How to choose a lead routing strategy

Routing strategy is the setting that decides what every lead in your operation is worth. Two operators can run identical campaigns, identical buyers, identical traffic, and see materially different revenue per lead — because one is routing on price discovery and the other is routing on position in a queue.

It's also the setting most operators pick once, during setup, based on whatever the platform defaulted to. Then they optimize everything else — creative, sources, filters — around a routing decision nobody revisited.

This is how to make that decision deliberately.

First, clear up the framing error

Almost every conversation about routing starts in the wrong place: "should we do ping/post or waterfall?"

That question doesn't parse. Ping/post is not a routing strategy. It's a transport.

Ping/post describes how you talk to a buyer: a thin, anonymized ping to get a bid or an interest signal, then a post carrying the full record to whoever wins. Waterfall, auction, weighted, and tiered describe how you decide who wins. These are independent axes. You can run a waterfall over ping/post buyers. You can run an auction over direct-post buyers with fixed prices. Any strategy runs over either transport.

Conflating the two is how operators end up believing they "can't do ping/post because we're on waterfall," or that switching to an auction requires rebuilding every buyer integration. Neither is true, and the confusion costs real money in deferred decisions.

Separate the axes:

• Transport — direct post (one call, full record) or ping/post (bid first, deliver to the winner).
• Strategy — the rule that picks the winner.

Everything below is about the second one.

The four strategies

Waterfall

Buyers are ordered. The lead is offered to the first buyer; if they accept, distribution ends. If they reject, it falls to the second, and so on. All buyers reject, the lead is rejected.

What it optimizes for: priority. You are explicitly saying "this buyer gets first look at everything, always."

When it's right: you have a preferred buyer relationship — a volume commitment, a strategic partner, a parent company — where the order itself is the business logic. Also correct when you have one or two buyers and an auction would be theater.

What it costs you: price discovery. Your first buyer accepts a lead they'd have paid double for, and you never find out, because nobody else was asked.

Highest bidder (auction)

Every eligible buyer is pinged simultaneously. Bids come back. You sort them, post to the highest, and if that buyer rejects the post, you fall to the next-highest.

What it optimizes for: revenue per lead. This is the only strategy that discovers what a specific lead is actually worth to a specific buyer at a specific moment.

When it's right: you have enough buyers to make a market — realistically three or more actively bidding — and your margin depends on price. If lead value varies meaningfully by attribute (geo, credit band, vertical intent), an auction captures variance that a flat price averages away.

What it costs you: it requires buyers who can actually bid. A buyer posting a static price to your ping isn't bidding, they're just quoting. An auction full of static quotes is a waterfall with extra latency.

Weighted

Each buyer carries an integer weight. Weights define target shares of volume. Each lead goes to whichever buyer is currently furthest below its target share.

What it optimizes for: distribution against volume commitments. If you've promised buyer A 50% and buyers B and C 25% each, weighted is the strategy that keeps that promise without you managing it by hand.

When it's right: you have contractual volume splits, or you're deliberately spreading volume to keep several buyers warm.

What it costs you: the same thing waterfall costs you — price discovery — plus a subtlety that catches people out, below.

Tiered

Tiers are ordered groups of buyers. Each tier runs its own sub-strategy over its own buyer subset. The lead tries tier 1; if every buyer there rejects, it falls through to tier 2, and so on.

What it optimizes for: mixing strategies in one campaign, which is usually what you actually want.

When it's right: more often than operators expect. The canonical shape: a premium tier running an auction among buyers who'll pay real money, with a fallback tier below running weighted distribution to spread whatever the premium tier didn't want. You capture price at the top and monetize the tail at the bottom.

What it costs you: configuration surface. Tiers introduce a floor per tier, an order per tier, and a buyer subset per tier. More knobs, more ways to misconfigure.

The decision framework

If your situation is…	Use
One or two buyers, fixed prices	Waterfall — an auction has nothing to auction
A preferred buyer with first-look rights	Waterfall
Three or more buyers who genuinely bid, and margin is the business	Highest bidder
Contractual volume splits to honor	Weighted
Premium buyers and a tail you want monetized	Tiered (auction on top, weighted below)
Buyers whose value you can't know until conversion	Tiered, with the conversion-priced buyers in a lower tier

The honest summary: waterfall and weighted optimize for relationships; auction optimizes for revenue; tiered lets you stop choosing. If you have the buyer count to support it, tiered is usually where you end up, because real operations have both a premium segment and a tail.

Weighted distribution deserves a warning

Two things about weighted routing are non-obvious and both have burned operators.

Weight targets share of leads won, not share of leads attempted. A buyer that gets offered every other lead and rejects most of them stays below its target share, so the engine keeps trying it first. Its deficit never closes. If you set weights expecting them to describe traffic allocation, and your buyers have materially different accept rates, realized volume won't match the weights — and the engine is behaving correctly.

The counter is usually all-time. This is the one that surprises people. Add a new buyer at weight 25 to a campaign that's been running for months, and that buyer is maximally below its target share from the moment it's created. It will receive a catch-up burst — a disproportionate slug of volume — until its realized share climbs to target. Same thing happens to a buyer coming back from a cap or a pause.

That's expected behavior, not a bug, but if you don't know it's coming you will interpret it as a routing failure and start changing weights mid-burst, which makes it worse. Add buyers to weighted campaigns knowing the first day is not representative.

Where conversion-priced buyers break the model

If you work with buyers who pay on conversion rather than on delivery, routing gets a wrinkle worth understanding before you configure it rather than after.

A conversion-priced buyer has no realized price at distribution time. Delivery books nothing; revenue lands later, when the buyer confirms — or never, if they don't. Which means:

Don't put conversion-priced buyers in a head-to-head auction with delivery-priced buyers. They have no bid to compete with. Either they're excluded by any price floor you've set, or they're included at zero and structurally lose every auction. Neither is what you meant.

Fallthrough triggers on rejection, not on non-conversion. This is the important one. When a conversion-priced buyer accepts the post, distribution ends — the lead is theirs. If the conversion later fails, lower tiers are never reached. The lead doesn't come back and try again. You've committed the lead to a buyer who ultimately paid nothing.

Both of these point the same way: put conversion-priced buyers in a lower tier with no floor, below your delivery-priced buyers. The top tier wins guaranteed revenue first. The bottom tier monetizes what's left on a maybe. That ordering reflects the actual risk, and it's the reason tiered exists.

Price floors are three different things

"Price floor" gets used loosely and it hides a real distinction:

• A campaign-level bid floor rejects bids below a threshold. Note that a bid floor only means anything when buyers are bidding — on a direct-post campaign with fixed prices there are no bids to floor, and operators regularly set one and wonder why it does nothing.
• A per-tier floor applies to one tier only and overrides the campaign floor for that tier's buyers.
• A per-buyer rule on the buyer's own response handling.

The per-tier floor has a sharp edge: a floor above zero silently excludes zero-price buyers. If you've built the recommended shape — premium tier on top, conversion-priced or revenue-share buyers in a fallback tier below — and that fallback tier has a floor greater than zero, your zero-price buyers qualify for nothing and the tier looks broken. Fallback tiers holding zero-price buyers need their floor explicitly set to zero.

Common mistakes

Mistake 1: Running an auction with buyers who don't bid

You switch to highest bidder, revenue doesn't move, you conclude auctions don't work in your vertical. The real problem is that all four of your buyers return the same static number to every ping. An auction only pays when there's dispersion in what buyers will pay.

The fix: before switching, check whether your buyers' bids actually vary by lead. If they don't, the auction is a waterfall wearing a costume, and your gain is in getting buyers to bid dynamically, not in the routing change.

Mistake 2: Picking a strategy for the campaign instead of for the segment

One strategy across every lead in a campaign assumes every lead has the same economics. They don't. Your high-intent, high-credit, dense-metro leads deserve an auction. Your thin-file tail deserves to be spread across whoever will take it.

The fix: tier it. This is precisely the problem tiers solve.

Mistake 3: Treating rejection as a routing failure

Buyers reject leads. That's the system working — a rejection is a buyer telling you something. When operators respond to rejection rates by reordering buyers or nudging weights, they're treating a quality signal as a routing problem.

The fix: read the rejection reasons before touching the routing. Most "routing isn't working" is a filter, cap, schedule, or source-quality problem wearing a routing costume.

Mistake 4: Never revisiting the choice

The strategy that was right when you had two buyers is not right when you have twelve. Buyer count is the variable that most changes the answer, and it's the variable that changes silently as you grow.

The fix: revisit routing every time your active buyer count materially changes. Crossing from two buyers to four is the threshold where auction economics usually start to pay.

Mistake 5: Changing strategy and volume at the same time

You switch to weighted on Monday and scale a new source on Monday. Revenue per lead moves. You have no idea which change did it.

The fix: change one variable. Routing changes need a clean baseline or you can't read the result.

The takeaway

Routing strategy is not a setup detail. It's the mechanism that converts your buyer relationships into revenue per lead, and the four options optimize for genuinely different things: waterfall for priority, auction for price, weighted for commitments, tiered for all of the above.

Start by separating transport from strategy — ping/post is not on the menu, it's the plumbing. Then pick based on your buyer count and where your margin actually comes from. If you have a premium segment and a tail, and most operations do, you want tiers.

And when you change it, change only it. The whole point is to find out what it was worth.