We use cookies to improve your experience and for marketing. Read our cookie policy & privacy policy.

Getting started · AI Ad Ops

Traffic Analyzer

StackUp TeamUpdated November 20266 min read

Traffic Analyzer shows how much of your traffic comes from people and how much comes from crawlers and automated agents, and what that automated traffic does in your Prebid auctions.

What you can do with it

Traffic Analyzer classifies every session by its UserAgent request source and by region. Use it to answer four questions:

  • What share of my sessions is likely human?
  • Which crawlers and agents account for the automated share?
  • Which regions send the most automated traffic?
  • How many bid requests, responses and renders does each bidder spend on automated sessions, and what is that worth?

Open Traffic Analyzer and set the report

  1. Sign in to the StackUp console.
  2. If you manage more than one account, choose the account in the switcher at the top of the page.
  3. In the left menu, under AI Ad Ops, select Traffic Analyzer.
  4. Use the domain dropdown to choose one domain or several. It opens on All domains.
  5. Use the date picker to choose a single day or several days. You can go back up to 7 days.
  6. Select Refresh to reload the report, or Export to download it as a JSON or CSV file.

The report is built from closed UTC days. The line under the page title tells you how many days are included, for example "Aggregated over 1 closed UTC day actually included". The line under the header card shows when the report was generated.

Check the headline numbers

The header card and the four cards below it summarize the selected period.

NumberWhereWhat it shows
SessionsHeader cardSession-days across the included days.
Bid RequestsHeader cardBid requests across all sessions.
Response RateHeader cardThe share of bid requests that received a response. The response count is shown beneath it.
Estimated CPMHeader cardCPM estimated from rendered value.
RendersHeader cardSuccessful Prebid renders.
Total SessionsSummary cardsDistinct daily sessions.
Likely HumanSummary cardsSessions classified as human, with their share of traffic.
Automated RateSummary cardsThe share of sessions classified as crawler, agent or API client. The session count is shown beneath it.
Automated Prebid Win ValueSummary cardsPrebid win value estimated from automated sessions.
Traffic Analyzer header card and summary cards

See how traffic is classified

The Human vs Crawlers / Agents panel splits your sessions into five classes, based on the UserAgent of each request. The ring shows the human share, and each class shows its session count and share.

ClassHow it is identifiedCounts as automated
Likely humanStandard browser UserAgent, after automation markers are excludedNo
CrawlerKnown or generic crawler marker in the UserAgentYes
Browser agentHeadless browser or automation framework markerYes
API clientProgrammatic HTTP-client marker in the UserAgentYes
UnknownEmpty or unrecognized UserAgentNo
💡
Unknown UserAgents are not included in automated traffic. The Automated Rate is the Crawler, Browser agent and API client sessions together, divided by total sessions.

Find where automated traffic comes from

Two panels break the automated share down further.

Request Source (UserAgent) lists the specific sources behind each class, for example Standard browser, Bing crawler, Google crawler, Other crawler, Headless browser and Unknown UA. Each row shows the source's class, its sessions and its automated rate.

Traffic by Region lists sessions, automated sessions and the automated rate for each collection region. The panel opens on the largest regions, with the rest grouped as Other. Select Show all regions at the bottom of the panel to list every region. Higher automated rates are highlighted in orange and red, so regions with unusual levels of automated traffic stand out.

Request Source and Traffic by Region panels

Measure the effect on your Prebid auctions

The Automated Prebid impact panel shows what happened in your auctions during sessions classified as crawlers or automated agents. The total at the top right, Automated Estimated Value, is the estimated win value across those sessions.

Each row is one bidder:

ColumnWhat it shows
BidderThe Prebid bidder.
Agent SessionsAutomated sessions in which the bidder took part.
Agent Bidder RequestsBid requests sent to the bidder during automated sessions.
Agent Bid ResponsesBid responses the bidder returned during those sessions.
Response RatioResponses divided by requests.
Agent RendersAds from the bidder that rendered during automated sessions.
Estimated Prebid Win ValueThe estimated value of the bidder's wins during automated sessions.
⚠
Sessions are distinct within a bidder row and are not additive across bidders. One automated session can involve several bidders, so do not sum Agent Sessions down the column.
Automated Prebid impact table by bidder

To see more detail for a bidder, select the arrow at the start of its row. The expanded row shows:

  • Agent render share: the part of the bidder's renders that came from automated sessions.
  • Total renders for bidder: all of the bidder's renders in the period.
  • A breakdown by request source, with agent sessions, bidder requests, responses, renders and win value for each source.
Expanded bidder row with breakdown by request source

For the definitions behind every number, open Methodology and metric semantics at the bottom of the page.

Common questions

Why is today not in the report?

The report only includes closed UTC days. A day appears once it has ended in UTC.

Why do the dates look off by a few hours?

Days are counted in UTC, not in your local time zone. The generation time under the header card is also in UTC.

Why is Unknown traffic not counted as automated?

An empty or unrecognized UserAgent is not proof of automation, so Unknown sessions are kept out of the automated count and the Automated Rate.

Why can't I add up Agent Sessions across bidders?

A single automated session can call several bidders, and it is counted once in each bidder's row. Summing the column counts the same session more than once.

Why is the Automated Prebid Win Value $0.00 when crawlers generated bid requests?

Bidders can receive requests during automated sessions without responding or winning. Compare Agent Bidder Requests with Agent Bid Responses and Agent Renders to see whether automated sessions led to any paid impressions.

Was this article helpful?