A creative refresh across hundreds of local ad accounts sounds simple until someone has to clone every ad, swap every page and URL, remember every location-specific exception, and then check the whole thing again.
At LT.agency, that used to be the job. For one 200-plus-location franchise client, a seasonal refresh could mean checking as many as 1,000 ads by hand. A single video swap could take eight to 10 hours. Larger refreshes could consume most of a work week.
Jess Petersen, VP of Digital Media at LT.agency, leads paid media work spanning programmatic, PPC, and paid social for multi-location clients. Her role puts her close to the part of advertising automation that matters after the demo: deciding which work can safely become a rule, which data has to stay clean, and which decisions still need a person.
LT.agency’s work with Fluency has cut some large creative-refresh workflows from roughly 40 hours to 15 to 20 minutes, according to the companies’ case study. But Jess’ account of the shift is less about handing campaigns to AI than about moving people away from repetitive execution and toward the places where judgment still matters.
Table of contents
- From 1,000 manual checks to a spot check
- The hard part is the exceptions
- Automation changed what the team could test
- Rules first, AI second
- The time goes back into judgment
From 1,000 manual checks to a spot check
Before automation, a new creative batch for a franchise client meant cloning ads for each store, changing the Facebook Page and URL attached to each one, and then verifying the result. The work was repetitive enough to invite mistakes and large enough to make quality control its own project.
“Then we had to quality-control potentially 1,000 ads individually to make sure every page and URL was correct,” she said.
The risk was not abstract. Meta could freeze during edits. A person could update four out of five ads and miss the fifth. Even when nothing went wrong, the team still had to prove that nothing had gone wrong.
Fluency changed the unit of work. Instead of recreating the same ad hundreds of times, LT.agency can place new creative into a single source, match it against the correct Page ID and URL, and let the system propagate the change across campaigns.
Jess said the team was more exhaustive during the initial build. Once the underlying data structure had been validated, the review process changed. “We still verify the work, but now it’s a spot check rather than manually reviewing every ad,” she said.
The hard part is the exceptions
The most dangerous part of multi-location advertising is often not the standard setup. It is the one store that needs something different.
A location may normally use a standard radius but require extra ZIP codes. Another may have a targeting rule that exists nowhere else in the account. Under a bulk-cloning workflow, the person doing quality control has to remember not only what the campaign should look like, but why one location is supposed to break the pattern.
“It wasn’t just a matter of checking whether the campaign looked right,” Jess said. “They also had to remember all of the individual exceptions and make sure they were reflected correctly across four different advertising channels.”
LT.agency now keeps those differences in a source of truth and pushes them across connected platforms. Jess said that changes the team’s responsibility from repeatedly implementing the same exception to validating the data that defines it.
That is a smaller operational task, but a bigger governance one. If the source is wrong, automation will reproduce the mistake quickly. If the source is right, the team no longer needs to rely on memory at every execution step.
Automation changed what the team could test
The operational savings mattered most when they changed what the media team had time to try.
For one homebuilder client, LT.agency had previously taken a broad approach because keeping many individual communities updated was too labor-intensive. After automating more of the campaign structure and inventory-feed work, the team could run more segmented campaigns within the same budget, build ads for individual communities, and test more audiences.
The case study reports a 456% year-over-year increase in form submissions, a 71% reduction in cost per lead, and a 14% reduction in cost per click for one multi-location client. Jess cautioned against reading the result as a simple budget story. “A lot of the increase in leads wasn’t tied to a budget change,” she said. “We weren’t spending more to generate those leads.”
Instead, the reclaimed time opened up more testing. One example was landing-page depth. Rather than sending every click to a broad page and asking visitors to navigate from there, the team could route people directly to a relevant home or community page.
“The time we saved with Fluency gave us the freedom to test new ideas and strategies,” Jess said.
Rules first, AI second
Jess’ comfort with automation depends on what the system is allowed to decide.
“My trust in Fluency comes from the fact that I’m not giving it free rein to go build whatever it interprets from a request,” she said. “It comes from having a strong data structure and clearly defined controls.”
That means defining the campaign inputs, the must-haves, and the exclusions before optimization begins. Once those controls are in place, the team can allow more automated recommendations, such as geo adjustments or budget-shift suggestions, where the client setup permits them.
The permissions are not identical across accounts. A franchise client may prohibit moving budget between campaigns, so that option stays off. A homebuilder client may manage budget at the division level, making the same action acceptable there.
“We know at the client level what they’re comfortable with,” Jess said, “and we’re not allowing decisions that don’t align with those expectations or with how we’re actually supposed to be running the media.”
The time goes back into judgment
The biggest change inside LT.agency is what happens after the repetitive work disappears.
Jess said the team used to spend so much time looking backward at reporting and troubleshooting that there was often little room left to act on what it learned. Automation did not remove the need for people. It changed what people could spend their time deciding.
“Now we spend much less time looking backward and much more time thinking about what’s next: What’s the next innovation? What’s the next test?” she said.
For franchise clients, that can mean more differentiated seasonal messaging by location. For homebuilder clients, it can mean more audience tests and smaller inventory segments. The work is still campaign management, but the bottleneck is no longer the number of repetitive steps a person can complete in a day.
That is also why Jess’ line between automation and judgment is useful beyond this one platform. The safest use of automation is not to ask software to decide everything faster. It is to remove the execution work that does not need human interpretation, then spend the saved time on the decisions that do.
Jess has a parallel version of that instinct outside work. She plays tabletop games, including Magic: The Gathering and D&D, and keeps a standing Wednesday game with friends. “Both at work and in my free time, I like having something to figure out and a problem to solve,” she said.
The tools can handle more of the moves. The point is to give people more time to think about which move to make.
