Quick answer
Real-time bidding (RTB) is a method of buying ad impressions in which each impression is auctioned the moment it becomes available, in under a second. The winning bid is a price for an impression, not for a visit or a conversion.
Formula: Spend = impressions ÷ 1,000 × CPM; cost per conversion = spend ÷ conversions
What is real-time bidding?
Real-time bidding (RTB) is a way of buying and selling ad impressions in which each impression is auctioned the moment it becomes available. When a page or app screen loads, details of the available ad slot are sent to potential buyers, they bid, and the highest eligible bid wins. Google's documentation describes RTB as the process in which digital advertising inventory is bought and sold, one impression at a time, and says the whole process occurs in less than a second.
RTB is one method of programmatic advertising, not a synonym for it. What makes it "real time" is that the price is decided per impression, at the moment of the request, instead of in a contract signed weeks earlier.
The point to carry through the rest of this page: the winning bid is a price for an impression, and an impression is not a visit. What happens after the auction, on your own site, is a separate measurement.
How does an RTB auction work?
The steps, following IAB Tech Lab's OpenRTB specification and Google's RTB documentation:
- Request. A person loads a page. The seller's system sends a bid request describing the impression to an exchange. The OpenRTB specification defines an exchange as a service that conducts an auction among bidders per impression.
- Broadcast. The request goes out to buyers' systems (see demand-side platform), often filtered first by criteria the buyer chose, such as language or geography.
- Evaluation. Each buyer checks the request against its campaigns and decides whether to bid and how much. A submitted bid includes a CPM price and either the ad itself or a pointer to it.
- Auction. The exchange applies its auction rules and selects a winner.
- Display and counting. The winning ad is returned and shown, and the impression is counted.
What the buyer receives in a request, according to Google's documentation, includes the referrer URL, targeting data, the IP address and an encrypted cookie ID. That is why RTB attracts data-protection attention: personal data travels to many parties for impressions that those parties do not win.
First-price versus second-price auctions
Exchanges run different rules for what the winner pays:
| Rule | Winner pays | Effect on bidding |
|---|---|---|
| First price | Their own bid | Buyers shade bids down to avoid overpaying |
| Second price | The next-highest bid (often plus a small increment) | Bidding the true value is safer |
Floors set by the seller can sit above both. A bid below the floor never wins, whatever the rule.
Sample data: four bidders offer $2.10, $3.40, $3.10 and $1.80 CPM, with a $2.00 floor. All but the $1.80 bid clear the floor. In a first-price auction the $3.40 bidder wins and pays $3.40. In a second-price auction with a $0.01 increment, the same bidder wins and pays $3.10 + $0.01 = $3.11. The same impression costs 29 cents more per thousand under the first rule. Always check which rule an exchange uses before comparing CPMs across sources.
How do you evaluate what RTB bought?
Because the price is set per impression, it is tempting to optimize for the lowest CPM. The cost that matters is cost per outcome on the landing page.
Sample data: two campaigns spend about the same budget through RTB.
| Campaign A | Campaign B | |
|---|---|---|
| CPM paid | $2.50 | $4.00 |
| Impressions bought | 1,000,000 | 500,000 |
| Spend | 1,000,000 ÷ 1,000 × $2.50 = $2,500 | 500,000 ÷ 1,000 × $4.00 = $2,000 |
| Landing sessions | 1,500 | 1,000 |
| Conversions | 15 | 30 |
| Conversion rate | 15 ÷ 1,500 = 1.0% | 30 ÷ 1,000 = 3.0% |
| Cost per conversion | $2,500 ÷ 15 = $167 | $2,000 ÷ 30 = $67 |
Campaign A has a 37.5% lower CPM ($2.50 versus $4.00) and a 2.5 times higher cost per conversion. The auction was won on price; the outcome was lost on audience quality. CPM is a price of delivery; it says nothing about whether the people it delivered to were interested.
A second check is the click-to-session ratio. If the platform reports 3,000 clicks and your analytics records 2,400 landing sessions, the 20% gap is normal to some degree (people leave before the page loads, bots, accidental taps). A gap that suddenly doubles points to a landing-page or tracking problem. See impressions for the definitional layer underneath.
Decision rules: what to do when the numbers move
Use these if-then rules on the weekly comparison of platform data and landing-page data:
- CPM falls and conversion rate falls faster (cost per conversion rises): the cheaper impressions are lower quality. Raise the bid floor on placements or narrow the audience.
- CPM rises and conversion rate rises faster (cost per conversion falls): the higher price is buying better audiences. Keep it.
- Clicks rise but landing sessions do not: suspect accidental or invalid clicks, a slow landing page or a tagging break before touching the bids.
- Sessions rise but conversions do not: the audience or the offer on the page is wrong; the auction is working as designed.
- Two sources disagree by more than a stable ratio: stop comparing until the tags and windows are checked.
Common mistakes with real-time bidding
- Comparing CPMs without comparing auction rules. A $3.11 CPM under one rule and $3.40 under another may be the same underlying value.
- Treating a won impression as a view. A won bid only means the ad was delivered to the slot. Whether it was seen, or by a person, is a separate verification question.
- Optimizing on clicks. RTB can find cheap accidental taps. Look at conversion on the landing page.
- Forgetting the data flow. Bid requests carry visitor data to many parties. Your own consent settings decide what your site sends; your legal team should review the supply-side configuration.
- Reading a rising win rate as improvement. Winning more auctions can simply mean bidding higher.
How to read RTB campaigns in TapCub
Take the two campaigns above. TapCub is not involved in the auction; it does not bid, buy or run ads. It measures what the visits from each campaign do.
- Tag each campaign's landing URL with
utm_source,utm_mediumandutm_campaign, putting the campaign name (A, B) inutm_campaign. See name UTM parameters. - Compare in Sources. Web analytics › Sources › UTM shows visitors, sessions, pageviews, bounce and duration per campaign, so you can see whether Campaign A's visitors leave faster than B's.
- Filter and count conversions. Click the campaign row to filter, then open Web analytics › Goals for the people-based conversion rate. TapCub separates recognised crawlers into Web analytics › Bots, so they do not sit in the denominator.
- Import spend and compare cost. In All-platform analytics › Business › Channel ROI, import daily spend by campaign; the page returns cost per click (spend ÷ sessions) and cost per acquisition.
- Reconcile clicks with sessions. Compare the platform's click count with the sessions TapCub recorded for the same campaign; a sudden change in the ratio is the first thing to investigate.
TapCub does not see impressions, bids or auction prices; those stay in the buying platform.
See also: What is header bidding? How it works vs the waterfall.
Sources
- Google Authorized Buyers Help, Introduction to real-time bidding (RTB).
- IAB Tech Lab, OpenRTB.
- IAB Tech Lab, OpenRTB 2.6 specification.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Product Adoption Consultant
Explores customer onboarding, live conversations, and self-service, turning common usage questions into practical guidance and improvements.