MatchOutcome · bidding algorithm
MatchOutcome
Our bidding software. It works out what a single impression is worth for the result you are buying, and bids accordingly.
01The gate
- The content read
- The placement record
- The campaign objective
- The contracted metric
One bid, set against your metric
Set after the read, before the buy.
- The content read
- The placement record
- The campaign objective
- The contracted metric
One bid, set against your metric
Set after the read, before the buy.
MatchOutcome works out what each impression is worth for your goal, and bids accordingly.
Every impression is priced against the goal you set.
What is this impression worth to this campaign. MatchOutcome answers it, for every impression, against the metric you bought.
What it decides
A bid for this impression, on this page, for this campaign. MatchPersona decided it is worth considering. MatchOutcome decides what that fit is worth.
Content Graph
asks whether the page is relevant
Placement Graph
asks whether the unit delivers for this outcome
MatchOutcome
sets the bid
Delivered results
retrain both graphs
What feeds it
The MatchPersona match
from MatchPersona, through the Content Graph.
The Placement Graph
what each placement has actually delivered.
The metric
from CampaignDNA.
The commitment
Metric and price agreed before the campaign. Reporting on the same metric.
The metric is named
The price is agreed
The outcome is delivered
Reported against the same metric the campaign was bought on.
What it does not do
MatchOutcome bids on what an impression is worth for your goal, on content we have read and vectorised, and reports on the same goal you set. The price follows your metric, and you can check every step.
Impression opportunities
Content read
Before any bid
Not read · no bid
Read · no brand fit
Read · fit · bid
Questions
MatchOutcome, explained
What does MatchOutcome decide?
The bid for each individual impression: a price for this impression, on this page, for this campaign, set against the metric the campaign was bought on, rather than a rate for a thousand impressions averaged across a placement.
What does it use to decide?
Three inputs. The MatchPersona match from MatchPersona, read through the Content Graph. The Placement Graph, which records what each audited ad unit has actually delivered by format and device. And the metric from CampaignDNA.
How does it work on YouTube?
On the Trusted Web the placement record informs the bid directly. On YouTube no placement record exists before the buy, so the MatchPersona match carries the decision and delivery is reported against the metric in CampaignDNA.
Does it optimise to a C Wire score?
No. The metric is yours, set in CampaignDNA. MatchOutcome optimises on delivered performance rather than predicted performance: what a placement has done is evidence, and what a model hopes it will do is not.
Does it chase cheap inventory?
No. Inventory that cannot plausibly deliver the result is not cheap; it is waste at a lower price. MatchOutcome only ever bids on content that has been read and matched.
What is agreed with the client?
The metric and the price, before the campaign runs. Reporting runs against the same metric, so the number we are held to is the number you chose. Change the metric and the bidding follows it.
Next step
Choose how to buy.
See the routes into the same matching and bidding system, or bring us a brief.