The freight agent scorecard, explained.

Why every published template fails

The scorecard literature that exists was written for a different relationship: a shipper buying trucking capacity from a carrier. Its columns — tender acceptance rate, on-time pickup, on-time delivery, claims per hundred loads — describe a one-directional purchase of a single service. The forwarder-to-agent relationship breaks that model three ways.

• It’s bilateral. Your agent in Colombo is also your customer — the nominations they route to you matter as much as the freehand you send them. No carrier scorecard has a reciprocity column, because no trucker nominates freight back.
• It’s multi-role. The same partner quotes your RFQs, executes your shipments, files your documents and settles your invoices. A carrier template scores one function; an agent must be scored on four.
• It’s protected — or it isn’t. Network payment-protection programs (Gold Medallion-class coverage, per-claim caps, cross-network rules) change what any given score is worth. A brilliant rate from an unprotected stranger is a different proposition than the same rate from a covered member. Carrier scorecards have no equivalent field.

So the honest starting point is this: there is no published forwarder-to-agent scorecard standard. More than 250 forwarder networks exist, and not one ranks its members on your data. What follows is a template, offered as one — not a benchmark dressed up as research.

The four dimensions that fit

Four dimensions cover the relationship: how they respond, how they price, how they execute, how they settle. The weightings below are an example starting point — a forwarder living on spot enquiries should push weight toward response speed; one running contracted lanes should push it toward execution and settlement.

Two fields ride alongside the four scores rather than inside them: protection status (which program covers this partner, per network membership, to what cap) and reciprocity(freehand sent versus nominations received). Neither is a performance grade — they’re context that changes what the grades mean when you decide where volume goes.

Seeding one from your own history

The good news: you already own the data. Every score above can be seeded from records sitting in your inbox, your WhatsApp threads and your job files right now. Ninety days is enough to make the first version honest.

• Pull every RFQ you sent in the last 90 days — partner, lane, timestamp sent, timestamp answered. Silence is a data point: log the non-responses, not just the replies.
• Normalize every quote received — all-in totals in one currency, validity noted — so “competitive” means something across a PDF, an email body and a WhatsApp message.
• Walk the shipment files for the same window: milestones hit or missed, documents right first time, PODs returned. Execution is the dimension memory flatters most.
• Pull the settlement record — SOA discipline, disputes opened and closed, netting behaviour — from accounts, not from the operations desk’s impression of it.
• Note protection status per partner, per membership — program, cap, cross-network rules — once, so it’s never reconstructed during a crisis.

Then keep it alive. A scorecard built once is a report; a scorecard that updates with every RFQ round is an instrument. This is, plainly, the part VendorFinder AI automates — every dispatched RFQ, parsed reply and settlement event updates the partner’s card without anyone maintaining a spreadsheet — but the method above works in Excel if you have the discipline. Most desks don’t, which is why the anecdotes win.

The three biases that ruin scorecards

• Recency bias. Last week’s late quote outweighs two years of clean files, because last week is what the desk remembers. Fix: score across every RFQ in the window, and make the window explicit — a 90-day view and an all-time view, side by side.
• Dinner bias. Conference warmth leaks into operational scores; the partner everyone likes gets graded on the relationship instead of the record. Fix: keep the scores mechanical and let the relationship inform the decision, not the data. The dinner belongs in the room where you decide — not in the OPS column.
• Cheapest-quote bias. Scoring competitiveness as “cheapest wins” rewards the partner who quotes low and executes badly — and the spread they save you is smaller than the milestones they miss. Fix: this is exactly why execution and settlement carry weight equal to price in the example table.