Three roles, five years

I started underneath the product. The job was support to a partner channel manager - build the training material, run the admin, keep the data clean - and it ran about seven months. It also put me next to sellers early, and that is where I learned what selling in a digital market actually demands. Most of what I understand about how this business makes money came out of that year.

In June 2022 I took over our largest organic SEO partnership. It was a white-label reseller channel: a national partner sold our services under their own brand, to their own clients, and I owned that relationship end to end for the next two years and ten months.

The channel grew from 125 concurrent monthly orders to 333 in that time, and roughly 2.5x in revenue, at about 37% compound annual growth. Those are run-rate figures rather than trailing-twelve-month actuals. The final business review reported $4.7M in annual revenue and $1.2M for the quarter, across 300+ active clients and 300+ concurrent campaigns spanning 80 partner markets, with 35% growth over the prior year.

Most of the daily work was activation. The pre-sales function produced over 1,500 audits a year and roughly 85 new opportunities a month, peaking at 350 completed audits in February 2025 with 43% of them turned inside three days. On new business I was the product expert in the room. I prepared or rewrote the pitch deck, ran the pre-call session with the seller, and delivered most of the call itself. Follow-up, audit revisions and the handoff into onboarding were mine too. Those sellers worked for the partner, not for us. Getting a channel to sell your product well is a training problem before it is a sales problem, and it never stops being one.

Four semiannual business reviews went to the partner's executive team over three years. I built and delivered more than 90% of each one. They ran as a single continuous argument rather than four separate updates: here is the diagnosis, here is the goal, here is what happened, here is where we go next.

Our public position on AI in content changed three times while I ran the channel, and I wrote all three. No AI involved, then ideation and research only with human writers retained, then AI as fundamental to how the work gets done. Each version was true when it was written. The market moved underneath all three, and the job was to move with it without spending client trust on the turn.

None of this came with direct reports. The account managers servicing the book reported to other managers, and what I had instead was a standing recommendation on their performance that those managers acted on. Running a channel this size through people you neither employ nor manage is a different job from running one you do, and that is the job I actually held.

The hardest stretch was a churn spike in late 2023 and the program that reversed it.

In April 2025 I moved into a director-level seat covering sales strategy and operations for all of Voice Media Group, reporting into the C-suite and working alongside two VPs of Sales across a multi-division organization.

The first thing the job needed was a view of sales performance that did not exist. I asked IT to build it and was told no, so I built it myself: three systems that shared no common identifier, joined on a key I designed, with entity resolution for sellers holding multiple CRM identities. Roughly 145,000 activity records, 193 sellers, six years. It became the data layer under the dashboards leadership ran on.

That model answered a question nobody at the company had been able to answer: how long a new sales hire takes to pay for itself. Across the post-2020 hire cohort the median was seven months to the first revenue benchmark and about ten to sustained profitability. The benchmarks showed the brake on time-to-revenue was onboarding rather than hiring quality, so I rebuilt the 90-day onboarding program around that finding - the version it replaced had run once and run badly.

Enablement was the other half of the seat. I authored the recurring sales curriculum - prospecting, objection handling, interest-gathering, renewal calls - along with the visual and brand spec governing how that material got produced. The last thing I built was a multi-agent system that produces training material end to end. Every statistic has to carry a publisher, date and URL to clear its provenance gate, and a conflict-halt protocol escalates contradictions rather than resolving them itself. I derived its limits by running roughly twenty structured evaluation cycles against the spec, which is how I found an undocumented generation ceiling in one of the underlying tools and designed around it.

Two of those things outlived the job. IT held meetings to rebuild the reporting model after I left rather than lose it, and my successor kept running the training system.