The most common way to build a campaign is this: pick a list, assign an agent, start calling. When the round finishes, read the report. Non-responders simply stay on the list.
Yet most of a campaign's value is created not in the first round but in the second step taken based on the first round's outcome.
What a chain is
A chain makes the output of one campaign the input of another. A simple example:
- Round 1 — calls. A thousand people are called.
- Responders are split by outcome: interested, not interested, call back later.
- Round 2 — message to non-responders. A short WhatsApp message goes to those not reached.
- Round 3 — calls to message responders. People who replied enter a new calling round.
- The interested go to the team, the uninterested to a holding pool.
Each step's input is the previous step's result. Nobody filters lists by hand.
How long to wait between steps
The most frequent mistake with chains is compressing the steps. Someone called and messaged on the same day experiences one irritation, not two touches.
Intervals that work in practice:
- Call → message: a few hours later the same day is fine; a written message does not intrude.
- Message → second call: at least one day.
- Second call → third touch: a few days, and usually anything beyond the third is unnecessary.
The most dangerous side effect: stale records
Chains are structures that keep running by themselves, which is good — until something breaks.
If a chain step fails, records waiting on that step pile up. When the fault is fixed, the chain resumes and hundreds of people who have been waiting for days get called all at once. The reason for the call is long stale; a customer is called about something they discussed a week ago.
We write this because we have lived it. A chain fix can revive campaigns frozen for days and start an unexpected wave of calls.
The rule: before repairing a chain, look at the freshness of what is waiting. How many records are pending, and how old is the oldest? Stale ones belong in a cleanup, not in a call queue.
Stop conditions
Every chain needs an exit. At least three conditions should end it for a given person:
- The customer responded positively and was handed to the team.
- The customer clearly said no. This must be recorded at account level, not per campaign; calling the same person from another campaign is the worst possible error.
- The attempt cap was reached.
Measure step by step, not in total
The report for a chained campaign is not a single conversion rate. Every step has its own, and the improvement opportunity usually hides in one of them: if the message step converts poorly the problem is the copy; if the call step converts poorly the problem is the agent's opening line.
A blended rate hides that distinction.
Tomorrow: when an agent's knowledge goes out of date — knowledge base freshness.