How to Measure the Impact of Traffic Shaping

Publishers are caught in a squeeze. Infrastructure costs are climbing as bid request volumes balloon. DSPs are increasingly aggressive about rejecting what they consider "spammy" requests. In many cases, platforms send more than 70% of their traffic to DSPs, yet fewer than 10% of those requests are actually considered. And yield remains frustratingly volatile despite sending more requests to more partners.

Traffic shaping has emerged as the consensus solution: intelligently filtering which bid requests go to which demand partners, when, and at what price.

But here's the problem most publishers face; everyone is talking about traffic shaping, but most are still looking for ways to prove it's working.

Traffic shaping changes auction dynamics. It affects who bids, how often they respond, and how competitive each auction becomes. Those changes often show up first in intermediate signals like bid response rate, win rate, bid density, and revenue per thousand requests, not immediately in topline revenue.

Without measuring those signals holistically, it is easy to misinterpret results. Bid response can improve while bid density weakens. Win rates can rise because fewer buyers are participating, not because demand is stronger. Revenue can remain flat while request efficiency improves or degrades underneath the surface.

This is why traffic shaping requires a different measurement approach. The question at hand is whether it is improving auction efficiency in a way that translates into incremental yield. Proving that requires visibility into how shaping decisions change request quality, buyer participation, and dynamics across demand partners and supply paths.

This guide walks through how to measure traffic shaping impact with precision, using frameworks from industry experts and real-world implementation patterns.

Metrics that indicate whether traffic shaping is improving yield

When evaluating traffic shaping, the most useful metrics are the ones that reflect changes in bid quality rather than volume. Traffic shaping primarily alters who participates in the auction and how competitive that participation is. These effects surface in bidder behavior before they appear in topline revenue.

The following metrics are commonly used by advanced publisher teams and industry experts to assess whether traffic shaping is strengthening demand competition and improving request-level yield:

Bid Response Rate

Bid response rate measures the percentage of bid requests that receive a response from demand partners.

In the context of traffic shaping, this metric indicates whether the requests you send are more relevant to buyers. An increase in bid response rate suggests that shaping logic is filtering out requests that bidders consistently ignore, resulting in a higher concentration of responsive demand.

This metric should be evaluated alongside request volume. A rising bid response rate paired with reduced request volume generally indicates improved request efficiency rather than suppressed demand.

Win Rate

Win rate represents the percentage of received bids that clear the auction.

For traffic shaping, win rate reflects whether buyers are encountering auctions where they are willing to compete and win. An increase in Win rate can indicate that shaping is directing impressions to buyers with stronger intent.

However, Win rate should not be evaluated in isolation. If win rate increases while bid density declines materially, it may indicate reduced competition rather than stronger demand. Healthy traffic shaping typically maintains or improves win rate while preserving competitive pressure.

rCPM (Revenue per 1,000 Requests)

rCPM measures revenue generated per thousand bid requests sent.

This is a core metric for traffic shaping because it reflects how much economic value each request produces. Unlike eCPM, which is impression-centric, rCPM evaluates efficiency at the request level, where traffic shaping decisions are applied.

If traffic shaping is effective, rCPM should increase as low-performing requests are filtered and remaining requests attract stronger bids. Stable or rising rCPM alongside reduced request volume is a strong signal that shaping is improving yield efficiency.

Buyer Diversity and Bid Density

Bryan Szekely, Head of Ad Strategy at Sigma Software Group, emphasizes these as critical KPIs: "I would look at potentially bid density [and] really understand the amount of buyers that are in the mix of your bidding platform."

Bid density refers to the number of bids received per auction, while buyer diversity reflects the breadth of participating DSPs.

Traffic shaping should reduce non-participating demand without collapsing competition. If bid density holds steady or improves while request volume declines, shaping is likely removing redundancy rather than limiting access to demand.

Similarly, track whether you're maintaining access to a healthy mix of buyers. Aggressive traffic shaping can inadvertently lock out DSPs that would have been willing to bid on certain inventory.

Fill Rate and No-bid Rate

Fill rate measures the percentage of impressions that successfully monetize, while passback rate reflects how often SSPs decline to fill an impression.

When traffic shaping is introduced, modest changes in fill rate can occur as low-value inventory is filtered. Significant or sustained declines, however, may indicate that shaping criteria are too restrictive.

No-bid rates provide additional context. Rising passbacks from specific SSPs can signal misalignment between the inventory being sent and that partner’s demand profile.

Path-Specific Performance

Gareth Glaser, co-founder of Gamera,inc  emphasizes measuring performance by channel and path.

Not all demand paths behave the same. As Gareth notes, some SSPs perform better via TAM or OB than they do through direct Prebid integrations, even though Prebid has fewer fees. 

This granularity matters for shaping decisions. Measuring performance by path allows shaping logic to adapt to where incremental yield is actually generated, rather than applying uniform rules across heterogeneous demand channels.

How to Measure Traffic Shaping in Practice

Measuring traffic shaping is not a one-time analysis. It requires a measurement framework designed to isolate the impact of shaping decisions from broader market movement and demand volatility. 

Its effects are best evaluated through controlled comparisons and longitudinal analysis rather than aggregate revenue snapshots.

Here's the operational framework:

Establish Baseline Logging

Before enabling any Traffic shaping, capture comprehensive baseline metrics over a significant period of at least two weeks. Shorter baselines often fail to account for campaign rotation, buyer testing, and intra-week demand patterns.

Log at the impression level:

• Bid requests sent by demand partner and path
• Bid responses and timeouts
• Bid density
• Win/loss outcomes
• CPMs and clearing prices
• Match rates
• Latency and processing time
• Infrastructure costs per request

Segment your baseline by meaningful dimensions: geography, device type, placement, time of day, traffic source. Traffic shaping rarely works uniformly, what optimizes mobile inventory in APAC might negatively yield for desktop traffic in the US.

Track Shaped vs. Skipped Decisions

Your measurement framework needs to distinguish between shaped impressions (subset of bidders called, dynamic floors applied) and skipped impressions (all bidders called, standard auction).

Rather than applying traffic shaping globally, reserve a consistent portion of traffic as an unshaped control. This allows direct comparison between auctions influenced by shaping logic and those operating under standard routing conditions.

Track the full metrics stack for each category: CPM, RPM, match rate, win rate, bid density, latency. 

If shaped impressions consistently show higher CPMs and win rates than skipped impressions, your decisioning logic is directionally correct.

Monitor uplift over time

Traffic shaping impact isn't static. Demand partners can adapt, campaigns change, inventory quality shifts. Your measurement system needs to be able to track trends.

Set up automated reporting that surfaces:

• Week-over-week changes in key metrics
• Degradation alerts if win rate or bid response drops below threshold
• Partner-specific performance shifts - has a previously low-performing SSP improved?
• Segment-level anomalies - is shaping working in some geos but not others?

Measure at the Path and Partner Level

Traffic shaping decisions should be evaluated at the demand path level rather than globally.

The same demand partner can exhibit materially different behavior depending on whether traffic flows through TAM, Open Bidding, or direct Prebid integrations. Fees, latency, and buyer access vary by path and influence yield outcomes.

Measuring performance by partner and path allows shaping logic to be refined based on where incremental yield is actually generated, rather than applying uniform rules across heterogeneous demand environments.

How Mile Approaches Measurement-Native Traffic Shaping

Mile builds measurement into the decisioning engine itself, using AI to adapt in real time based on what's working.

Impression-Level Shaped vs. Skipped Tracking

Mile's AI-powered RTD module logs every impression as either "shaped" or "skipped," tracking which bidders were called, which were excluded.

For both shaped and skipped impressions, Mile tracks CPMs, RPMs, match rates, win rates, and bids per auction, enabling direct comparison. 

Publishers can drill into exactly which requests were shaped and what the outcome was using real time dashboards. 

AI-Driven Real-Time Decisioning

Mile's traffic shaping operates at the impression level using machine learning models trained on historical, site-specific auction data. For each impression, the AI evaluates which demand partners are worth calling based on:

• Historical bid rate 
• Match rate 
• Win rate  
• Historical CPMs and clearing prices

These inputs are used to determine which demand partners are likely to contribute meaningful competition for similar impressions. As auction outcomes are observed, the models update continuously, allowing shaping decisions to adapt as bidder behavior changes.

Continuous Experimentation

Mile automatically maintains a portion of unshaped traffic, ensuring that shaping is always benchmarked. This exploration loop detects when filtered bidders regain value and prevents the system from over-optimizing into blind spots.

Why this matters for yield optimization

Measurement-native traffic shaping makes it possible to understand not just whether auctions are cleaner, but whether those changes are translating into stronger competition and higher request-level yield over time.

By pairing impression-level decisioning with continuous benchmarking, publishers can use Mile to evaluate traffic shaping impact with clarity and confidence, and to optimize based on observable auction outcomes. 

For publishers looking to test this approach under real auction conditions, Mile offers a six-week free trial that runs traffic shaping and dynamic floor pricing together on live traffic across client-side and server-side demand.Book your demo here.