The Influencer Program That Ran on One Person's Memory

The program worked. That was the problem.

One person at a client ran the entire influencer program. They knew the influencers personally and had built those relationships over time. They knew who had been paid, what the rates were, and what content was waiting on approval. The program produced content and drove sales, so from the outside it looked like a functioning operation.

That outside view was wrong. The whole thing was one person holding a set of relationships together in their own head.

When that person became unavailable

The exact circumstances don't matter here. The moment this person couldn't be in the seat, for whatever reason, the program stopped working.

Posts went live with nobody reviewing them first. Some of it hadn't been properly approved under the brand guidelines, and it went out anyway.

There was no record of which influencers had active agreements. Nobody had written down what each one had been paid, what rate they'd agreed to, or when a payment was supposed to trigger. Payouts ran through Refersion on a $100 threshold, but outside this one person, nobody could say with any confidence who was enrolled, where their payments stood, or what we still owed.

We didn't know what was live, what had been approved, or what we'd committed to pay, never mind who had already been paid.

The rebuild process

When you inherit a program with no documentation, you start from scratch and treat everything as unknown until you've verified it yourself.

The first step was a full audit of Refersion: who was enrolled, what their payout history showed, what was still pending. That gave us a list of influencers with at least some payment history behind them. It was a starting point, but nowhere near the full picture.

Then we built the structure that should have existed from the start: an Asana project with defined stages, running from outreach and agreement through content submission, review and approval, post live, and payment triggered. Each influencer relationship became a card moving through those stages, and every card carried the agreement terms, the payment rate, the content status, and the approval status.

The last piece was defining what "approved" actually meant: the review criteria, who held approval authority, what the turnaround time was. Before, those answers lived in one person's judgment. Now they were on paper, so anyone with access to the Asana project could make the same call.

The goal was to get the rebuilt program running by October 1st, in time for holiday season. That timeline was tight given that we were starting from zero documentation, but hitting it mattered because the influencer content pipeline needs lead time to build.

Why this pattern is predictable

The single-point-of-failure problem turns up in marketing programs at almost every company under about 50 people. Someone builds the thing because they had the energy and the relationships for it. They do it because they care and they're good at it, and because nobody else stepped up. It works partly on that person's competence and partly because the relationships are theirs personally.

But the program becomes dependent on their presence in a way that's invisible until they're not there. Nobody notices the fragility while it's working.

This is different from depending on a revenue-generating founder or a technical lead. Those dependencies are at least visible. Everyone knows the business would be in trouble if that person left. The influencer program dependency is invisible precisely because the program keeps working. Since it's working, everyone assumes no process is needed. What it actually means is that an informal process is quietly covering for the absence of a formal one.

The basic setup for an influencer program (an Asana project structure, documented rates and payment rules, an approval workflow, a current list of enrolled influencers) takes maybe two hours to put together properly. Those two hours are worth far more than the months of lost pipeline you get when the person holding it all together in their memory becomes unavailable.

Document it while it's working. That's the whole lesson, and almost nobody does it until after something has already broken.