Operations
Part of Display advertising: signal versus noise
Two display advertising case studies compared for publisher and audience gaps
Display advertising case studies need review for publisher, selection, dates, audience, creative, inventory, measurement, fees, costs, outcomes, and transfer limits.
What to take away
- Both cases are vendor published. Garnier India dynamic creative is Google's account of L'Oreal work; Qantas Programmatic Guaranteed is Google's account of the airline's display buys.
- Publisher gapneither case names a publisher, a placement or an inventory source a buyer could call.
- Audience gapneither case shows an independent count of who saw the ads, and neither runs a control group.
- No budget, creative cost, fee split or campaign date is public for either case.
- Copy the mechanism, then test it locally. Treat every percentage as a lead, not a benchmark.
What each case publishes, and what it withholds
Google's Garnier India story describes a dynamic creative campaign for L'Oreal: audience groups, automated product feeds and many creative variations. Google's Qantas story describes tailored display ads for the Australian carrier, bought through Programmatic Guaranteed at prices fixed in advance.
Neither story gives a campaign start or end date. Neither gives media spend, creative production cost, a fee split or a profit figure. A number quoted on another site is unverified until you see the primary document. Display advertising coverage on this site explains why vendor write-ups normally leave those fields empty.
Before citing the Garnier case, check six items.
- Campaign start and end dates
- Definitions of audience group and conversion
- The comparison used as a baseline
- Media, creative and fee costs
- Whether the campaign was typical or picked after it worked
- Who counted the result, and on which system
Verify Before Concluding
- Check campaign dates
- Check audience definitions
- Check comparison method
- Check costs
- Check selection criteria
- Check underlying evidence
The publisher gap, side by side
Publisher gap means the distance between the party that served the impression and the party that wrote the result. In the Garnier case both are Google. Inventory comes from Google's own buying platform, so no outside publisher confirms placement, viewability or the fee split.
In the Qantas case, Programmatic Guaranteed fixes price and volume with publishers before the campaign runs. That structure beats an open auction on terms. The public story still withholds publisher names, ad positions, viewability floors and fee terms.
Garnier India dynamic creative
- Buying route
- Google's own buying platform
- Publisher named
- No
- Placement or position
- Not published
- Viewability count
- Not published
- Fee split
- Not published
- Independent ad server named
- No
Qantas Programmatic Guaranteed
- Buying route
- Fixed-price Programmatic Guaranteed deals
- Publisher named
- No
- Placement or position
- Not published
- Viewability count
- Not published
- Fee split
- Not published
- Independent ad server named
- No
The audience gap, side by side
Garnier's audience groups come from the advertiser's own signals. Qantas segments come from the airline's traveler and customer records. In both cases the seller of the inventory also defines the audience.
Neither case reports a count of who saw the ads from a source outside the platform. Neither uses a control group. Without one, a click or conversion comparison cannot separate creative from seasonality, price moves or audience mix. Keep platform attribution apart from caused effect when you read those numbers.
Nine mechanisms, listed plainly
The nine lanes are a reading structure, not a ranking. Each lane asks a different question of a case study.
- Responsive combinationshow were incompatible claims and crops filtered?
- Dynamic creativewhich feed fields and failure rules governed scale?
- Contextual placementhow was page meaning checked rather than assumed?
- Direct dealswhat transparency and remedy did the contract add?
- Frequencywhich identity gaps remained?
- Retargeting suppressionhow were completed and ineligible states excluded?
- Native recognitioncould ordinary users tell it was advertising?
- Lift measurementhow were assignment, power and spillover handled?
- Supply pathwhich intermediaries and fees were removed or kept?
One current documented case per lane is enough, and undated entries should be dropped. Display advertising examples shows how to tell a documented campaign from a promotion.
Example: a bounded local test with a stop rule
This is a design template, not a reported result.
- Take one claim from a case, such as dynamic creative lifting click-through.
- Budget it as a typical rangethree to four weeks at a low four-figure US dollar spend, confirmed against your own quotes.
- Run it on Maximize Conversions with a target CPA cap, so the system does not spend the month in week one (Google Ads support).
- Hold out part of the audience on the old creative, and keep both arms in one ad server.
- Stop ruleat day 14, pause if cost per qualified lead sits more than 30% above the pre-test figure and the seven-day trend is flat.
What a buyer can trust from these two cases
Trust the mechanism and the buying route. Do not trust the percentages, the transfer to your market or any hint of incremental profit.
In the United States, objective performance claims need a reasonable basis before they are published, per the FTC advertising substantiation policy. A friendly customer story does not lower that bar.
The FTC guidance on reviews and incentives covers undisclosed relationships and conditioned incentives, which is what a paid placement in a case-study list amounts to.
Before you pick a platform on the strength of either case, compare display advertising tools by inventory, fees and exit terms. Neither story names an ad server, so that comparison has to run on your own requirements.
