Real-Time Bidding: How RTB Auctions Work in Programmatic Advertising

Real-Time Bidding: How RTB Auctions Work in Programmatic Advertising

What Is Real-Time Bidding

This real time bidding guide explains the auction technology that powers most programmatic advertising transactions today. Real-time bidding, commonly abbreviated as RTB, is the process by which digital ad impressions are bought and sold through instantaneous auctions that occur in the time it takes a webpage to load, typically under 100 milliseconds.

Every time you visit a website that displays ads, an RTB auction likely determines which ad you see. The publisher’s ad server sends information about the available impression to an ad exchange. The exchange solicits bids from connected demand-side platforms representing advertisers. Each DSP evaluates the impression against its campaigns’ targeting criteria and submits a bid. The highest bidder wins and their ad is displayed. All of this happens before the page finishes rendering.

RTB revolutionised digital advertising by replacing manual negotiations and bulk purchases with automated, impression-level buying. This means advertisers can evaluate each individual impression and decide whether it is worth buying based on the specific user, context and timing. The efficiency gains are enormous compared to traditional media buying approaches.

Understanding RTB mechanics is essential for anyone involved in digital marketing, particularly as programmatic spending continues to grow. RTB is the engine behind the programmatic advertising ecosystem that Singapore businesses increasingly rely on.

How RTB Auctions Work Step by Step

The process begins when a user navigates to a webpage. The publisher’s web server starts loading the page content while the ad server identifies available ad slots on the page.

For each ad slot, the publisher’s supply-side platform creates a bid request. This request contains data about the impression including the page URL, ad placement size, the publisher’s floor price and any available user data such as cookie IDs, device type and geographic location derived from the IP address.

The SSP sends this bid request to one or more ad exchanges. The exchange broadcasts the request to all connected demand-side platforms. Each DSP receives hundreds of thousands to millions of bid requests per second across all its connected exchanges.

Each DSP evaluates the bid request against active campaign criteria. Does the impression match any campaign’s targeting parameters? What is the user’s predicted value based on historical data? Is there remaining budget available? The DSP’s algorithms calculate a bid price within milliseconds.

Bids are submitted back to the exchange, which runs the auction. The winning bid is determined by the auction type. The exchange notifies the winning DSP, which responds with the ad creative. The ad is served to the user’s browser, completing the process.

This entire sequence, from page load to ad display, occurs in roughly 50 to 100 milliseconds. The speed is achieved through distributed computing infrastructure, with bid evaluation happening in data centres geographically close to the exchanges to minimise latency.

First-Price vs Second-Price Auctions

The digital advertising industry has transitioned from second-price to first-price auctions. Understanding the difference is crucial for effective bid management.

In a second-price auction, the winning bidder pays one cent more than the second-highest bid. If advertiser A bids SGD 5.00 and advertiser B bids SGD 3.00, advertiser A wins but pays only SGD 3.01. This model encouraged bidders to bid their true valuation since they would rarely pay their full bid amount.

In a first-price auction, the winning bidder pays exactly what they bid. Using the same example, advertiser A wins and pays the full SGD 5.00. This model requires more sophisticated bidding strategies because overbidding wastes money while underbidding risks losing impressions.

Most major ad exchanges including Google Ad Exchange moved to first-price auctions by 2020. This shift has made bid shading essential. Bid shading is a technique where DSPs algorithmically reduce bids below the advertiser’s maximum to avoid overpaying while still winning a competitive share of auctions.

Floor prices set by publishers add another dimension. Publishers establish minimum bid thresholds below which they will not sell their inventory. In private marketplace deals, floor prices are negotiated between specific buyers and sellers, often at premium levels that reflect the quality of the inventory.

Signals That Influence Bid Decisions

User data is the most valuable signal in RTB. Information about who the user is, what they have browsed, what they have purchased and their demographic profile all influence how much an impression is worth to an advertiser. Remarketing audiences, where the user has previously visited the advertiser’s site, typically command the highest bid values.

Contextual signals describe the content environment. The page topic, content quality, domain authority and position on the page all affect impression value. An ad on a trusted news site within a relevant article is worth more than one buried in low-quality content.

Device and format signals include the type of device, operating system, browser, ad size and format. Mobile impressions may be valued differently from desktop. Larger ad formats generally command higher bids because they offer more visibility.

Temporal signals such as time of day, day of week and proximity to key dates influence bidding. A retailer might bid more aggressively during evening hours when conversion rates are higher. Campaigns aligned with Singapore events like 11.11, Chinese New Year or year-end sales may increase bids during these periods.

Competitive dynamics affect clearing prices. Popular audience segments and high-demand inventory attract more bidders, driving prices up. Less contested segments may offer better value. Understanding your competitive landscape helps set realistic bid expectations.

Effective SEO provides first-party data that enriches these signals. Users who arrive through organic search and are later remarketed through RTB provide valuable conversion data that improves bidding algorithms.

RTB Bidding Strategies

Fixed bidding sets a static bid price for all impressions matching your targeting criteria. It is simple but inefficient because it pays the same amount regardless of impression quality. Fixed bidding is useful for initial testing when you lack performance data.

Dynamic bidding adjusts bid prices based on the predicted value of each impression. Machine learning models analyse historical performance data to estimate the probability of a desired outcome for each impression. Impressions with higher predicted value receive higher bids. This is the standard approach used by modern DSPs.

Target CPA bidding optimises towards a specific cost per acquisition. The algorithm adjusts bids to achieve the target CPA across the campaign. This requires sufficient conversion data for the model to learn effectively, typically at least 30 to 50 conversions per week.

Target ROAS bidding optimises for a specific return on ad spend. It requires revenue data attached to conversions and works best for e-commerce campaigns. The algorithm considers both conversion probability and expected order value when setting bids.

Bid multipliers allow manual adjustments on top of algorithmic bidding. You might set a 1.5x multiplier for mobile devices if they convert at higher rates, or a 0.5x multiplier for certain geographies. Use multipliers to incorporate business knowledge that the algorithm may not capture.

Integrating RTB bidding with your Google Ads strategy ensures consistent bidding logic across platforms and avoids competing against yourself in auctions where both your DSP and Google Ads campaigns are active.

Optimising RTB Campaign Performance

Win rate analysis reveals how competitive your bids are. If your win rate is below 10 percent, your bids may be too low. If it exceeds 50 percent, you might be overbidding. Aim for a win rate between 15 and 30 percent for open auction campaigns, adjusting based on your specific objectives.

Supply path optimisation identifies the most efficient route to reach publishers. When the same impression is available through multiple exchanges and SSPs, each with different fees and auction dynamics, choosing the optimal path reduces costs. Leading DSPs automate this process, but manual review of supply sources can yield additional savings.

Frequency management prevents wasted spend on users who have already seen your ad multiple times. Set frequency caps based on your campaign objective. Brand awareness campaigns might allow 5 to 10 impressions per user per week, while remarketing campaigns might cap at 3 to 5 per day.

Bid landscape analysis examines the relationship between bid levels and outcomes. Gradually testing different bid levels across audience segments reveals the point of diminishing returns where higher bids no longer improve performance proportionally.

Creative performance directly impacts RTB efficiency. Ads with higher click-through and conversion rates generate better returns at the same bid level. Test creative variations continuously and retire underperforming ads. Dynamic creative that personalises messaging based on user and contextual signals can significantly improve conversion rates.

A strong website with fast load times and clear conversion paths is essential for converting the traffic that RTB campaigns deliver. Poor landing page experiences waste the impressions you have paid for.

RTB Challenges and Solutions

Ad fraud is a persistent challenge in RTB environments. Fraudulent publishers create fake impressions using bot traffic to earn ad revenue. Invalid traffic detection tools, pre-bid fraud filters and post-bid verification help combat this issue. Read our detailed guide on ad fraud prevention for comprehensive strategies.

Viewability varies significantly across RTB inventory. Impressions served below the fold or in hidden placements may never be seen by users. Set viewability targets and use pre-bid viewability prediction to filter out low-quality placements. Our article on ad viewability standards details measurement approaches and industry benchmarks.

Data deprecation from cookie phase-outs and privacy regulations is changing how RTB works. Contextual targeting, first-party data strategies, cohort-based targeting and publisher-provided identifiers are emerging as alternatives to third-party cookies. Prepare for this shift by investing in first-party data collection now.

Latency issues can cause bid timeouts and lost opportunities. DSPs must respond to bid requests within strict time limits, typically 50 to 100 milliseconds. Infrastructure quality and geographic proximity to exchanges affect response times. Choose DSPs with strong APAC infrastructure for Singapore campaigns.

Transparency remains an ongoing concern. The programmatic supply chain involves multiple intermediaries, each taking a fee. Demand transparency from your partners, use supply chain verification tools and audit your programmatic spending regularly to ensure your budget reaches actual media rather than disappearing in hidden fees.

Frequently Asked Questions

How fast does real-time bidding actually happen?

The entire RTB process from bid request to ad display takes approximately 50 to 100 milliseconds. DSPs typically have 10 to 50 milliseconds to evaluate a bid request and respond. This speed is achieved through distributed computing infrastructure optimised for low-latency transactions.

Does RTB always result in the cheapest ad price?

Not necessarily. RTB prices are determined by competition. Popular audiences and premium inventory can be expensive in open auctions. Private marketplace deals and programmatic guaranteed transactions may offer more predictable pricing for specific inventory. The value of RTB lies in efficiency and precision rather than lowest price.

What data do bid requests contain about users?

Bid requests typically include cookie or device IDs, IP-based location data, device type, browser, operating system and sometimes demographic information. The amount of user data varies based on publisher consent practices and applicable privacy regulations.

How does RTB handle user privacy?

RTB platforms comply with privacy regulations by supporting consent management, honouring opt-out signals and implementing data protection measures. As third-party cookies phase out, RTB is evolving to use privacy-preserving technologies like contextual targeting and aggregated audience signals.

Can small businesses benefit from RTB?

Yes, though most small businesses access RTB through simplified platforms like Google Ads rather than dedicated DSPs. Google Ads display campaigns use RTB mechanics under the hood, making impression-level buying accessible without requiring programmatic expertise or large budgets.

What is the difference between RTB and programmatic?

RTB is a subset of programmatic advertising. Programmatic refers to all automated ad buying, including RTB open auctions, private marketplace deals, preferred deals and programmatic guaranteed. RTB specifically refers to the open auction model where any advertiser can bid on available impressions.

How do I know if my RTB campaigns are working?

Monitor key metrics including win rate, effective CPM, click-through rate, conversion rate and return on ad spend. Compare performance against your campaign objectives and industry benchmarks. Regular analysis of these metrics reveals whether your bidding strategy, targeting and creative are effective.

What happens when no one bids on an impression?

If no DSP bids above the publisher’s floor price, the impression may go unfilled, showing no ad or a house ad. Some publishers use passback chains to offer the impression to secondary ad networks. Others use header bidding to maximise fill rates by soliciting bids from multiple demand sources simultaneously.