The rise of header bidding overloaded the programmatic supply chain with bid requests. So bid throttling emerged as a much-needed detox.
Header bidding required DSPs to compete for the same ad inventory auctioned by multiple SSPs in real time. But this caused an explosion in duplicated bid requests. And bid duplication was driven further by rampant reselling of ad impressions among SSPs.
Bid throttling is publishers’ strategic response to this overwhelming flood of bid requests and the need for more precise inventory management.
“The rise of header bidding introduced a lot of disruption,” said Linda Chen, Senior Director of Ad Strategy and Operations at Chegg Inc, “publishers suddenly found themselves exposing all inventory at once and inadvertently competing against themselves multiple times.”
Bid throttling is partly about blocking duplicated bid requests to avoid wasted effort from pinging noncompetitive bidders. This allows publisher pages to load more quickly because they’re unburdened by unnecessary auction activity. But it’s also about creating a more efficient programmatic ecosystem by prioritizing high-yield bidders – potentially increasing publisher revenue in the process.
As DSPs and SSPs wrestle for control over bid optimization, publishers like Chegg and the Daily Mail are developing nuanced approaches to manage their ad inventory. Here’s their advice for how publishers can build a bid throttling strategy that will work for their unique needs.
Taming Duplicate Demand
According to Chen, it’s important to distinguish bid throttling from other types of controls.
“Bid throttling,” she explained, “is about limiting the number of bid requests sent to DSPs, particularly in cases where multiple paths to the same impression could lead to duplicate bidding.”
With every SSP accessing the same impression, buyers face an influx of duplicated opportunities and a severe case of wasted queries per second (QPS).
In response, some DSPs pushed to handle filtering on their own. “Some wanted full access to all available inventory so they could apply their own algorithms,” Chen said. “Others preferred a more selective approach, asking SSPs for pre-filtered supply to reduce the volume they had to process.”
These bid-filtering efforts by DSPs led to a wave of supply-path optimization (SPO) logic from the SSP and DSP side. But these SPO solutions largely left publishers in the dark on how DSPs and SSPs were filtering bid requests. In response, publishers wanted to take a more measured approach in the form of publisher-side bid throttling.
“At Chegg, we only start throttling when it’s clear the inventory isn’t going to monetize,” Chen said. “If an ad slot gets called 20 or more times with zero bids, we stop making requests and serve content or a house ad instead.”
But Chen cautioned against publishers being overzealous with their bid throttling. “I’m not limiting after just five or six no-bids,” she said. “That’s too aggressive.”
The approach, she explained, has two key benefits: it reduces latency and ensures that the bid requests sent to buyers are more likely to generate value.
“But I try not to over-engineer it,” she added. “I don’t have visibility into every buyer’s or SSP’s algorithm. If we over-throttle, we risk missing out on demand we didn’t anticipate.”
“Things change constantly,” she said. “A buyer could suddenly value a new ID or campaign type, and if I’ve already filtered that inventory out, I lose the opportunity.”
Instead, her strategy is about minimal, data-driven control. If there’s a consistent pattern of no-bid inventory, it gets deprioritized. “It’s not about cutting off the long tail completely,” she said. “It’s about knowing when it’s time to stop trying.”
Throttling With Precision
To more effectively balance between demand losses and revenue gains, some publishers have taken a more infrastructure-driven approach to bid throttling, building it directly into their internal tech stacks. That’s because going in-house can give publishers more control over what bid requests they’re sending out than working with third parties.
For example, the Daily Mail experimented with a third-party bid-throttling solution, “but the uplift in yield and [user experience] didn’t justify the loss in bid volume,” said Jasper Liu, the publisher’s senior programmatic yield analyst. “So we transitioned to an in-house approach.”
Today, the Daily Mail runs rolling A/B tests based on page type, ad format, environment and browser, evaluating SSPs across auction-level bidding behavior and latency, Liu said. Partners are dialed up or down in real time depending on performance within each test slice. And the Daily Mail’s server-side wrapper handles both traffic throttling and demand seat rotation, enabling tighter control without excessive manual work.
This infrastructure has delivered steady performance improvements, according to Liu. “We’ve seen modest but consistent improvements, particularly in reducing page latency and minimizing any impact from bidders that are resource hungry,” he said. That, in turn, has improved core metrics like eCPM, bid rate and win rate for participating SSPs, he added.
Inventory Throttling: Controlling What Gets Sold
But deciding how to send bid requests and to which partners is just part of the problem publishers have to tackle. They also have to decide whether it’s even worth it to send programmatic bid requests for some of their ad inventory.
That’s where the concept of inventory throttling comes into play.
“Inventory throttling is deciding what inventory you even want to allow to be bid on,” explained Chegg’s Chen. “It’s not about limiting the number of bid requests; it’s about choosing which pieces of your inventory go to which partners in the first place. Even with bid throttling, publishers don’t get to control how inventory is surfaced and what gets bid on.”
SSPs have increasingly taken control of inventory throttling through curated deals and SPO packages. But Chen sees a problem with SSPs controlling these decisions: “Their algorithms aren’t built for individual publishers; they’re built for a marketplace,” she said. “So what’s optimal for a massive network might not work for me.”
In recent years, new tools have offered publishers more control, allowing them to decide what inventory gets surfaced to the buy side. But Chen remains cautious about going all in on these solutions.
Guardrails for Premium Demand
To protect high-value opportunities, the Daily Mail also applies its bid throttling selectively.
“From our end, throttling applies only to open auction traffic,” Liu said. “Deals—PMP, PG and first-look deals—are whitelisted and exempt from throttling, ensuring no impact on high-value or guaranteed campaigns.”
The Daily Mail has also developed custom dashboards to monitor and rank demand partners by integration type, tracking performance across monthly, quarterly and annual intervals. This data helps inform throttling decisions and ensures the system stays calibrated as demand patterns shift, Liu said.
Start Simple, Stay Flexible
While bid throttling may seem like a precarious balancing act, with various trade-offs between revenue yield and demand opportunities, there are a few simple things publishers should consider before diving in.
Linda Chen’s advice to publishers thinking about bid throttling is straightforward: You don’t have to do it, and if you do, it doesn’t need to be overly complex.
“It’s not a must-have,” Chen said. “And it doesn’t have to be sophisticated to be effective.”
For those just getting started with bid throttling, she recommends high-level, low-risk tactics. If a bidder consistently underperforms in a particular region, stop sending them inventory from that region. Similarly, if a partner struggles with video, exclude them from video auctions.
“You can keep it simple without building out a complex model,” Chen added.
If publishers do want to explore more advanced throttling, Chen noted that numerous vendors offer this kind of support. But she leans toward a more conservative approach for Chegg.
“Things are always changing, so I avoid aggressive models,” she said. “What looks fine today might have very different results over time.”
Because bid throttling decisions need to be adjusted over time, Chen also emphasized the importance of building in a long-term learning phase when deciding on a strategy. That way, publishers can track performance and impact beyond the short-term metrics.
In the end, Chen’s philosophy is straightforward: Test carefully, keep it flexible, and don’t overengineer a solution when a more simple approach will do just fine.