A solar agency managed ads across 50+ accounts
A solar agency used data-triggered tasks to share media-buying work across about six buyers. An account change generated a specific job, and the agency distributed those jobs across the team. A buyer's time off became easier to cover because the work could go to someone else.
I worked with this shop as a consultant, helping build the data infrastructure and a dashboard that brought its portfolio into view. It managed 50+ accounts for separate solar customers, who together usually spent between $300,000 and $400,000 a month on advertising. Those customers paid for leads. The agency handled their Facebook ads.
That arrangement put the daily responsibility for campaign decisions inside the agency. Its buyers needed to act across a portfolio of accounts, and the dashboard let the team see which clients were doing well and which weren't. The next step was getting the resulting work to someone who could do it.
Corrective work still needed doing when buyers were away
A media buyer's client list and an agency's daily task list describe different things. One says which accounts belong to a person. The other says which actions need doing, account by account. Keeping every action attached to its usual buyer makes that person's availability part of the delivery process.
The solar agency had recurring corrections that could be expressed as tasks. A buyer might need to increase an ad set's budget, work on an individual ad or change a campaign setting. Those actions occurred across the book of accounts.
The agency used round-robin task distribution to make those actions shareable. The team could cover time off and other absences by distributing the generated work among its buyers. The change concerned who could receive the next job, while the task still concerned a specific customer's advertising.
Data triggers generated tasks for the buyers
Triggers monitored deviations from an account's average and anomalous results. A trigger generated a task when its condition was met. Make.com automations built the task list, which was then randomized and distributed to the media buyers.
The list covered around 80 kinds of task. Tasks included increasing an ad set by 10%, making changes at the ad level for a customer and adjusting campaign optimization settings. Each example names an action tied to an account, so another buyer has a specific job to carry out.
The work followed an ordered path:
- Account data met a trigger condition.
- The automation generated a task describing an action.
- The agency randomized and distributed the tasks across its buyers.
- A buyer carried out the assigned media-buying work.
The automation generated and routed work for people to do. The record doesn't establish that it made every advertising change itself.
I described this as round-robin media buying. The recorded distribution was randomized. It doesn't establish an exact buyer-by-buyer rotation or equal workloads. The operational point was that another buyer could receive an account's task when its usual buyer wasn't available.
The data tools and task automation did different jobs
I used Supermetrics to aggregate the portfolio data, BigQuery to store it and Google Data Studio to visualize it. The business owner wasn't interested in a large infrastructure budget, and Supermetrics made aggregation affordable for that project. Make.com generated the media-buying tasks.
Those tools answered different operational questions. The dashboard gave the agency a combined view of its clients' performance. The task automation turned detected conditions into work that could be distributed. Both were part of managing the same portfolio, but a chart showing an account's performance and a task asking a buyer to change something serve different purposes.
The available project record doesn't include the exact query, trigger thresholds or connection sequence between these tools. Rebuilding this workflow would require those implementation decisions. A stack list alone wouldn't supply them.
Round-robin distribution made the team easier to manage
This was the agency's first use of a round-robin approach for its media buyers. I reported that it made management much easier and gave the team flexibility when buyers took time off or were otherwise unavailable. The documented result is qualitative. The project record supplies no measured reduction in response time, cost per lead or staffing cost.
The portfolio's buying playbook also gave the work a repeatable basis. Buyers compared the current day's performance with the previous three days and the previous seven days. They kept ads that were getting better and stopped ads that were getting worse.
That was the working playbook I described, with buyers making the comparisons. It shouldn't be read as a complete set of automatic trigger rules or a universal instruction to stop any ad after a bad day. The account doesn't specify the thresholds or exceptions that would make such a rule safe to reproduce.
The team could distribute actions across the portfolio
The change gave managers a different unit of work to assign. A whole account contains many decisions and a history of previous decisions. A task can describe a bounded change within that account. The agency could distribute those changes across its buyers without requiring one person to do every action for a given customer.
That's the staffing implication of the workflow. Coverage still depended on buyers understanding the work they received. Sending a task to another person doesn't, by itself, establish that the person has the knowledge or authority to complete it.
The combined dashboard contributed visibility across the portfolio. Managers could see which clients were performing well and which weren't, while the generated task list identified actions arising from monitored conditions. An account could remain visible to management even when a particular buyer was away. The record supports that operating model, without supplying a count of missed tasks before and after.
Shared task definitions made flexible assignment possible
The useful part to carry into another agency is the definition of the work. Increasing an ad set by 10% describes a bounded action. Asking someone to improve a customer's advertising leaves the next person to diagnose the account and decide what to do. Those requests need different amounts of context, even if both appear as a single task on a board.
My recommendation is to separate repeatable actions from decisions that require deeper account knowledge before sharing the queue. A qualified buyer should be able to identify the account, understand the requested action and know when to stop and ask for a review. Those are requirements for adapting this approach, rather than features documented in the original automation.
Random distribution also leaves questions about workload open. Tasks can require different amounts of effort. A randomized list provides no evidence that buyers received equal amounts of work or that every task had equal priority. Those questions need their own rules and measurements.
This approach fits agencies with repeatable work and qualified buyers
Round-robin task routing is worth considering when an agency runs similar campaigns across many accounts and has buyers who can cover one another. The solar agency had both a portfolio-wide view and recurring actions specific enough to generate as tasks. Its reported gain was easier management when staff availability changed.
I'd start an assessment by checking whether another buyer could complete a recurring action from the task and its account context. A job that still needs its original buyer to explain every decision isn't ready for shared assignment. Improving the task definition comes before changing who receives it.
A small shop with a few clients and one buyer has nobody else to route work to. I'd keep its process simpler. The reason to adopt this model is a real coverage need and enough repeatable work to support it. Another qualified buyer must be able to take the task.