Agents, dual-WAN failover and outage credits: three upgrades for a growing WiFi network

Most hotspot operators hit the same ceiling. Sales depend on you being at the shop. Revenue stops the moment the upstream link dies. And when there is an outage, angry customers are handled one WhatsApp message at a time — badly, and only the ones who complain loudest.

Three capabilities break that ceiling: an agent network that sells while you sleep, a second WAN link that carries the network when the first one fails, and automatic outage credits that make it right when service is lost anyway. Here is how each one works and why they belong together.

1. Agents: let other people sell your vouchers

Every shop, kiosk and barber near your coverage is a potential sales point. The problem has never been finding sellers; it has been the accounting. Who took how much stock? Who has paid? Who is quietly selling cards and pocketing everything?

A float-based agent model settles all three. The agent tops up a float balance with you — by mobile money, from their phone. When you issue them voucher stock, their float is debited the face value minus their commission. An agent on 10% who takes 10,000 shillings of stock pays 9,000. They sell at face value and keep the difference.

The important consequence: the margin is settled at issuance, so there is no payout to chase later. You are never owed money by an agent, because they bought the stock before they sold it. No commission run at month end, no arguments about which codes were sold.

What the agent sees

Agents get their own portal login — separate from your admin panel, with nothing of yours exposed. From it they can:

• See their float balance and top it up by mobile money without calling you
• Browse the voucher batches issued to them, and the codes inside each
• Track which of their codes have been redeemed, and when
• Export batches and sales to Excel for their own records

On your side, every voucher stays attributed to the agent who took it, so per-agent performance is a report rather than a guess. Retire an agent and the record survives: the float ledger stays auditable and old vouchers still resolve to their name.

2. Dual-WAN: stop selling one link's downtime

A single upstream link means your business has a single point of failure that you do not own. When the fibre is cut or the ISP has a bad afternoon, you are refunding customers for something you cannot fix.

A second link — LTE, a different fibre provider, a wireless backhaul — solves it, in one of two modes:

• Failover. The backup sits idle and takes over only when the primary dies. Simple, and the cheapest way to buy insurance on a metered LTE SIM.
• Load balancing. Both links carry traffic at the same time, split by a weight you choose. Give the fibre a weight of 2 and the LTE a weight of 1, and the fibre carries twice the connections. You get more total capacity and resilience, because if one link drops the other absorbs everything.

The detail that makes or breaks it

Naive failover watches the gateway — the modem in your rack. But the most common real-world failure is an upstream outage behind a modem that is still powered on and still answering. The router sees a healthy gateway, keeps routing into a black hole, and your customers are offline while your dashboard says everything is fine.

The fix is to probe past the modem: point each link at a public internet host and route the check recursively through that link. If the host stops answering on WAN 1, traffic moves to WAN 2 — even though the modem never went down. That is the difference between failover that works on paper and failover that works at 8 p.m. on a Friday.

One more setting worth understanding: how connections are pinned to a link. Balancing per packet breaks things — banks, payment pages and game servers see your customer's traffic arriving from two different public IPs and end the session. Classifying by the source and destination address keeps all of a given customer's traffic to a given host on one link, so logins survive.

3. Outage credits: the answer when you do go down

Even with two links, you will have bad days. Power fails, a tower goes down, a switch dies. What separates a network people stay with is not never failing — it is what happens next.

Most operators handle it reactively: whoever complains gets an extension, everyone else silently loses the time they paid for and remembers it at renewal. It is unfair, it is inconsistent, and it rewards the loudest customer.

The alternative is to treat outages as ledger entries. Every time a router goes offline and comes back, the gap is timed and recorded against that router: when it started, when it ended, how long it lasted. Short blips are logged for visibility but not acted on. Anything past your threshold becomes an actionable outage — credited automatically, or held for you to review, whichever you prefer.

When a credit is processed, each affected customer is handled according to their state:

• Still active: their expiry is pushed out by the lost minutes. They simply get the time back.
• Already expired: a refund voucher is issued for the lost time, so the credit is not silently wasted on someone whose plan ended during the outage.
• Either way: an SMS tells them what happened and what they got.

Two practical touches matter here. First, quiet hours: crediting at 3 a.m. means texting hundreds of people at 3 a.m., so credits can be deferred to a sensible window. Second, an audit trail — a per-customer record of who got what, which is exactly what you need when someone asks whether they were compensated.

The commercial argument is simple. Proactive credit costs you time you already failed to deliver. Losing the customer costs you every renewal after that.

Why these three go together

They are the growth loop for a small network. Agents widen your reach beyond the people who walk past your shop. Dual-WAN protects the revenue those agents generate, because a wider customer base makes downtime more expensive, not less. And outage credits keep customers through the failures that remain, so growth compounds instead of leaking out the back.

Run one without the others and it shows. Agents selling into an unreliable network just distribute your reputation problem faster. Perfect uptime with no sales channel is a network nobody knows about.

Getting started

None of this requires new hardware. Agents are set up in your admin panel — add the person, set their commission, issue their first batch. Dual-WAN is configured on the MikroTik you already own, from the router's page in the dashboard. Outage credits need one decision from you: the threshold at which an outage becomes creditable, and whether to credit automatically or review first.

All three ship on every plan, with no extra licence. Next up: read our guide on pricing time and data vouchers so your agents have something worth selling.