By: Eva Keller
A brand raises its Meta budget, and the platform dashboard reports more purchases. The Shopify revenue line stays where it was. Krista Karpan and Nicolette McIntyre, co-founders of the marketing agency Milked Media, argue that this pattern is rarely an ad account problem. When spend climbs and sales do not, the cause usually sits in three places the ad platform cannot show: how the brand measures success, what it gives customers to get excited about, and what happens after the click.
Milked Media is a team of ten working with fashion, jewelry, beauty and home brands across paid social, email, Google and creative production. Krista, the agency’s CEO, has a clear process for diagnosing a plateau. Nicolette, on the other hand, has seen how clients misread their own numbers and how the agency keeps them from doing it.
Measure the business, not the platform
Krista’s first move is to stop treating Meta’s report as the verdict. Her team evaluates overall Shopify revenue, compared year over year, because seasonality makes month-to-month comparisons unreliable. A November will always beat an October.
The deeper problem is attribution. “Meta reports on a one-day view, seven-day click,” Krista said, a window that fits poorly with high-consideration purchases. Many of her clients sell fine jewelry. “I’m probably not going to click on an ad and buy a $5,000 ring within three days,” she said.
The failure she most wants to catch is divergence between what Meta reports and what the store earns. “It doesn’t matter that Meta is reporting more purchases. If you’re not bringing in more money, something is broken,” Krista said. She would “almost rather see” the reverse, with Meta performance looking soft while overall Shopify revenue rises.
To get around attribution, the agency tracks marketing efficiency ratio, or MER, which compares total ad spend with total revenue no matter which platform claims the sale. It “takes attribution out of it,” Krista said. Meta now points in the same direction. Its conversion lift tests, once available mainly through a Meta rep, compare people who saw an ad with people who did not, and they are open to any advertiser who meets a minimum spend.
Clients still tend to fixate on platform ROAS, Krista said, because it is easy to read. Nicolette sees the daily version of the same habit. “We have clients that are in the account every day just nitpicking,” she said, reacting to a down day after an up day. Her agency’s answer is education. “We really try and train our clients to look at the bigger picture and a more holistic picture.” Talking clients off the ledge, as Nicolette described it, is part of the job.
Give the algorithm something new to sell
The second place Krista looks is the product calendar. A brand stalls, she said, when it is “not really offering anything new or exciting to their audience.” The usual patch is a sale, and sales carry a cost. The brand ends up “training their customer to wait for sales.”
She points to Heretic, a perfume client. For a stretch, the brand launched little, and some products sold out, so the agency leaned on evergreen products. Once Heretic settled into a steady launch schedule, the ad account had fresh material to work with. Limited-edition messaging performs especially well. “It really creates that sense of urgency,” Krista said. A Heretic collaboration with the film Nosferatu also produced the campaign that won Milked Media the 2025 Meta Agency Award for Performance Campaign of the Year in the boutique agency category.
Newness only works if the brand knows who is buying and why. Krista said the jewelry category is crowded with similar-looking brands, and the ads have to answer a simple question: “give me a reason to shop from you.” Without a clear answer, she said, “the ads will struggle.” Audience data often corrects assumptions here. A brand that looks like it skews young can turn out to be bought mostly by women between 55 and 65, which changes the creative and the messaging.
Creative is the targeting, and the site has to close the sale
Meta’s targeting tools have shrunk to the point that testing audiences no longer pays. Comparing a 3% lookalike with a 1% lookalike “just doesn’t work anymore,” Krista said, and a new campaign structure is not worth the risk. “You’re probably going to just blow up the account.”
What replaced audience selection is the ad itself. Krista’s team keeps targeting broad, sometimes all women in the United States, then builds creative aimed at specific buyers. A mom-focused ad, for example, speaks to mothers. Meta reads the creative and decides who sees it. “You got to focus in on the creative,” she said. The agency is also leaning on user-generated content and partnership ads, which run from a creator’s account rather than the brand’s. Krista said Meta’s guidance has moved from suggestion to expectation. “It’s now like, no, you need to have it.” She added that Meta’s new Andromeda algorithm favors creative diversity, creative volume, and partnership ads. “If you’re not doing it, you’re almost actively working against what the algorithm wants.”
Strong creative still fails when the page it lands on is weak. “If we’re spending all of this money on ads to drive traffic to a website that is not set up for conversions, you might as well burn your money,” Krista said. Every new client gets a website audit in week one, covering the homepage, collection pages, product pages, and checkout. For fine jewelry, the product page has to carry the sale. That means a ring size guide, photography showing the piece on a hand or around a neck, reviews and real customer stories, and a visible payment-plan option for expensive items.
The counterargument: the ad account still matters
Nothing here means the ad account is irrelevant. Krista said some real levers remain, such as optimizing for the value of conversions rather than their volume. Her point is about order of operations. A team that adjusts bids and budgets before checking measurement, product newness, buyer clarity, and page conversion is adjusting the wrong thing.
Nicolette adds a fourth check: concentration. Brands that depend on one channel carry risk no dashboard flags. “You can’t just rely on one channel to be the backbone of your revenue,” she said. “It’s our most successful clients who show up in all of the channels equally and have that distribution of revenue.”
What to do first
For a brand stuck at a revenue ceiling, the diagnostic starts outside the ad platform. Compare Shopify revenue and marketing efficiency year over year. Check when the last launch happened and whether a sale is doing the work a product should. Confirm the brand can say why a customer should choose it. Audit the pages that follow the click. Only then does it make sense to open the ad account and test new creative.
Even then, Krista and Nicolette don’t treat testing as a phase that ends. Their first 90 days with a client cover the audit, budget planning, a profitability analysis, and heavy creative testing. Once winning ads emerge, the team can test less and scale more, but it never stops. “It’s continuous testing,” Krista mentioned.
The ad account reports what happened to the ads. A plateau is about what happened to the business.
