of small decisions: how a prompt was worded, which model rendered the product best, which of six takes went out, where the logo sits in the 9:16 version. When the whole ad is built on one canvas, from the first image to the export, each of those decisions sits in the step where it was made. Turned into a workflow, that canvas carries all of them into the next ad.

On the next brief, the workflow gets new material: next week's gameplay, a new offer or a new market. Motn rewrites the prompts for it and sets the steps up again, and your team reviews each step and picks the takes. Every fix made along the way goes back in. A workflow is never finished. It's as good as everything your team has taught it so far, and it gets better with each brief.

How this differs from agents, apps and templates

Most AI ad tools handle the workflow in one of three ways, and each leaves the team with less control over the steps and less of what it learns.

Agents that do the work for you. Sett's agents generate game ads, put them into campaigns and learn from the results inside Sett's own system. Layer's Creative Agent builds a workflow from a brief. Meta's Advantage+ creative tools change ads on their own, which is how REI's approved bike photo got a second set of handlebars. With an agent the team gets the ads but not the steps behind them, so it can't fix one step and keep the fix. The full argument is in why a canvas, when an agent can make the video.

Workflows published as apps. Layer, Figma Weave, Krea, FLORA and Magnific let the person who built a workflow publish it as an app with inputs and a Run button. Layer's teammates run it "without opening the canvas", and in FLORA the people running a saved workflow see the graph read-only unless they use Detach, which turns it into an editable copy. The people running the app mostly work through a form, so what they notice tends not to reach the workflow unless someone opens it up.

Templates that repeat the same steps. Runway saves workflows as templates the team runs again. A template repeats the steps it was saved with, and improving it is a separate job someone has to remember.

Some of these canvases stop at images or clips and leave the cut, the sizes and the export to another editor. Layer's Video Editor node and FLORA's Timeline now assemble finished videos too, so the real difference is whether the sizes, languages and fixes live in the same workflow the team edits.

Motn keeps the workflow an open canvas for the whole team. Everyone on the plan sees every step and can change any of them, and each change stays for the next run. Motn sets the steps up again for new material, but nothing goes out that someone on the team hasn't looked at and chosen. The canvas also runs all the way to the finished ad, with the cut, every size and every language, so everything the team decides about an ad lives in one place. The comparisons go tool by tool: Layer, Figma Weave, FLORA and Sett.

Shared libraries: everything your team approves, ready for everyone

Everything your team approves goes into shared libraries. The saved branding is there, with fonts, colours, logo, type sizes and motion style. So are approved footage, gameplay captures, product shots, music, end cards, legal lines, and the openings, titles and transitions the team has made.

Anyone on the plan can pull from the libraries into any ad. A junior designer's ad starts from the senior designer's decisions on the first frame. Last month's winning opening can lead this week's ad. The disclaimer legal approved goes into every version word for word. Up to 15 people share one plan with free seats, so any approved piece can go into anyone's ad the day it's approved.

What each person brings stays with the team

Everyone's work on an ad ends up in the same workflow. The designer's look and motion style are there. The project manager runs reviews on the live canvas, so comments and sign-offs stay with the work. The UA manager has the Meta Ad Library plugged in, with ads that have run in the category for 30 days or more next to the team's own. Legal's approved wording sits in the library as live text.

Nobody writes any of this up separately. It lands in one place, so every later decision, from the next opening to test to the offer for a new market, is made with all of it in view. When someone leaves or a new designer joins, the knowledge stays: the new person opens the workflow and sees every step, prompt and model the team has used.

What the workflow learns from tests and mistakes

Because every step stays on the canvas, the workflow also keeps what tests and mistakes taught the team. When the ad platform shows which opening won, the winner goes into the library and leads the next round, and the openings that lost stay on record, so the next test starts from what the team already knows. When a model keeps getting something wrong on your product, such as redrawing a letter in the logo or cutting the price out of the 9:16 version, the fix the team found stays in that step, and the next brief doesn't repeat the mistake. A lucky take is kept the same way. The step that produced it holds the exact prompt, model and source image, so the team can use it as a starting point for related shots, even though a new run won't repeat the take exactly.

The platforms are offering to keep it for you

The pressure to hand all this over is commercial. Meta's stated aim is to automate ad creation, targeting and budgets end to end, and its generative tools reached a reported $10 billion run-rate. Agent vendors make the same offer from the other side: Sett's agents learn from campaign feedback inside Sett. Either way the lesson is real, and the brand can't read it. Knowledge kept on the team's own canvas can be read, changed and taken to the next channel, which is the whole of DeAngelis's point.

Parts that already worked become new ads

A workflow reruns on new material, and the libraries add something else: any part can be connected to any other. The opening that held people last month can lead this week's product shot, and a logo reveal can play inside a 3D scene from another project as an end card nobody has seen. A golf game can join gameplay from one hole, the week's leaderboard, a score counter from the motion library and a "Beat this score" headline into a 15-second format no template has. The team decides which parts meet, because it knows why each one worked the first time, and AI does the assembly. Each part already passed review, so the new ad needs a much shorter one. Worked examples are in connecting the nodes.

The senior's decisions in everyone's output

Most of a team's ads are now made by juniors, marketers and AI, and the senior designer can't check 48 versions a week frame by frame. A brand guide records the colours and fonts but rarely how they're used, and motion drifts first, because every animator rebuilds it with slightly different values. So the senior sets the brand library once and builds the logo reveal, the end card and the title animations as live code with their easing and timing intact. A junior who drags them into a new ad gets the senior's motion, not an approximation of it, and when the senior spots a problem, the fix goes into the library rather than into a comment on one file.

Reviews on the live canvas

A workflow only keeps what reaches it, and feedback sent on exports never does. In a StreamWork survey of 500 US marketers, more than 60% spent up to a day a week chasing approvals. When the brand lead, legal and the UA lead review the canvas the designer is working on, each comment sits on the step it's about, the takes that weren't picked are visible next to the one that was, and a wrong price is fixed while the reviewer is still looking. Agencies that review with clients will find the rest in Motn for agencies.

Takes side by side, and the chosen one feeds the next step

Every step on the canvas can give several takes at once, so the team compares them and picks before the next step spends a credit on the wrong one. The chosen product shot is what the motion step animates, and the chosen motion is what goes into the cut. The takes that weren't picked stay beside it, so a reviewer can see what was tried and why.

Free seats, so the whole team can be in it

All of this depends on the people who make the calls being on the canvas, and per-seat pricing keeps them out. A team of eight on FLORA's Pro plan pays $400 a month in seat fees, which include a pooled usage allowance, so the designers get seats, and everyone else can comment on a shared link but can't change the work. Motn includes up to 15 people who can work in and change the workflow, on one shared balance, and credits go only on generation; editing, comments, previews and storage cost nothing. The argument is made in full in why every reviewer should have a seat.

Starting points that become your own

A team that starts from a ready-made logo reveal or a kinetic type layout doesn't have to stay with it. The starting point opens on the canvas as live code inside the brand library, the senior adjusts its motion and layout, and the first ad built on it is saved as the team's own workflow. Every fix after that goes into the team's version, not the template, as making a starting point your own shows.

Where to start

Start with the ad your team makes most often: the weekly refresh, the monthly event or the store preview you update with every release. That's where knowledge builds up fastest, so it's where keeping it pays back first. Make it once with this week's material, save it, and run next week's brief through it. The $10 Try pack is enough to do that on your own footage, or book a demo and bring the brief.

Make yours on Motn.