Si an-Se illa, Ido; Pa ham, Pa ick; McGuigan, Lee
A icle
"Cookie-less" iden i ica ion o /agains p i acy?
In e ne Policy Re iew
P o ided in Coope a ion wi h:
Alexande on Humbold Ins i u e o In e ne and Socie y (HIIG), Be lin
Sugges ed Ci a ion: Si an-Se illa, Ido; Pa ham, Pa ick; McGuigan, Lee (2025) : "Cookie-less"
iden i ica ion o /agains p i acy?, In e ne Policy Re iew, ISSN 2197-6775, Alexande on Humbold
Ins i u e o In e ne and Socie y, Be lin, Vol. 14, Iss. 3, pp. 1-27,
h ps://doi.o g/10.14763/2025.3.2025
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/324163
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/3.0/de/deed.en
Volume 14 |
‘Cookie-less’ iden i ica ion o /agains
p i acy?
Ido Si an-Se illa Uni e si y o Ma yland
Pa ick Pa ham Uni e si y o Ma yland
Lee McGuigan Uni e si y o No h Ca olina a Chapel Hill
DOI: h ps://doi.o g/10.14763/2025.3.2025
Published: 6 Augus 2025
Recei ed: 19 Sep embe 2024 Accep ed: 4 Feb ua y 2025
Funding: The au ho s did no ecei e any unding o his esea ch.
Compe ing In e es s: The au ho has decla ed ha no compe ing in e es s exis ha
ha e in luenced he ex .
Licence: This is an open-access a icle dis ibu ed unde he e ms o he C ea i e
Commons A ibu ion 3.0 License (Ge many) which pe mi s un es ic ed use,
dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly
ci ed. h ps://c ea i ecommons.o g/licenses/by/3.0/de/deed.en
Copy igh emains wi h he au ho (s).
Ci a ion: Si an-Se illa, I., Pa ham, P., & McGuigan, L. (2025). ‘Cookie-less’
iden i ica ion o /agains p i acy? In e ne Policy Re iew, 14(3). h ps://doi.o g/
10.14763/2025.3.2025
Keywo ds: Cookie-less ad e ising, Consume p i acy, Online ad e isemen ,
Comme cial su eillance
Abs ac : The ad e ising indus y’s an icipa ed shi away om hi d-pa y cookies led o he
p oli e a ion and no malisa ion o i s -pa y iden i ica ion a chi ec u es online. Ma ke ed as
‘p i acy- iendly,’ he new echnologies p omise o deli e he e iciencies ha ad e ise s ha e
become accus omed o, while add essing p i acy conce ns om hi d-pa y cookies. Such ension
calls o a be e unde s anding o he p i acy implica ions om i s -pa y online iden i ica ion
a chi ec u es. We e alua e i s -pa y use iden i ica ion mechanisms by (1) su eying he li e a u e
o c ea e a ypology ha syn hesises exis ing p i acy conce ns in hi d-pa y cookie-based
iden i ica ion, and (2) applying ou ypology o e alua e he p i acy o p ime examples in wha we
ame as h ee dis inc ypes o i s -pa y ID a chi ec u es — Uni e sal IDs, Onboa ding ID, and
Walled Ga den ID. We analyse echnical documen a ion and code eposi o ies om each
a chi ec u e ype and show how i s -pa y ID solu ions s ill enable c oss-si e acking o e longe
pe iods o ime and encou age sensi i e use a ge ing. Fi s -pa y ID solu ions do c ea e
mechanisms o ease op ing ou om acking, bu he implemen a ion o hose mechanisms is
ques ionable. Ou indings demons a e how he ad e ising indus y is ying o main ain i s
exis ing s uc u e and eplica e he acking unc ionali ies on which i has g own elian .
Issue 3
1 - In oduc ion
Con inuous iden i y-building o online consume s has been he c own jewel o he
digi al ad e ising indus y (Alaimo, 2022; Gandy, 2021; Melle & Beau isage,
2020). By using indi idual da a poin s ha go beyond iden i ica ion de ails, ad e -
ising ac o s ha e been in ensi ely c a ing use iden i ies o p o iling pu poses,
wi hou use awa eness (Chan , 2021; EU Commission, 2023), iola ing consume s’
p i acy igh o eely and con inuously econs uc he sel in he ace o e e -
changing con ex s (Ag e & Ro enbe g, 1997). Taking con ol o e he cons uc ion
o consume sel es (Benois , 2008; Elme , 2003; Zwick & Dholakia, 2004), compa-
nies ha acili a ed pe sonalised ad e ising made hemsel es in eg al o he
mone isa ion o he in e ne (C ain, 2021; Kan , 2021; Tu ow, 2011). Iden i y se -
ices became in as uc u al o online ad e ising ma ke s ha , in he US alone,
gene a ed USD 225 billion in e enue in 2023 (IAB, 2024).
The pe asi e iden i y-building o consume s has been enabled by online iden i i-
ca ion a chi ec u es. These a chi ec u es allow con inuous a ibu ion o collec ed
da a o speci ic use s and eed he ep esen a ion o indi iduals as co ela ed da a
subjec s, based on pas beha iou , o en wi hou he knowledge o he da a sub-
jec s hemsel es (Hildeb and , 2006). Online iden i ica ion eme ged sho ly a e
he bi h o he In e ne . Unique use iden i ie s in he o m o hi d-pa y (3p)
cookies became he mos popula ancho s o collec ed use da a (Tu ow, 2011).
O iginally designed o enable con inui ies in b owsing expe iences (namely, le ing
p oduc s placed in online shopping ca s emain he e while a use na iga ed o
o he pages, o con inue shopping, pe haps), hi d-pa y cookies became he go- o
ins umen o iden i ying online sessions and acking consume s’ beha iou o
iden i y-building p ac ices (Jones, 2020). Ad e ising ac o s ha e been using hi d-
pa y cookie iden i ie s by d opping hem on b owse s o pe sis en consume
iden i ica ion ac oss he web (Si an-Se illa & Poudel 2025), he collec ion o
b owsing his o y, and o conduc cookie syncing p ac ices o linking consume
iden i ies ac oss de ices and be ween di e en ad e ising ac o s (Solomos e al.,
2020). They a e called hi d-pa y cookies because he companies deploying hem
do no ha e a di ec (o i s -pa y) ela ionship wi h he use — he p esump ion be-
ing ha i s -pa ies, like he websi es o me chan s a use consciously engages
wi h, a e legi ima ely au ho ised o collec use da a, whe eas hi d-pa ies ha e
less au ho isa ion.
Recen ly unde p essu e o e o m online p i acy p oblems, Apple and Mozilla
ha e es ic ed he use o 3p cookies on hei b owse s, while Google has epea -
edly made, and hen wi hd awn, ha same pledge (Mo gan & Mazalon, 2024). The
2 In e ne Policy Re iew 14(3) | 2025
looming inabili y o use hi d-pa y cookies o use iden i ica ion c ea ed a acu-
um o ad e ising ac o s. Wi hou a ibu ing collec ed da a o consume s, ad e -
ising e enue was pe cei ed o be a isk (B odhe son e al., 2021). Ou s udy o-
cuses on one highly-publicised me hod o illing his acuum: i s -pa y iden i i-
ca ion a chi ec u es, which a e ma ke ed by he In e ac i e Ad e ising Bu eau
(IAB), he leading ad e ising indus y ade associa ion, as ‘p i acy i s solu ions’
(Eng, 2024; IAB Canada, 2021). In iguingly, wha we e p e iously ma ginal con-
sume iden i ica ion mechanisms ha e become inc easingly popula and no -
malised as ‘pos -cookies’ & ‘p i acy- i s ’ use iden i ica ion based on i s -pa y
use da a (Mel ze , 2020; Tabisz, 2023). The ad e ising indus y calls hese “uni-
e sal” o “cookie-less” iden i ie s; we sugges ha a mo e i ing name is “ i s -pa -
y iden i ica ion a chi ec u es,” and ask, wha a e he p i acy implica ions o i s -
pa y iden i ica ion a chi ec u es?
‘Fi s -pa y iden i ica ion a chi ec u es’ p o ide a means o he digi al ad e ising
indus y o legi imise and main ain ce ain abili ies o iden i y consume s. Fi s -
pa y iden i ica ion a chi ec u es a e dis inguished om p io hi d-pa y sys ems
based on he sou ce o da a and iden i ica ion mechanisms; in he o me , compa-
nies wi hin digi al ad e ising ma ke s (e.g., ad e ise s, ad ech in e media ies, and
publishe s) supply hei own p op ie a y da a, including emails, names, and phone
numbe s, o ind consume s ac oss ad-suppo ed media enues, a he han elying
on he passi e su eillance execu ed ia hi d-pa y cookies. This name is some-
wha imp ecise, in ha hese a chi ec u es s ill ely on he sha ing o i s -pa y
da a be ween di e en digi al ad e ising ecosys em pa icipan s; he indus y’s
sel -se ing use o ‘ i s -pa y’ glosses o e shi ing classi ica ions o wha is i s -
pa y e sus hi d-pa y da a in hese con ex s (McGuigan e al., 2023). Ne e he-
less, we sugges ha his e m e ec i ely cap u es he na u e o he p ojec —o
using he legi imising conno a ion o i s -pa y ela ionships o au ho ise con in-
ued acking and iden i ica ion. We a gue ha h ee ypes o i s -pa y iden i ica-
ion a chi ec u es ha e eme ged o help acili a e he iden i ica ion o consume s
om di e en posi ions in he ad e ising supply chain.
We ecognise h ee dis inc i s -pa y iden i ica ion a chi ec u es ha ha e
eme ged in he online ad e ising supply chain: (1) Publishe s1a e deploying ‘Uni-
e sal ID’ solu ions o pe sis en ly iden i y logged-in consume s; (2) Ad e ising
echnology pa ne s p o ide ‘Onboa ding Iden i ica ion’ mechanisms o link i s -
pa y ad e ise da a wi h use iden i y ac oss ad e ising pla o ms; and (3)
1. In digi al ad e ising e minology, a publishe is any en i y ha exhibi s con en and sells ad e is-
ing oppo uni ies on i s websi e(s), whe he a la ge news o ganisa ion o an independen blogge .
3 Si an-Se illa, Pa ham, McGuigan
‘Walled-Ga den Iden i ica ion’ enables dominan ac o s ha a e ac i e on bo h he
buye and selle sides o online ad e ising, such as Google and Me a, o link hei
i s -pa y use da a ac oss he di e en p oduc s o he same ‘walled-ga den’ pla -
o m.
Ou pape sys ema ically compa es he capabili ies and ecommended uses o i s -
pa y iden i ica ion a chi ec u es o he known su eillance and p i acy conce ns
su ounding hi d-pa y cookies, he eby allowing a igo ous assessmen o
whe he he o me meaning ully sol e—o essen ially ep oduce— he p oblems
associa ed wi h he la e . We i s conduc an in e disciplina y e iew o li e a u e
on he p i acy conce ns o 3p cookie-based iden i ica ion and acking. We ol-
lowed a snowball app oach o collec li e a u e un il eaching sa u a ion in e ms
o spo ing new hemes o p i acy issues. We o ganised he hemes ha eme ged
om ou e iew in o a no el i e-pa ypology o 3p cookie-based p i acy con-
ce ns. We hen used ha ypology o e alua e exempla s o each i s -pa y iden i-
ica ion a chi ec u e: The T ade Desk’s Uni e sal ID 2.0 solu ion, Li eRamp’s
RampID onboa ding iden i ie , and Google’s Cus ome Ma ch walled-ga den iden i-
ica ion solu ion. Ou selec ion o p ime examples o analyse was in o med by ou
eading o ade publica ions co e ing he digi al ad e ising indus y, he equen-
cy o epo ing on speci ic i s -pa y solu ions, and he expec ed adop ion a es o
hose solu ions by di e en publishe s and AdTech ac o s (Asim, 2021, 2022).
All h ee ypes o i s -pa y iden i ica ion a chi ec u es a e simila in ha hey ely
on he enc yp ion o i s -pa y da a o iden i y use s. S ill, we obse ed dis inc
iden i ica ion objec i es ac oss hese solu ions: Uni e sal IDs ope a e ac oss si es
wi hin ne wo ks o pa icipa ing publishe s; Onboa ding iden i ica ion helps buy-
e s o ads a ge use s in almos any si e o pla o m; and Walled Ga den Iden i i-
ca ion helps buye s a ge use s in speci ic pla o m en i onmen s. We collec ed
da a on each i s -pa y iden i ica ion case om echnical documen a ion ha he
de elope s o he di e en solu ions ha e made public. Fo each solu ion, we
analysed he code eposi o ies, ela ed indus y publica ions, media epo ing, and
ele an documen s, in ligh o ou de eloped e alua ion ypology o exis ing p i-
acy conce ns wi h 3p cookie iden i ie s.
Con a y o indus y claims ha i s -pa y ID solu ions ad ance consume s’ p i a-
cy, we ind ha i s -pa y iden i ica ion ac ually in ensi ies impo an aspec s o
online su eillance: he iden i ie s and he da a p o iles associa ed wi h hem can
be mo e pe sis en ly a ached o indi iduals; he pe sonal da a p o iles can be
mo e comp ehensi e; and he iden i ie s can allow o sensi i e o ms o p o iling
and disc imina ion o pe sis and e en ski e o s o disable hem. The e is nu-
4 In e ne Policy Re iew 14(3) | 2025
ance o hese indings. We do ecognise p og ess ega ding he se o ac o s ha
can ack use s and some p omising anspa ency and use agency mechanisms
ha could po en ially inc ease he accoun abili y o online acke s. S ill, he ap-
plica ion o ou ypology sugges s some conce ning conclusions abou he p i acy
implica ions o i s -pa y iden i ica ion a chi ec u es.
Fi s -pa y iden i ica ion is a om p i acy-inducing. These e o ms ail o add ess
he s uc u al condi ions, incen i es, and powe ela ions ha emain in con adic-
ion wi h many legal p o isions (Veale & Bo gesius, 2022) and any meaning ul de-
ini ion o p i acy (Ag e & Ro enbe g, 1997; Ba ocas & Nissenbaum, 2014;
Wach e , 2019). Consume acking, p o iling, and a ge ing a e deeply en enched
as no ms and an unde lying ma ke in as uc u e in hese a chi ec u es.
The nex sec ion de ails ou esea ch me hodology and i s limi a ions. We hen
summa ise exis ing li e a u e on he a ious p i acy implica ions o hi d-pa y
cookie iden i ie s, es ablishing a ypology o la e assess i s -pa y iden i ica ion
a chi ec u es. Sec ion 4 applies ou c i e ia o compa e he p i acy implica ions o
i s -pa y iden i ica ion a chi ec u es, analysing a p ime example om each i s -
pa y iden i ica ion ype. Sec ion 5 discusses he implica ions o ou indings and
concludes, de ailing u u e esea ch ques ions.
2 - Me hods
The s a ing poin o ou da a collec ion was he ad e ising indus y's claims ha
‘ i s -pa y’ iden i ica ion a chi ec u es imp o es use s’ p i acy, as popula b owse s
shi ed away om hi d pa y cookies, and he indus y sough new use iden i ica-
ion mechanisms (Eng, 2024; IAB Canada, 2021; Mel ze , 2020; Tabisz, 2023). We
we e ai ly skep ical, and decided o i s syn hesise exis ing p i acy conce ns in
he online ad e ising supply chain ha a e d i en om he eliance o ad e ising
on hi d-pa y cookies.
Da a collec ion s a ed wi h a su ey o exis ing li e a u e abou acking ia hi d-
pa y cookies and i s p i acy consequences. We began wi h he key wo ds o ‘ hi d-
pa y cookies.’ We used ACM’s digi al lib a y and he Google Schola sea ch engines
and collec ed s udies om compu e science con e ences, law and policy s udies,
and ela ed ields o ma ke ing and media esea ch. The sea ches yielded 287 a i-
cles, which we e hen en iched by inding addi ional s udies ha ci e o a e ci ed
by wo ks in he ini ial sample, expanding in a snowball p o ocol un il eaching sa -
u a ion. We ead and il e ed hese wo ks and ended up wi h 67 ele an pape s
ha we e manually classi ied based on he p i acy hemes ha eme ged in he
5 Si an-Se illa, Pa ham, McGuigan
con ex o hi d-pa y cookies. Pape s we e selec ed o analysis i hey discussed
any p i acy implica ion om he use o hi d-pa y cookies o acking and a ge -
ing pu poses. This i s da a collec ion s ep led o he es ablishmen o ou ypolo-
gy, which names i e dis inc p i acy conce ns enabled by hi d-pa y cookie iden-
i ie s, a icula es hei ypes and de ini ions, and lis s a sample o ele an s udies
on each p i acy issue (see nex sec ion).
The second s ep o da a collec ion e ol ed a ound he collec ion o in o ma ion on
he sugges ed i s -pa y iden i ica ion a chi ec u es. Based on he expe ise and
indus y expe ience o one o ou co-au ho s, we b oke down sugges ed i s -pa y
iden i ica ion a chi ec u es o h ee ypes: Uni e sal ID solu ions, On-boa ding
iden i ie s, and Walled Ga den iden i ie s. We hen selec ed h ee echnologies o
ep esen he h ee di e en ypes o iden i ica ion a chi ec u es based on: (1) The
T ade Desk UID2.0’s posi ioning as he leading indus y wide solu ion ha has
been es ed by signi ican numbe o ad e ise s and seen con inued adop ion
(Asim, 2022; Ba be , 2024; He che , 2024); (2) Li eRamp RampID’s de elopmen
om he con inued leading onboa ding i m in he indus y (He che , 2019;
Shields, 2023); and (3) Google Cus ome Ma ch’s posi ion as he solu ion om
la ges Ad ech i m and pla o m by ma ke sha e and e enue (Yuen, 2024).
We hen di ed in o hese h ee iden i ica ion a chi ec u es and collec ed da a om
hei gi hub eposi o ies, indus y publica ions, and media epo ing. Ou main
sou ces include o icial documen a ion and indus y epo s om magazines such
as Digiday, AdExchange , and eMa ke e . We also looked in o he code eposi o ies
and websi es o he companies ha own he solu ions and used he ma e ials p o-
ided o assess how each solu ion wo ks in he b oade ad supply chain.
We analysed he collec ed da a on each i s -pa y solu ion o unde s and i s
b oade p i acy implica ions in he con ex o exis ing p i acy conce ns wi h hi d-
pa y cookies. The lis o hemes ha eme ged in s ep #1 se ed as a use ul guide-
line o unde s anding he newly p oposed solu ions, since hey add ess long- e m
p i acy conce ns on wha is enabled by hi d-pa y cookies, and se e as a poin o
e e ence o unde s and ou main puzzle - whe he cookie-less iden i ica ion a chi-
ec u es imp o e o unde mine use s’ p i acy.
Impo an ly, and as explained in he in oduc ion, we ollow he aming o he
digi al ad e ising indus y when conside ing wha o include as ‘ i s -pa y’ iden i-
ica ion mechanisms. Ins ead o elying on passi e su eillance conduc ed by hi d-
pa y cookies, he i s -pa y solu ions unde s udy supply hei own p op ie a y
da a o c ea e use iden i ie s and sync hose be ween ac o s in he ad supply
6 In e ne Policy Re iew 14(3) | 2025
chain. Fi s -pa y, in ha sense, is no necessa ily i s -pa y o he use , bu i s -
pa y o he ac o ha iden i ies he use in he ad supply chain (publishe , ad ne -
wo k, o walled ga den se ice). The indus y uses legi imising conno a ion o
‘ i s -pa y’ ela ions o jus i y con inued acking and iden i ica ion, and we aim o
assess whe he ‘ i s -pa y’ ela ions be ween he sou ces o he da a and he ac o
ha conduc s he iden i ica ion a e indeed ha mless om a use p i acy pe spec-
i e.
Ou li e a u e e iew is no an exhaus i e lis o all hi d-pa y cookie-based ack-
ing s udies. We ollowed he snowball app oach, and elied on ou expe ience and
expe ise in s udying his ield, o come up wi h wha we conside he mos impo -
an p i acy implica ions o hi d-pa y cookie ad e ising. Addi ional limi a ions
a e ela ed o ou abili y o decode and na iga e he de ails o each i s -pa y ID
speci ica ion. Despi e ou e o s o examine and analyse hese i s -pa y iden i i-
ca ion a chi ec u es, he a ailable documen a ion does no clea ly disclose all he
ele an de ails o hei unc ionali ies and implemen a ion. We ied o e i y ou
unde s anding wi h di ec ques ions o he companies bu we e unable o e ie e
mo e e ealing de ails. Ou indings may no ha e a long shel li e. Any changes o
implemen a ion p ac ices could a ec he p i acy implica ions o he di e en so-
lu ions.
3 - Li e a u e e iew: he p i acy implica ions om
hi d-pa y cookie based ad e ising
Table 1 below cap u es an a ay o di e en p i acy conce ns ha eme ged om
analysing he li e a u e on hi d-pa y cookies. The i s wo conce ns a e echnical
- he usage o hi d-pa y cookies o iden i y and ack use s ac oss Web, mobile,
and sma echnologies has enabled (1) acking use s ac oss di e en si es, and
(2) acking use s o e ime, as hese cookies ge synced and pe sis o long pe i-
ods o ime. The nex wo conce ns a e abou use igh s. Thi d-pa y cookie based
acking (3) makes op ing-ou e y di icul o use s, wi h many acke s o en ig-
no ing use s’ p e e ences, while p o iding e y li le, i any, (4) acke anspa en-
cy on who exac ly acks he use o e an a ay o si es and apps. A i h p i acy
conce n ha eme ged om he li e a u e is he abili y o acke s o use sensi i e
ca ego ies o a ge ing use s. This is a consequence o p e ious a o dances en-
abled by hi d-pa y cookies. The sample column in he able ep esen s he mos
comp ehensi e empi ical wo ks ha s udied he p i acy conce n. The pa ag aphs
below summa ise and demons a e each conce n om he e iewed li e a u e.
7 Si an-Se illa, Pa ham, McGuigan
TABLE 1: Use P i acy conce ns enabled by hi d-pa y cookies
PRIVACY
CONCERN
TYPE OF
CONCERN DEFINITION
SAMPLE OF
RELEVANT
LITERATURE
1 - CROSS-SITE
TRACKING TECHNICAL
THE ABILITY TO TRACK USERS
ACROSS MULTIPLE SITES BY VARIOUS
ACTORS BASED ON THEIR 3P COOKIE
IDENTIFIER.
ENGLEHARDT &
NARAYANAN,
2016; KARAJ ET
AL., 2019;
FOUAD ET AL.,
2020.
2 -
LONGITUDINAL
TRACKING
TECHNICAL
THE ABILITY TO TRACK USERS OVER
TIME BASED ON THEIR 3P COOKIE
IDENTIFIER.
SAMARASINGHE
& MANNAN,
2019;
PAPADOPOULOS
ET AL., 2019;
FOUAD ET AL.,
2022
3 - OPT-OUT
LIMITATIONS USER RIGHTS
THE EXTENT TO WHICH USERS CAN
OPT-OUT FROM 3P COOKIES
TRACKING OR OTHERWISE EXERCISE
THEIR PREFERENCES.
MATTE ET AL.
2020; HABIB ET
AL. 2022;
SANCHEZ-ROLA
ET AL., 2019;
GRASSL ET AL.,
2021; TREVISAN
ET AL., 2019
4 - LACK OF
TRACKER
TRANSPARENCY
USER RIGHTS
USERS' INABILITY TO KNOW WHAT
DATA IS COLLECTED ABOUT THEM
AND BY WHOM.
DEGELING ET AL.
2019; LIBERT &
BINNS, 2019;
FOUAD ET AL.,
2022
5 - SENSITIVITY
OF TARGETING CONSEQUENTIAL
LINKING 3P COOKIE IDENTIFIERS
ACROSS ADVERTISING ACTORS TO
TARGET INDIVIDUALS BASED ON
CATEGORIES SUCH AS RACE, GENDER,
AND RELIGION, THAT HAVE SPECIAL
STATUS IN LAW AND/OR ARE
PROHIBITED BY CORPORATE
POLICIES.
ALI ET AL., 2019;
WEI ET AL.,
2020;
BEAUVISAGE ET
AL., 2023
3.1 C oss-si e acking
One o he mos commonly discussed limi a ions on use s’ abili y o keep hei
iden i y agmen ed online is he use o hi d-pa y cookie iden i ie s o iden i y
and ack use s ac oss si es. Schola s ha e been measu ing he olume o c oss-si e
acking conduc ed by a ious ac o s, e ealing acking le els based on ype o
8 In e ne Policy Re iew 14(3) | 2025
a e hen me ged wi hin he AbiliTech pla o m o ep esen use s (Li eRamp,
2022c). Li eRamp consolida es o line PII abou a known indi idual including:
email add esses, pos al add esses, phone numbe s, and enables esolu ion whe e
indi iduals ha e mul iple iden i ie s in hese ca ego ies (Li eRamp, 2022c). The
Li eRamp algo i hm hen ansla es his pai ing o a RampID ia pseudoanony-
mous PII om he pa ne s (Li eRamp, 2022c). O line PII and Online de ice iden i-
ie s can now be ma ched o ep esen he use as a RampID, and enable o he o -
line and online da a sou ces abou he use o be pai ed (Li eRamp, 2022c). The
ma ching ID, ‘RampID,’ is enc yp ed in he eal- ime bidding p ocess, when buye s
a emp o buy ad space based on he RampID, and can only be dec yp ed by sup-
ply-side pla o ms (SSPs) and demand-side pla o ms (DSPs) who a e pa o he
Li eRamp ne wo k (Li eRamp, 2022d).
Since RampID is based on i s -pa y da a collec ed o pu chased by ad e ise s,
use s ha e e y limi ed isibili y o poin s o in e en ion in he p ocess. S ill,
use s can choose o op -ou om he RampID solu ion when hey log in o websi es
and sha e hei email add ess in exchange o con en . Thei email add ess can ei-
he be connec ed o a RampID h ough Li eRamp’s API, o disconnec ed om he
RampID solu ion in case use s choose o op -ou (Asim, 2022; Li eRamp, 2022c).
4.1.3 Google’s Cus ome Ma ch / Ads Da a Manage - ‘Walled Ga den Solu ion’
On Oc obe 11, 2023, Google p esen ed i s ‘Google Ads Da a Manage ,’ a pla o m
ha enables he use o i s -pa y iden i ie s held by ad e ise s and ma ke ing
agencies o he a ge ing o consume s ac oss Google’s walled-ga den p oduc s.
The pla o m aims o simpli y he managemen o i s -pa y da a connec ions be-
ween buye s o online ads and Google’s p oduc s. Acco ding o media epo e s, i
shows he desi e o Google o le e age i s massi e logged-in use base as a e-
placemen o hi d-pa y cookies (Joseph, 2024).
This i s -pa y iden i ica ion a chi ec u e is simila o he p e iously de ailed
RampID solu ion. I is gene a ed om he will o an ad e ise / ma ke ing agency
o a ge po en ial cus ome s based on a se o i s -pa y iden i ie s held by ad
buye s, such as email add esses, phone numbe s, names, and home add esses (Ja-
na dhan, 2023). Those i s -pa y iden i ie s a e hen ma ched wi h Google’s i s -
pa y iden i ie s o logged-in cus ome s ac oss Google p oduc s - Sea ch Engine,
Gmail, YouTube, and Google’s Display Ne wo k ( he a ay o websi es whe e Google
exis s as a hi d-pa y, no necessa ily ia cookies). The newly in oduced Google
Ads Da a Manage inco po a es Google’s exis ing i s -pa y a ge ing solu ion,
Cus ome Ma ch, h ough which he ID ma ching akes place (Google, n.d.-c, 2024).
15 Si an-Se illa, Pa ham, McGuigan
In con as o RampID which can enable ad e ise s o a ge use s ac oss di e en
ad e ising pla o ms, his Google solu ion is na i e o Google p oduc s, and en-
ables ad e ise s o ind use s wi hin Google’s (sp awling) walled ga den. Ad e is-
e s can upload hei i s -pa y da a by connec ing sophis ica ed i s -pa y da a
managemen pla o ms, such as cus ome ela ionship managemen (CRM) pla -
o ms and cus ome da a pla o ms (CDPs), o h ough simply uploading a *.cs ile
o cus ome da a o he Google Ads pla o m ha con ains he heade s “Email,”
“Phone,” “Fi s Name,” “Las Name,” “Coun y,” and “Zip" (Google, n.d.-b, n.d.-d, n.d.- ;
Jana dhan, 2023).
Ad e ise s also ha e he op ion o p o ec he PII da a o hei cus ome s om ex-
e nal s akeholde s ou side he Google ne wo k using he SHA256 algo i hm o
one-way hashing (Google, n.d.-e). Google ma ches he uploaded PII wi h co e-
sponding da a associa ed wi h Google Accoun s o c ea e he inal Google Ma ch
segmen ha can be used o a ge ing by Google (Google, n.d.-a).
4.2 P i acy implica ions o i s -pa y ID a chi ec u es
The sec ion below examines he p i acy implica ions o i s -pa y iden i ica ion
a chi ec u es, in ligh o exis ing p i acy conce ns om hi d-pa y cookies.
4.2.1 C oss-si e acking
The abili y o iden i y use s ac oss an inc easing numbe o websi es and ad e is-
ing pla o ms, and by an a ay o ad buye s, has been one o he main p i acy con-
ce ns enabled by hi d-pa y cookie iden i ie s. The di e en ypes o i s -pa y
iden i ica ion a chi ec u es eplica e his echnical p i acy conce n, bu limi he
se o ac o s who can do so. S ill, he Uni e sal ID ope a o can pe sis en ly iden i-
y use s ac oss he si es o any publishe s pa icipa ing in he Uni e sal ID solu-
ion; use s a e also pe sis en ly iden i ied by buye s o online ads, ac oss di e en
ad e ising pla o ms and walled ga den p oduc s. F om he selle 's side, c oss-si e
acking is only enabled on publishe s who pa icipa e in he Uni e sal ID solu ion,
po en ially limi ing he amoun o ac o s who ge c oss-si e isibili y on use s.
UID2.0 enables c oss-si e acking only o he UID ope a o who s ill ge s o iew
use s in he bids eam ac oss hei b owsing expe iences. UID okens a e cons an -
ly upda ed o ensu e ha unapp o ed en i ies wi h access o he bid s eam can-
no build p o iles based on UID okens o e ime. Li e amp’s RampID a chi ec u e
also enables c oss-si e acking by ma ching and sha ing Ramp IDs ac oss ad e -
ising pla o ms. Google’s Cus ome Ma ch enables c oss-si e acking as well,
since ad e ise s a e able o a ge Google use s ac oss hi d-pa y publishe s ha
16 In e ne Policy Re iew 14(3) | 2025
a e pa o he Google Display Ne wo k. S ill, he limi s placed on he numbe o
pa ne s able o iden i y indi iduals h ough he in oduc ion o enc yp ion mecha-
nisms could be conside ed an imp o emen o e he amoun o c oss-si e acking
based on hi d-pa y cookie iden i ie s.
4.2.2 Longi udinal acking
The abili y o lea n abou use beha iou o e ime is cu en ly enabled by a ious
hi d-pa y cookie-based acking echniques such as he espawning and syncing
o cookie IDs. In i s -pa y iden i ica ion a chi ec u es, his capabili y is enabled by
he in oduc ion o much mo e pe sis en iden i ie s, which a e based on PIIs such
as use emails o phone numbe s. This enables acke s o lea n abou use beha -
iou o e longe pe iods o ime, based on an iden i ie ha is less likely o change
equen ly in compa ison o hi d-pa y cookie IDs, and is o en used in o he o -
line da a ansac ions, linking oge he a ich his o y o use eco ds.4
UID2.0 a chi ec u e enables longi udinal acking o ad e ising and a ge ing
pla o ms by elying on use s’ PII-based login da a. Fo Li eRamp, he seed PII ha
builds he iden i ie is no expec ed o equen ly change o e ime, po en ially en-
abling pa icipa ing ac o s o ack he use o e longe pe iods o ime in compa -
ison o cu en hi d-pa y cookies-based iden i ie s. Simila ly, Google’s Cus ome
Ma ch associa es use ’s PII wi h he use 's Google login, which is unlikely o
change equen ly.
In all h ee iden i ica ion solu ions unde s udy, he use iden i ie is mo e pe sis-
en han a pseudo-anonymous hi d-pa y cookie iden i ie in he p e ious ecosys-
em. This migh lead o po en ially mo e pe sis en use acking o e ime.
4.2.3 Op ing-Ou
All i s -pa y iden i ica ion a chi ec u es in oduce echnical imp o emen s o he
abili y o use s o exe cise hei igh s, wi h dedica ed op -ou mechanisms om
pe sonalised ad e ising ha send op -ou signals o all ac o s in he bidding
p ocess abou use ’s p e e ences (Asim, 2022; Google, n.d.-a; Li eRamp, 2022b;
Uni iedID2, 2022). As opposed o cu en usage o hi d-pa y cookies, whe e ack-
e s can easily escape o no p o ide unc ional op -ou op ions (Papadogiannakis
e al., 2021), he i s -pa y iden i ica ion a chi ec u es enable an op -ou op ion
om he s a , by in oducing pop-up windows ha ask o use consen upon log-
4. Solu ions such as Apple’s hide my email can help pe iodically, bu he e is a limi o he ob usca ion
a e age use s a e capable o . Le alone when conside ing addi ional PIIs ha a e used o ace use
iden i y o e ime.
17 Si an-Se illa, Pa ham, McGuigan
ging in o a websi e.
S ill, he likelihood ha hese op -ou mechanisms will allow use s o ully ealise
hei p e e ences emains ques ionable. Use s may eel p essu ed agains op ing-
ou as he decision could mean losing access o publishe s’ con en . In ac , he
iden i ica ion indus y speci ically s a es ha he onus is on publishe s o con ince
use s o op -in o a ge ed ad e ising in exchange o online con en (Ti one,
2021). Publishe s a e he ones who need o decide on he alue exchange o use s
who do no wish o be a ge ed. They migh p e en hose use s om iewing
hei con en , making he designed op -ou mechanisms non- easible ac oss he
di e en solu ions. Fo walled ga den p oduc s and o he ad e ising pla o ms,
use s can e use iden i ica ion and a ge ing in he pla o ms’ se ings, bu will
ha e o be consis en . Resea che s will ha e o e i y ha consen pop ups a e no
icking use s in o accep ing acking e ms hey do no ully unde s and. Addi ion-
ally, while Google’s use choice and con ol documen a ion speci ies ha use s can
op ou om pe sonalised ads ha a e se ed based on eco ds o web b owsing
and app usage, i does no speci y how use s can manage hei p e e ences when
ad e ise s di ec ly upload hei PII in o Google’s Ads Da a Manage and Cus ome
Ma ch o a ge ing.
4.2.4 T acke anspa ency
T anspa ency ega ding how hi d-pa y cookies a e used o acking and wha da-
a is collec ed abou use s has been insu icien (Fouad e al., 2022). The complex
dynamics o acking we e no ully de ailed and disclosu es on p i acy policies
ailed o desc ibe obse ed acking in p ac ice (Papadogiannakis e al., 2021). In
con as , all i s -pa y iden i ica ion a chi ec u es p omise o p o ide g ea e
acke anspa ency, bu he ex en o which use s will be able o iew eco ds
collec ed on hei ac i i y emains qui e ambiguous. UID2.0 and Li eRamp’s Ramp
ID p o ide li le insigh in o hei p omises o acking anspa ency. They all p o-
ide ‘ anspa ency mechanisms,’ bu i is unclea how hose mechanisms will p o-
ide up- o-da e in o ma ion o use s on how hey a e being iden i ied and by
whom in a dynamic ad e ising deli e y en i onmen . Technically, in he complex
se ings o use iden i ica ion and a ge ing o ads, i is almos impossible o p o-
ide use s wi h accu a e in o ma ion on who is acking hem a any gi en mo-
men . The i s -pa y iden i ica ion a chi ec u es unde s udy a e unclea on how
exac ly hey will b idge his gap.
4.2.5 Sensi i e a ge ing
All i s -pa y iden i ica ion a chi ec u es p esen signi ican conce ns o sensi i e
18 In e ne Policy Re iew 14(3) | 2025
a ge ing, since hey ha e he po en ial o make i easy and a ac i e o ad e is-
e s o a ge use s based on sensi i e a ibu es (e.g., ace, eligion, e c.) ha a e
subjec o legal o o he limi a ions. Ad e ise s can pe sis en ly iden i y use s
based on hei i s -pa y cus ome da a, and hen ask a ge ing pla o ms o as-
semble ‘look-alike audiences’5based on ma ched IDs. Wha ha means is ha an
ad e ise could use an a ibu e like ace o eligion o cons uc an audience,
based on hei i s pa y-da a, and hen ha e an ad pla o m ind hose a ge s, e -
ec i ely o e iding he es ic ions agains his so o disc imina ion ha some ad
pla o ms ha e made h ough hei co po a e policies. Fo ins ance, suppose an ad-
e ise wan s o a ge A ican-Ame icans on a social media pla o m ha does
no allow a ge ing by ace. Tha ad e ise could ake a i s -pa y da ase o
which in o ma ion abou ace is included (and o which a pe sis en iden i ie like
an email add ess in known), hen so i in o an audience de ined by ace, and hen
push ha cus om audience in o he pla o m o ac i a ion wi hou ha ing o
speci ically selec ace- ela ed a ge ing pa ame e s om ha pla o m. Fu he ,
he ad e ise could eques ha he pla o m iden i y new a ge s, whose email o
o he IDs a e no pa o he ad e ise i s -pa y da a se , ia lookalike modeling
(by o he p oxies such as pos al code and income), which would be p ep ocessed
acco ding o a ace a iable. Such ‘look-alike’ audiences can now be a ge ed, en-
abling he a ge ing o A ican-Ame icans wi hou speci ically s a ing so o he a -
ge ing pla o ms.
Ad e ise s’ capaci ies —and e en incen i es— o de ine and de elop hese p op i-
e a y audiences, be o e i s -pa y use iden i ie s a e encoded o a ge ing, ep e-
sen a conce ning ea u e o he sys ems unde s udy he e. Ad e ise s cons an ly
pu chase use da a om da a b oke s o expand p o iling o use s beyond wha
passi e su eillance o b owse beha iou can p o ide. The emphasis on i s -pa y
da a ac oss he iden i ica ion a chi ec u es is encou aging ad e ise s o u he
le e age hei exis ing cus ome and sales da a, amass addi ional da a a ound
hose iden i iable cus ome s, and o ind o he s wi h simila ai s (Li eRamp,
2020). Compa ed o ad e ise s’ e o s in he cu en hi d-pa y cookie-based
ecosys em, ad e ise s a e now in es ing mo e in echnology ha can ma ch i s -
pa y cus ome da a o o he da ase s (Va gas, 2022). This wo k is ca ied ou ia
exis ing ‘iden i y g aph’ and new ‘clean oom’ echnologies ha allow ad e ise s
o manage indi idual-le el da a and encode he da a h ough an ID me hod o
5. ‘Look-alike’ audiences a e po en ial ad e ising cus ome s who esemble exis ing cus ome s ac oss
many da a a ibu es, and a e he e o e highly likely o sha e o he da a a ibu es wi h exis ing au-
diences. Ad e ise s / ma ke ing agencies use look-alike audiences when hey ask ad e ising pla -
o ms o a ge i s -pa y iden i ie s o consume s ha esemble exis ing consume s wi hin he up-
loaded PII (Schneide , 2023).
19 Si an-Se illa, Pa ham, McGuigan
choice; his p o ides a cen alised sys em o me ge online and o line iden i ie s
in o a consolida ed p o ile o pai wi h pu chased da a om da a b oke s and ac i-
a e selec ed audiences wi h mul iple pa ne s.
Wi hin The T ade Desk Uni ied ID 2.0, o ins ance, he company men ions ‘Fi s -
Pa y Rela ionships’ capabili ies, whe e an ad e ise is able o upload i s -pa y
da a o be encoded o he UID2.0 o ac i a ion ac oss publishe si es (Uni iedID2,
2022). Simila ly, Li eRamp o e s ad e ise s he oppo uni y o ‘onboa d’ hei da-
a whe e PII can be uploaded in o de o be con e ed in o RampIDs and o ganised
by segmen so ha hey can be ac i a ed in mo e han 500 di e en pa ne pla -
o ms (Li eRamp, 2022a). These capabili ies esemble he p ima y se ice p o ided
by Google’s Cus ome Ma ch p e iously desc ibed.
Impo an ly, he e is no mechanism o e i y how ad e ise s ha e segmen ed i s -
pa y da a be o e impo ing in o hese sys ems, and he i s -pa y iden i ica ion
a chi ec u es make sensi i e popula ion segmen a ion and bidding e y a ac i e
o ad e ise s. As de ailed in Sec ion 3, p e ious esea ch has iden i ied ins ances
whe e ad e ise -uploaded lis s ha e iola ed he policies se by a ious ad selle s
o in e media ies o p e en he a ge ing o use s including ca ego ies such as
ace, eligion, poli ics, sex li e, o heal h (Wei e al., 2020b). We a gue ha i s -pa -
y iden i ica ion a chi ec u es place a signi ican amoun o impo ance on le e ag-
ing i s -pa y da a ha encou ages ad e ise s o look o ways o link i s -pa y
and pu chased da a h ough ID solu ions, allowing o a ge ing p ac ices wi h li -
le o no o e sigh by p ima y pla o ms ( o example: Me a, 2021; Twi e , 2022).
20 In e ne Policy Re iew 14(3) | 2025
TABLE 2 below summa ises he p i acy implica ions ac oss i s -pa y iden i ica ion a chi ec u es
and compa es hem o exis ing p i acy conce ns in hi d-pa y cookie-based iden i ica ion se ings.
PRIVACY
CONCERN UID 2.0 RAMPID GOOGLE CUSTOMER
MATCH
1 - CROSS-SITE
TRACKING
ENABLED, BUT ONLY
FOR THE UID
OPERATOR. SSPS AND
DSPS ONLY GET TO
WORK WITH
ENCRYPTED USER IDS
AND EACH PUBLISHER
GETS A DIFFERENT
TOKEN..
SHARING & MATCHING
OF IDS IS ENABLED
ACROSS ADVERTISING
PLATFORMS AND SITES
FOR PARTICIPATING
ACTORS.
ADVERTISERS ARE
ABLE TO TARGET
GOOGLE USERS
ACROSS THIRD-PARTY
PUBLISHERS THAT ARE
PART OF THE GOOGLE
DISPLAY NETWORK &
ACROSS GOOGLE
PRODUCTS.
2 -
LONGITUDINAL
TRACKING
PII-BASED USER
IDENTIFIER IS
CREATED, BASED ON
USER LOGIN TO
PUBLISHERS’
WEBSITES. THIS IS
MORE DETERMINISTIC
THAN THIRD-PARTY
COOKIE IDENTIFIERS
AND CAN TRACK USERS
OVER POTENTIALLY
LONGER PERIODS OF
TIME.
THE SEED FIRST-PARTY
PII THAT BUILDS THE
IDENTIFIER IS NOT
EXPECTED TO
FREQUENTLY CHANGE,
POTENTIALLY ENABLING
PARTICIPATING ACTORS
TO TRACK THE USER
OVER LONGER PERIODS
OF TIME IN COMPARISON
TO 3P COOKIES-BASED
IDENTIFIERS.
GOOGLE’S CUSTOMER
MATCH ASSOCIATES
THE SEED FIRST-PARTY
PII WITH THE USER'S
GOOGLE LOGIN, WHICH
IS UNLIKELY TO
CHANGE FREQUENTLY,
ENABLING TRACKING
OVER LONG PERIODS
OF TIME.
3 - OPTING-
OUT
OPT-OUT MECHANISM
HAS BEEN DESIGNED,
BUT THE ONUS IS ON
PUBLISHERS TO
CONVINCE USERS TO
OPT-IN IN EXCHANGE
OF CONTENT, MAKING
OPT-OUT CHOICES
QUESTIONABLE IN
PRACTICE.
OPT-OUT MECHANISM
HAS BEEN DESIGNED.
USERS CAN OPPOSE
IDENTIFICATION AND
TARGETING BY RAMPID.
THE ONUS IS ON THE
VARIOUS PARTICIPATING
ACTORS FROM THE
SELLER AND BUYER
SIDES TO CONVINCE
USERS TO OPT-IN IN
EXCHANGE OF CONTENT.
OPT-OUT MECHANISM
HAS BEEN DESIGNED.
CONSUMERS OF
DIFFERENT WALLED
GARDEN PRODUCTS
CAN CHOOSE TO OPT
OUT FROM BEING
TARGETED IN THEIR
LOGGED-IN PRODUCTS.
4 -
TRANSPARENCY
IN TRACKING
SELF-GOVERNED
TRANSPARENCY
MECHANISM WAS
DESIGNED, BUT THE
INCLUDED
INFORMATION IS
UNCLEAR. IT IS
TECHNICALLY ALMOST
IMPOSSIBLE TO
SELF-GOVERNED
TRANSPARENCY
MECHANISM WAS
DESIGNED, BUT THE
INCLUDED
INFORMATION IS
UNCLEAR. IT IS
TECHNICALLY ALMOST
IMPOSSIBLE TO
SELF-GOVERNED
TRANSPARENCY
MECHANISM WAS
DESIGNED, BUT THE
INCLUDED
INFORMATION IS
UNCLEAR. IT IS
TECHNICALLY ALMOST
IMPOSSIBLE TO
21 Si an-Se illa, Pa ham, McGuigan
MONITOR FOR USERS
ALL THE TRACKERS
THAT COLLECT THEIR
DATA AND TARGET
THEM FOR ADS.
MONITOR FOR USERS
ALL THE TRACKERS THAT
COLLECT THEIR DATA
AND TARGET THEM FOR
ADS.
MONITOR FOR USERS
ALL THE TRACKERS
THAT COLLECT THEIR
DATA AND TARGET
THEM FOR ADS.
5- SENSITIVITY
OF TARGETING
ENCOURAGE
TARGETING BASED ON
SENSITIVE CATEGORIES
BY PAIRING
ADVERTISER 1ST-PARTY
& PURCHASED DATA
AND THROUGH ‘LOOK-
ALIKE’ AUDIENCE
TARGETING.
ENCOURAGE TARGETING
BASED ON SENSITIVE
CATEGORIES BY PAIRING
ADVERTISER 1ST-PARTY
& PURCHASED DATA
AND THROUGH ‘LOOK-
ALIKE’ AUDIENCE
TARGETING.
ENCOURAGE
TARGETING BASED ON
SENSITIVE CATEGORIES
BY PAIRING
ADVERTISER 1ST-PARTY
& PURCHASED DATA
AND THROUGH ‘LOOK-
ALIKE’ AUDIENCE
TARGETING.
5 - Discussion and conclusion
Ou s udy de ails and e alua es p i acy implica ions o he inc easingly popula
i s -pa y iden i ica ion a chi ec u es being used ac oss he online ad e ising
supply chain. We apply a i e-pa ypology o show how he p oposed i s -pa y
ID a chi ec u es main ain and, in some cases, e en magni y he p i acy conce ns
associa ed wi h hi d-pa y cookies. . Despi e he ma ke ing o hose ‘cookie-less’
solu ions as ‘p i acy-conscious’ o ‘p i acy- i s ,’ and he a emp o companies o
dis ance hei b ands om he nega i e publici y su ounding online acking,
hese i s -pa y iden i ica ion a chi ec u es ep oduce AdTech’s ma ke logic o da-
a-d i en op imisa ion and e iciency h ough use iden i ica ion & a ge ing, mak-
ing e y li le p og ess, i any, o use s’ p i acy.
The building o dossie s on use s h ough c oss-si e acking is enabled by all he
solu ions we examined. Use iden i ie s a e becoming mo e de e minis ic and pe -
sis en (o en using pe sonal da a such as phone numbe s and e-mail add esses),
po en ially enabling acking o e longe pe iods o ime. The e icacy o he op -
ou and sel -go e ned acking anspa ency mechanisms enabled by hese new ID
a chi ec u es emain ques ionable in p ac ice, as i is hea ily dependen upon
p ope sel -implemen a ion o go e ning a angemen s.
The bu den o ob ain consen o i s -pa y-based iden i ie s is on publishe s, pa -
icipa ing ac o s, and walled ga den p oduc manage s, who need o con ince
use s o accep he a ge ing in o de o mone ise hei con en . These i s -pa y
iden i ica ion me hods also incen i ise ad e ise s o ci cum en a ge ing policies
22 In e ne Policy Re iew 14(3) | 2025
by p o iding g ea e means and incen i es o a ge use s based on sensi i e ca e-
go ies. Wi h he pi o ing o he indus y owa d ‘iden i y g aphs,’ ‘clean ooms,’ and
i s -pa y da a, ad e ise s a e now incen i ised o a ge consume s based on he
ich da a p o iles hey hold on each use . In case sensi i e ca ego ies like ace o
eligion would sound appealing o ad e ise s, hey can easily o e ide exis ing
a ge ing es ic ions h ough ‘look-alike’ audience a ge ing schemes.
Ul ima ely, he business models and wo king p ac ices o he ad e ising indus y
lead o a p i acy-conce ning ad landscape e en wi hou hi d-pa y cookie iden i-
ie s. The implemen a ion o i s -pa y iden i ica ion a chi ec u es by he AdTech
complex enables simila isibili y on consume s, longe consume acking, and
he assembling o mo e sensi i e consume a ge ing, wi h use agency and acke
anspa ency s ill e y ques ionable. AdTech acke s can s ill lea n abou use s’
demog aphics da a, in e es da a, in en da a (in en ion o pu chase), and measu e-
men da a (a ibu ion), main aining hei da a-d i en business model ha does
no go hand-in-hand wi h meaning ul p i acy.
Despi e hei appa en esigna ion o p o-p i acy he o ic, ad e ise s, ad e ising
echnology companies and he publishe s who sell access o audiences ha e no
abandoned hei commi men s o da a-d i en e iciency and op imisa ion (Veale,
2022); hey s ill wan o ecognise, as p ecisely as possible, he p obable beha -
iou s o consume s and he expec ed e u ns on hei ad e ising in es men s
(McGuigan, 2023). C ea o s o i s -pa y iden i ica ion a chi ec u es push o a -
ge ing o use s based on hei i s -pa y da a, c ea ing an ad e ising landscape
whe e only hey can ully capi alise on he shi away om hi d-pa y cookies. The
way hose companies ma ke ‘ i s -pa y p i acy-p ese ing Ad Tech’ appea like a
smokesc een o he eal a ge ing and iden i ica ion solu ions hey a e pushing
and ad oca ing o .
To di e ge om his pa h, we call o he use o use iden i ies in he echnical
sense o be cons ained o p o ec iden i y cons uc ion in he sense o sel hood
(Hildeb and , 2013). Such a e o m would equi e a s uc u al change in how on-
line ad e ising cu en ly wo ks. Ins ead o allowing ad e ising ac o s o collec
and pu chase mo e use s’ da a, egula o s would ha e o limi he a ibu es ha
ad e ise s can collec on use s, and cons ain hei abili y o d aw in e ences ou -
side o he con ex in which use da a we e collec ed. Tha should also include
limi s on ad e ise s’ abili y o assemble a ge ing c i e ia.
The his o y o echnological de elopmen o he deli e y o ads has p o en ime
and again ha change will no come om use s’ abili y o op -ou and unde s and
23 Si an-Se illa, Pa ham, McGuigan
he le el o acking hey ace. No will i eme ge om ad e ise s' conce ns abou
use s’ p i acy. Change has o come om public-in e es egula o s who can ully
app ecia e he s uc u al gaps ha enable p i acy abuse in he online ad e ising
ecosys em.
Re e ences
Alaimo, C. (2022). F om people o objec s: The digi al ans o ma ion o ields. O ganiza ion S udies,
43(7), 1091–1114. h ps://doi.o g/10.1177/01708406211030654
Beau isage, T., Beusca , J.-S., Coa oux, S., & Melle , K. (2024). How online ad e ising a ge s
consume s: The uses o ca ego ies and algo i hmic ools by audience planne s. New Media &
Socie y, 26(10), 6098–6119. h ps://doi.o g/10.1177/14614448221146174
Choi, H., Mela, C. F., Balsei o, S. R., & Lea y, A. (2020). Online display ad e ising ma ke s: A
li e a u e e iew and u u e di ec ions. In o ma ion Sys ems Resea ch, 31(2), 556–575. h ps://doi.o
g/10.1287/is e.2019.0902
C ain, M. (2021). P o i o e p i acy: How su eillance ad e ising conque ed he In e ne . Uni O
Minneso a P ess.
Eng, A. (2024, June 6). Key ends in digi al ad e ising: Adap ing o he new p i acy no m. IAB. h p
s://www.iab.com/blog/2024-key- ends-in-digi al-ad e ising/
Eu opean Commission. Di ec o a e Gene al o Communica ions Ne wo ks, Con en and Technology.
(2023). S udy on he impac o ecen de elopmen s in digi al ad e ising on p i acy, publishe s and
ad e ise s: Final epo . Publica ions O ice. h ps://da a.eu opa.eu/doi/10.2759/294673
Fouad, I., Bielo a, N., Legou , A., & Sa a ijano ic-Djukic, N. (2020). Missed by il e lis s: De ec ing
unknown hi d-pa y acke s wi h in isible pixels (No. a Xi :1812.01514). a Xi . h ps://doi.o g/10.48
550/a Xi .1812.01514
Fouad, I., San os, C., Legou , A., & Bielo a, N. (2022). My cookie is a phoenix: De ec ion,
measu emen , and law ulness o cookie espawning wi h b owse inge p in ing. P oceedings on
P i acy Enhancing Technologies. h ps://pe symposium.o g/pope s/2022/pope s-2022-0063.php
Fou cade, M., & Healy, K. (2017). Seeing like a ma ke . Socio-Economic Re iew, 15(1), 9–29. h ps://d
oi.o g/10.1093/se /mww033
Google. (n.d.-a). Abou he cus ome ma ching p ocess. Google Ads Help. h ps://suppo .google.com/
google-ads/answe /7474263
Google. (n.d.-b). C ea e a cus ome lis . Google Ads Help. h ps://suppo .google.com/google-ads/ans
we /6276125
Google. (n.d.-c). Fo ma you cus ome da a ile. Google Ads Help. h ps://suppo .google.com/google-
ads/answe /7659867?sjid=13277765107387728213-NA#zippy=
Google. (n.d.-d). How Google uses Cus ome Ma ch da a. Google Ads Help. h ps://suppo .google.co
m/google-ads/answe /6334160?sjid=13277765107387728213-NA
24 In e ne Policy Re iew 14(3) | 2025
