George Popstefanov's first computer was a Commodore 64 with a German manual. He learned German, then BASIC. Years later, the pattern repeated: confronted with a system he could not quite use, he learned its language and built a new one. In 2010 that system was the advertising business. Popstefanov left a $35,000 job after his employer, Range Online Media, had been acquired by iProspect. He no longer felt he fit. At 27, he put $960 into a new agency in Fort Worth and called it PMG.
The origin story has the satisfying thrift of a garage startup, minus the garage mythology. Early accounts included RadioShack, J.Crew and Travelocity. PMG reportedly made about $1 million in its first full year, $2.6 million in the second, $5.6 million in the third and more than $10 million after year four. It was cash-flow positive and profitable from the start. There is no heroic near-death pivot in the public record. What failed first was simpler: Popstefanov's fit inside an acquired employer. The company he built in response would spend years preparing itself not to create that feeling when it eventually became an acquirer.
An agency with an operating system
PMG plans and buys media, develops strategy and creative, handles search and social, produces content, analyzes performance, and now runs commerce and influencer programs. Its customers are enterprise brands with complicated businesses and even more complicated data: Apple, Nike, Dropbox, Experian, Sephora and Whole Foods are among the names PMG publishes. D CEO counted about 60 clients in 2024. The firm now says more than 1,000 people in 13 hubs manage over $7 billion in annual media spend.
That sounds like a large independent agency. The more revealing description is a services company organized around software. PMG's proprietary system, Alli, connects data, tools, partners and teams. It unifies reporting, supports forecasting and audience modeling, automates campaign operations, lets users query billions of rows in plain language and measures activity in real time. PMG says every client and every team uses it. Clients get access to the same environment as the agency, not a ceremonial dashboard with three charts and a login nobody remembers.
The problem PMG is solving is not a lack of marketing tools. Most big brands have too many. Media plans live in one place, sales data in another, retail signals somewhere else and creative approvals in an inbox that has become a small archaeological site. The delays between insight, decision and execution eat the value of the insight. PMG's pitch is coordination: a shared data foundation, a shared context and a shared loop from planning to action to measurement.
“Data is everywhere. Intelligence is not.”PMG's compact diagnosis of modern marketing
The distinction matters in a market full of holding-company networks, digital consultancies and performance specialists. PMG's alternative is neither a pure SaaS dashboard nor a loose federation of agency departments. It is an integrated service team using one operating layer. Alli includes human validation and approval gates, an unfashionably sensible acknowledgment that an AI can generate 5,000 ad variations before lunch and still misunderstand the brief.
The switch from patience to purchases
For 13 years PMG did not acquire another company. Popstefanov later explained the hesitation plainly: growing too fast could have ruined the culture. Then his mind changed. He began thinking less about being a good founder and more about becoming a good CEO. The question shifted from “Can we protect this?” to “Have we built enough structure to extend it?”
The buying began in late 2023. Dallas-based Camelot Strategic Marketing & Media brought video, over-the-top television and connected TV expertise. RocketMill added a full-service UK operation and a European beachhead. Together, those first two deals added roughly 400 people and more than $50 million in revenue to the system. RocketMill completed its rebrand as PMG in September 2024, a telling move toward one global operating model rather than a collection of agency nameplates.
In June 2025, PMG bought Momentum Commerce, whose Velocity platform tracks more than 37 million products and 880,000 brands across marketplaces. That deal created a formal commerce capability spanning Amazon, Walmart, Target and Instacart. In January 2026 came Digital Voices, a 70-person influencer agency with offices in London, New York and Costa Rica, plus its Chord and Composer campaign tools. Four deals in about 27 months filled four specific gaps: television, geography, commerce and creators.
What did this spree cost? PMG has not disclosed the purchase price of RocketMill, Momentum Commerce or Digital Voices, and no reliable total is public. That missing number is important. Readers can judge the strategic fit and the added capabilities, but not the return on invested capital. PMG remains privately held, with no published valuation and no audited public revenue statement. Anyone turning this into a tidy acquisition case study should leave that cell blank.
What customers actually buy
The commercial package begins with people: strategists, media buyers, analysts, creatives, engineers and commerce specialists. Alli supplies the connective tissue. In 2025 PMG added Alli Marketplace, an app exchange that launched with 29 foundational partners and more than 100 existing native integrations. Amazon, Google Cloud, LinkedIn, Reddit, Roku, Snap and TikTok were among the names. A brand can enable tools for audience enrichment, forecasting, creator discovery or creative generation inside its existing environment instead of beginning another procurement and integration cycle.
PMG does not publish a rate card. Enterprise agency work is scoped around markets, channels, media spend, data complexity, production and technology, so the honest answer to “what does it cost?” is “a negotiated amount.” It is also not presented as a swipe-a-card, self-serve SaaS product. A buyer is paying for an operating model and a team, not merely seats in Alli.
The evidence of performance is necessarily a mix of client results and agency claims. PMG says its BYD work lifted brand demand 114 percent in four months across Europe; its Travelex content overhaul increased traffic 33 percent year over year. The BYD campaign won a Shorty Award. The company reports 90 percent client retention across five years, while MediaPost reported 20 percent revenue growth in 2025. Those numbers do not make every campaign a winner. They do suggest the integration pitch is doing more than decorating proposals.
The parts worth stealing
A founder cannot copy PMG by naming an internal dashboard and ordering matching fleece. The useful moves are structural. First, put builders close to the work early. PMG began with engineers in its core, long before “AI agency” became a conference badge. Second, make the internal system useful enough that adoption is compulsory by gravity, not policy. Third, acquire distinct capabilities only after the operating spine and culture can absorb them. Finally, keep humans in the approval loop when automation touches a customer's reputation.
Connect planning, execution and measurement so each campaign makes the next decision better.
Embed engineering beside practitioners. The best workflow ideas usually hide inside repetitive work.
Choose acquisitions that add a capability or geography, not another overlapping org chart.
Automate volume and drudgery while preserving approvals for strategy, quality and brand risk.
This model also has conditions. It works best for organizations with multiple markets, large media budgets, fragmented data and enough repeated activity for automation to compound. It is a poor fit for a small advertiser that needs one campaign, a company without clean data or executive alignment, or a team seeking a standalone self-serve tool. Integration can remove handoffs inside PMG's system, but it cannot make a client agree with itself. Nor can proprietary software guarantee distinctive creative work.
Culture is the other claimed safeguard. PMG has appeared on Ad Age's Best Places to Work list for 11 consecutive years, most recently at number six for 2026. It runs graduate and technical development programs, employee resource groups and a Digital Career Accelerator with Tarrant County College intended to train more than 1,000 students by 2028. PMG says employees contributed 6,000 paid hours to more than 40 community groups in a recent year. These programs are not proof against the strain of integration. They are evidence that management knows what is at risk.
The most interesting thing about PMG is therefore not that a small Texas agency got large. Agencies do that, usually with money, mergers and a small blizzard of new titles. It is that PMG spent more than a decade building a common language before it started adding accents. The next test is whether Alli can keep strategy, creative, media, commerce and creators coordinated as the headcount and map expand. The company built the machine first. Now it has to prove the machine can digest what it is being fed.