Managing Fan Conversations: Triage, Priority, Backlog

An inbox at volume behaves nothing like an inbox at small scale. Below a few dozen live threads you answer everything and the sort order does not matter. Above that, you stop finishing the queue and start choosing from it. Every hour you do not choose deliberately, the platform’s default sort chooses for you. This guide is about that choice: what gets opened first, what gets left, and what happens to the large quiet remainder that nobody has time for.

What does managing fan conversations mean at volume?

It means running a triage, not a queue. The work splits into four decisions that repeat all day, and none of them are about wording:

• What to open. Which threads justify a human reading them right now.
• In what order. Which of the open threads gets the next reply.
• Where in each thread to push. Whether this conversation is ready for an offer.
• What to close out. Which threads you stop working, on purpose, with a record of why.

Handled well, these four are invisible. Handled badly, they show up as an inbox where the newest message always wins and every thread older than a day quietly dies.

How do you triage an inbox you cannot read in full?

By sorting on money and momentum rather than recency. Newest-first is the default in almost every tool, it feels productive, and it is the worst policy available: it guarantees that a thread that went quiet yesterday will never be seen again, because something newer always arrives on top.

A workable priority order:

Band	What it contains	Why it goes here	When it gets worked
1	Live threads mid-conversation, especially past the tenth exchange	Highest chance of closing, and latency kills them	Immediately, peak hours
2	Recent buyers still talking	Cheapest revenue you will ever get	Peak hours
3	New arrivals, first contact	The thread that does not start today rarely starts	Same session, short replies
4	Quiet and lapsed threads	Volume is large, per-thread odds are low	Scheduled slot, off-peak

Two rules hold the table together. Nothing from band 4 is allowed to eat peak hours, and nothing in band 1 waits behind a band 3 pleasantry. Sorting fans by value rather than arrival time is what fan segmentation is for.

Which conversation gets answered first?

The one closest to a decision, not the one that arrived last. Concretely, that means the thread where a real exchange is already running and the fan is answering in full sentences. That is a conversation with momentum, and momentum is perishable.

Signals worth routing on, in rough order of usefulness:

• Depth. How many exchanges deep the thread already is. Depth is what the timing rule below is built on.
• Purchase history. Someone who has bought before will buy again more readily than someone who never has. High-value fans (see whale) deserve a named owner, not whoever is on shift.
• Response latency. A fan replying within seconds is at their phone. A fan replying the next morning is not in a conversation, they are in a correspondence.
• Time of day. Our data shows evenings maximise conversion while the 2am-6am window collapses it. Overnight, keep threads alive; do not spend them.

When should a pitch land in a live thread?

After about ten exchanges, and never before the sixth. This is the most robust finding we have across 1.5 million messages: pitching before the sixth message drops conversion by roughly a third, and the optimum sits after roughly ten exchanges, worked through thread by thread in when to pitch a sale. Everything about the message is secondary to where it lands, because it is the one mistake that repeats thousands of times a week without anyone noticing.

Three things to avoid at the moment of closing, all measured:

1. Ending the sales message with an ellipsis. The worst-performing closer in the corpus.
2. Closing on a closed question. A yes/no at the moment of the offer costs several points of conversion: it hands the fan a one-word exit.
3. Placing a personal callback at the exact moment of the pitch. It costs points, and it is the most counter-intuitive result we have; personalising a sales message works through where the callback belongs instead.

What should you deliberately drop?

Threads that consume peak hours without ever converting, and one-word threads that have never gone anywhere. Dropping is a decision you make once and record, not a thing that happens by neglect. The difference matters, because a recorded drop can be reviewed later and a neglected thread cannot.

Reasonable candidates to stop working during peak hours:

• Threads that have absorbed long sessions repeatedly without a single purchase.
• Fans who only respond to broadcasts and never to a direct message.
• Threads where the fan has clearly said no to the whole category, more than once.

None of these should be deleted. They move to the tail, where they cost nothing to hold and can be reached with a broadcast. What you are protecting is the evening, and the evening is the scarcest thing the operation owns.

What happens to the dormant tail?

Nothing, unless it is scheduled. This is the most consistent operational failure in the trade: a chatter starting a shift opens the newest threads, because that is what the screen offers, and the older half of the inbox is never reopened by anyone. It does not decay loudly. It just sits there.

The fix is mechanical, not motivational:

• Reserve a fixed off-peak slot for oldest-first work, deliberately.
• Use a broadcast to make the tail move at all, then work the replies as real conversations.
• Measure the share of dormant threads opened per week. Unmeasured, it stays near zero.
• Treat lapsed subscribers as their own workflow, not as inbox overflow (see win-back).

It also helps to stop treating the tail as one undifferentiated mass. It is not:

Tail segment	Typical state	What actually restarts it
Bought once, then went quiet	Still subscribed, stopped replying	A direct reference to what they bought, off-peak
Talked a lot, never bought	Engaged, price-resistant	A lower tier, offered without a pitch attached
Lapsed subscriber	Gone from the feed entirely	A win-back workflow, not an inbox reply
Never replied at all	Cold from the start	A broadcast, and nothing more expensive than that

The first two rows are worth a person’s time. The last two are worth a scheduled send.

How do you tell a conversation problem from a volume problem?

Compare the same chatter across accounts and different chatters on one account. If everyone performs badly on one creator’s inbox, the problem is the brief or the price grid, not the writing. If one person is weak everywhere, it is a skills problem. If everyone is fine but revenue is flat, you have a coverage problem, and the answer is in the rota rather than in the messages, which is where running a chatting agency picks up.