How to start a B2B travel agent portal

Every agency that decides to build a B2B portal starts with the same mental image: a search box, a list of flights, an agent clicking book. That part is real, but it is also the part that is largely solved. The decisions that determine whether the portal works are made elsewhere, and they are usually made too late.

What follows is the order those decisions should be taken in, based on what tends to go wrong when they are not.

Decide what an agent is before you decide anything else#

A B2B portal is a pricing system with a search page attached. Before you look at a single supplier, you need to know how many tiers of agent you will have, what distinguishes them, and whether an agent can have their own sub-agents.

This sounds administrative. It is not. Tier structure determines your markup rules, your credit policy, your reporting and your onboarding process, and changing it after you have onboarded three hundred agents is genuinely painful. Two or three tiers is normal. Some consolidators run five. The number matters less than deciding it deliberately.

Credit is the hard part, not search#

The single most common failure in agency-run portals is credit exposure discovered too late. An agent books through the month, the ledger is reconciled at month end, and by the time anyone notices the balance the money is gone.

A working portal blocks at the point of issue, not after. The agent sees the shortfall, requests a top-up, and the request lands in an approval queue. That is a small feature that prevents a large category of loss, and it needs to be in place before your network grows rather than after your first bad debt.

Alongside it you need a ledger your agents will accept. Receipts, credit notes, adjustments and a statement they can export. Disputes over balances consume more counter time than almost anything else, and they are almost always caused by a ledger the agent cannot see or does not trust.

Supplier content: start narrow#

The temptation is to connect everything on day one. Resist it. Every additional source adds search latency and, on some agreements, a cost per shopping request. Start with enough coverage to serve your actual routes, then let the data tell you what to add.

A common starting point is one consolidator covering both air and hotel, because it gets you selling quickly on a single agreement. Add a GDS when your own contract lands, add a bedbank when your hotel volume justifies it, and add a second air source when the reports show you are losing bookings on routes your first source covers thinly.

The reports that make this decision are win rate and margin per supplier. If a source is searched ten thousand times and wins forty bookings at thin margin, it is costing you latency for very little.

Markup rules will be more complicated than you think#

Almost nobody prices with a single percentage. Real markup logic varies by supplier, by airline, by route, by class, by agent tier, and often by date range for seasonal adjustment. It stacks, and it needs a minimum margin floor so a stacked discount cannot take a booking below cost.

Before you launch, get someone to write down how you currently price. It usually lives in one person's head, and extracting it is the most valuable two hours of the project. It is also the moment most agencies discover that their pricing has inconsistencies nobody had noticed.

What to have ready before go-live#

Your agent tiers and what each should see. Your credit policy and who approves changes to it. Your markup logic, written down. Your terms, cancellation policy and invoice format. Your payment gateway, or a plan to open one. Supplier credentials where you already hold contracts.

None of that is technical, and all of it is on your side. The build is rarely what delays a launch.