Lifecycle with a team of two

Ten or more lifecycle programs run concurrently at Forbes on a team of two: BDM, high net worth, consumer, onboarding, reengagement, Forbes 250, and the DanAds self-serve product among them. People assume that means automation did the work. Automation did some of it. Most of it was writing things down before we needed them.

The first thing that got written down was a standard. A lifecycle program here has a defined shape: an entry trigger, a sequence with a set number of touches, a suppression rule, a routing rule for anything that turns into a sales conversation, and a reporting view that ties sends to pipeline. Once that shape existed, adding a program stopped being a design problem and became a fill-in problem. The DanAds lifecycle I built solo is eight programs and twenty-nine emails, plus a revenue model and a 2027 forecast that is now in use for a vendor contract renegotiation. One person could build it because it did not have to invent its own shape.

The second thing was a QA checklist. Every send goes through the same list: links, tokens, suppression, rendering in the clients that matter, the routing test, the seed. The checklist is boring, and it is the reason a two-person team can ship at volume without the failure that ends careers, the wrong email to the wrong list. A checklist also means the second person can cover the first. When I built leave coverage and an onboarding playbook, most of the work was already done, because the checklist was the job description.

The third thing was the prompt library. Before any AI touched production work, the recurring writing tasks (subject lines, variants, summaries, the first draft of a nurture) had prompts with rules and examples attached. Custom GPTs on ChatGPT Enterprise then took the trainable share of that work. The order matters: the standard, the checklist, and the prompts existed first, and the AI inherited them instead of improvising around their absence.

The fourth thing was the two-bucket contractor model. Work splits into trainable execution and specialized expertise. Trainable execution, building an email from an approved brief, loading a list, running the QA list, can go to a contractor or an agent with the playbook in hand. Specialized expertise, deliverability, the attribution model, the routing logic into Salesforce, stays with the two of us. Drawing that line on paper is what let me stop asking for headcount for the first bucket. The ask usually shrinks once the playbook exists.

That is the honest answer to how two people run ten programs. There is one more piece, and it took the longest to learn: judge a quarter by what got added to the machine rather than by what got sent. A program that runs itself next quarter is worth more than a campaign that spiked this one. The team stayed at two because the machine kept getting bigger.