A B2B travel portal for sub agents, next to your public site

It usually starts on WhatsApp. A sub-agent in Sharjah sends a screenshot of a Dubai–Lahore fare and asks whether you can do better. You check, you quote, they confirm, someone on your team issues it, and the payment turns up three days later, maybe. Multiply that by twenty agents and a busy week and the business you actually run is a chat queue with a ticketing desk behind it. A B2B travel portal for sub agents is the way out of that queue — but only if it sits on the same inventory as your public website without ever letting the two see each other's price.

That last condition is the whole article. Putting agents on a login is easy. Keeping a traveller from finding your net-plus-small-markup price, and keeping an agent from finding a public price lower than the one you charge them, is where the design decisions are.

Why the login matters more than the logo

Plenty of agencies treat the sign-in screen as a courtesy: agents get an account so their bookings are grouped together. The real reason for it is contractual.

A lot of wholesale hotel content and a lot of negotiated fares are sold to you on the understanding that the rate goes to the trade, not to the open web. Some contracts say it in so many words; others enforce it quietly, by watching which rates turn up in public metasearch and cutting off the account that leaked them. If your agent prices can be reached by anybody who types a URL, you have put the supplier relationship at risk on behalf of a sub-agent who books five tickets a month.

On this platform a storefront can be set to require sign-in before any search runs, and it can register business customers — agencies — separately from individual travellers, with an optional manual check before an agency account goes live. That check is worth switching on. It is the moment you confirm the company behind the sign-up is a real agency and not a competitor or a price-scraper who filled in a form.

One inventory, two audiences: what actually differs

The inventory is the same. Everything around it is not. This is the part worth printing out before you configure anything, because each row is a decision somebody on your team will otherwise make by accident.

Reasonable people disagree about the second row. Some agencies show agents the net and let them add their own markup on top; others show a single agent price and keep the net private. The second is safer when your agents also resell to each other, because a net that leaks once is a net every agent in the city knows by Friday.

Markup per agent, without a spreadsheet per agent

The spreadsheet version of per-agent pricing is familiar: a tab per agent, a column per airline, a colleague who remembers which ones are out of date. It works until the colleague goes on holiday.

Here markup rules attach to office categories rather than to individual people. Every agency is an office; your public customers sit in an office of their own; a category groups offices that should be priced alike. A rule can narrow by carrier, route, country, trip type and booking or travel dates, and it can be a percentage or a fixed amount, with commission and discount fields beside it. So in practice you build tiers — a category for your large agents, one for new ones, one for the agent you have given a special rate on one airline — and move an agency between them instead of rewriting its prices.

A single agent who genuinely needs its own pricing gets its own category. That should be the exception. Twenty categories for twenty agents is the spreadsheet again, only harder to read.

The public office has its own rules too, and that is what keeps the two sides apart: an anonymous visitor is priced as the public office, a signed-in agent as its own office's category. Check both prices for the same flight on the same day before you invite anybody. If the agent price is ever higher than the retail one, you will hear about it — from the agent, on WhatsApp, with a screenshot.

Credit limits and settlement

Credit is where a sub-agent operation makes or loses its money, and it is the one thing that should never live in a spreadsheet.

Each agency carries a credit limit. When one of its users issues a ticket or a hotel voucher, the platform reserves that amount against the available credit before it asks the supplier to issue, and refuses the issue if there is not enough left. The reservation is atomic, which matters more than it sounds: two staff in the same agency issuing at the same minute cannot both squeeze under a limit that only covered one of them. The hold is released whether the issue succeeds or fails, so a supplier timeout does not leave an agent's credit frozen.

Settlement runs on invoices. An invoice is a monthly snapshot per agency and currency — fares, taxes and extras, minus commission, plus penalties, minus refunds — and once generated it is frozen, so a late refund does not quietly rewrite a document the agent has already paid against. An invoice past its due date shows as overdue; you mark it paid when the money lands.

None of this removes the commercial judgement. You still decide who gets a limit and how big. What it removes is the category of loss where a limit existed on paper and nobody checked it at two in the afternoon on a Thursday.

What getting it wrong costs

Three failures, in the order agencies usually meet them.

• The leaked net. Agent pricing reachable without a login, or a public rule that undercuts an agent tier. The first costs you a supplier contract; the second costs you the agent, who stops booking and does not say why.
• The overdrawn agent. A limit that was not enforced at issuance. The tickets are real, the airline wants its money, and your recovery is a phone call to a company that may no longer answer.
• The departed agent who still has a login. Covered below, because it is the one people forget.

Each of them is cheaper to prevent in configuration than to fix in collections.

When a sub-agent leaves

Agents leave. They go to a competitor, they close, a partner walks out and takes the client list. The question is what their access looks like the next morning.

Deactivating an agency's office on this platform is a switch, not a deletion. It refuses new logins for every user in that office and cuts the sessions that are already open, immediately — not whenever a token happens to expire. Their bookings stay on record, their invoices stay on record, and whatever they owe is still owed. If it was a misunderstanding, you switch it back and nothing was lost.

Do it the day you hear, not the day the paperwork arrives. An agent who has decided to leave and still has a live credit line is the most expensive user on your system. And their travellers are not yours to call. That was the deal when they signed up, and the reputation you keep by honouring it is worth more than the handful of bookings you might win by breaking it.

Where to start

Most agencies build the public site first and bolt the agents on later. It is easier the other way round: decide the tiers, set the credit rules, then open the public storefront knowing which office it prices as. If you are still weighing whether to run your own engine at all, the trade-offs are laid out in white label versus building your own; if your agents sit in several countries, read where margin leaks on a multi-currency site before you set a single rate.

And if what you are really building is a network — agencies buying your content under their own brands, not just booking through yours — that is a consolidator account rather than a storefront setting. Here is what a consolidator account covers, including selling white-label sites to the agencies you already supply.