Oldcastle APG Mainspring Pilot
Trade School Enterprise Campaign Creation-at-Scale | February-March 2026
Mainspring’s first “official” run out of the gate was a big one.
One client, but three of their brands.
All told, it would be 964 assets in 47 days… a new record for us in terms of multi-brand complexity. The official Trade School case study says only 716 assets because project management lumped together several of the individual deliverables based on “ad units” – but the Creative team should get credit for 964 pieces of content!
We first heard about the possibility of the project in October 2025, but getting the MSA signed and the media plan locked took until January 30, 2026.
This was all for Q2 brand campaigns, so from the time we heard “Go!” it was a race.
My biggest pain point was not having a locked media plan… or locked templates. Without those, there was nothing the team could do to work ahead. Mainspring requires coding for everything, starting with the Components and Frames in Airtable and ending with the templates in Figma. I kept telling the larger team, “I need EXACT units and specs for us to build ANYTHING. We can’t work off ‘vibes’ or ‘guesses’ no matter what the client thinks the timeline should be.”
The Creative team was essentially “dead in the water” until both of those were approved.
This project took twice as long only because the clients pivoted on the template design in Round 2.
Otherwise, this was a very straightforward content-building campaign. Mainspring generated the RD1 kickoff deck based on the matrix; the team jumped in and made any necessary edits in Figma. That was then reviewed by the clients and feedback came in just like any normal campaign. Because all three were being done simultaneously, Jasmine, LaNorris and Rachel were each assigned a different brand to handle the feedback. Kris initially started with all the videos but we had to bring over freelancer Dan McCool from his work for us on The Home Depot to help Kris out.
The human touchpoints on a Mainspring-generated campaign are essential, especially at the final proofing and QA stage. Because platforms change their ad specs so often, we’ll always have human eyes do the final check against the latest specs documentation.
964 assets in 47 days
Q2 Campaign work for three brands in the Oldcastle APG portfolio: Catalyst (fencing), RDI Railing, and Sakrete (bagged concrete mix). The clients wanted two variations of every ad unit so the creative could be swapped out/refreshed mid-flight. So for every “A” version we created, there was a matching “B” version for the same audience.
All three brands asked for ads for every phase of the Sales Funnel (Awareness -> Consideration -> Conversion).
Catalyst and RDI only had two audiences to serve at each point in the sales funnel: Homeowners and Pros; but Sakrete also had Dealers to advertise to, which exponentially increased their number of deliverables.
This would be the first real pressure test of Mainspring and also the first time Trade School’s Creative team would work with the Media team from our sister agency, 22Squared.
964 assets were initially generated in under 30 minutes (between 7 and 10 minutes per brand) – so the “starter deck” was already populated before most of the team even got kicked off. That alone shaved 75% off the typical deck-building and concept development phase timeline.
This also marked the shift from clients proofing in a Google Slides or PowerPoint deck into proofing everything in Figma via a link, which saved a lot of download-upload time during the review and proofing stages.
As a tool, it was Mainspring’s first real pressure test, and we learned a lot.
For one, locked templates were absolutely necessary or no time savings would result. RDI changed templates in RD2, which meant the team had to essentially restart all the design work on that brand.
Catalyst’s templates were designed to use one long image across the background of their carousel units with transparent PNGs of a different fence in each frame – and Mainspring couldn’t handle transparent PNGs yet (it was on the roadmap). Which meant Mainspring would generate the background image, and a Designer still had to go in and manually add all the “hero” fence images.
For a first pass at “starter” copy, the AI-generated copy was acceptable – comparable to a junior writer’s first pass.
I built, dev’d, and tested all the templates between Figma and Airtable.
That was a lot.
But for a first test of the platform, I wanted to make sure <I> understood how it all worked before I brought in any other members of my team to take over the template design and development given the timeline. It would be too much pressure to throw onto anyone else at the speed it had to go.
According to the official Case Study put out by Trade School, the team delivered 716 assets in 47 days… but that’s because some of that total number is counting “ad units” rather than actual deliverables / pieces of content.
This unit count is a regular fight I have with Project Management because the Creative and QA teams have to touch and edit WAY more layouts than we’re given credit for. It makes a big difference to QA whether they’re reviewing “five META PAC carousel ad units” (5) versus “five ‘units’ in three aspect ratios with five frames for each ratio” (5x3x5=75)
Below are the actual numbers of final assets the team delivered.
964
Total Assets Delivered
47
Days from Kickoff to Final File Delivery
168
Videos (54 for Catalyst, 48 for RDI and 66 for Sakrete)
210
OLA Banners (28 for Catalyst, 56 for RDI and 126 for Sakrete)
586
Social Static Ads (224 for Catalyst, 146 for RDI and 216 for Sakrete)
