Twindo Ad Campaigns

Overview

Twindo needed refreshed creative for Meta and Performance Max campaigns, but conventional location filming, motion graphics, and digital effects would have made the volume we needed slow and expensive. I developed an AI-enabled production workflow that let the team create this kind of advertising roughly ten times faster and at a fraction of the traditional production budget.

The work was part of a broader effort to integrate AI tools across marketing and the organization: using them where they could remove production bottlenecks without giving up judgment, accuracy, or final craft.

A complete story in 15 seconds

Most impressions and performance were concentrated in the 15-second format, leaving very little room to explain a technical product. The priority was still to tell Twindo’s primary customer story in full: scan a real space, create a digital twin, and receive a fully edited CAD design.

For the Plan to CAD offer, the sequence became even more direct: submit a floor plan and get a clean CAD model back. Each ad had to make the transformation legible immediately, even when viewed silently in a fast-moving feed.

The square Plan to CAD cut leads with the deliverable, aimed at the audience that cares most about what lands in their CAD software.

Building the AI workflow

I used ChatGPT to develop and refine storyboards, current video-generation models including Seedance for footage, and Suno and Gemini’s music tools for soundtrack exploration. AI could accelerate the raw production dramatically, but it could not—and should not—do the whole job. The strongest material still needed to be selected, assembled, corrected, and finished in After Effects.

The most difficult part was not generating motion; it was getting generations that were recognizably on brand and showed Twindo’s work accurately. Over the course of the campaign, I improved both the prompting system and the reference-image setup. The discarded footage shown below represents only a fraction of what I generated and sorted through while honing the process.

That first campaign required more experimentation, but the reusable workflow made subsequent ad-generation projects dramatically shorter and gave the marketing team a practical foundation for producing more creative without scaling the budget at the same rate.

Static and Performance Max

The static system used the same visual logic as the motion work. Cyan represents captured space; magenta and violet represent the editable digital environment Twindo creates. Consistent framing, type hierarchy, and product treatment allowed the ads to flex across Performance Max and paid social while still reading as one campaign.

Results

The generative-AI creative refresh contributed to an 8.09% click-through rate from May through July 2026, compared with a typical 0.8–1.2% cold-audience benchmark. Just as importantly, the project established a repeatable workflow for generating, evaluating, and finishing new campaign concepts far faster and for a fraction of the conventional production budget.