Take-Aways (AI)
  • The FAC qua­li­fi­es syste­ma­ti­cal­ly lin­ked infor­ma­ti­on on residential/life situa­ti­on and occu­pa­ti­on as per­so­na­li­ty pro­files under the FADP.
  • Dis­clo­sure of such pro­files to third par­ties is not per­mit­ted wit­hout legal grounds; cons­ents to third par­ties have not been dee­med effective.
  • The FAC weig­hed up the over­all inte­rests and gave prio­ri­ty to the pro­tec­tion of tho­se affec­ted over the purely finan­cial inte­rests of Moneyhouse.
  • Court con­di­ti­ons: 5% Ran­dom check of data accu­ra­cy; 3% Manu­al checks of cre­dit­wort­hi­ness queries deter­mi­ned to be appropriate.
Note to this sum­ma­ry: Money­hou­se was repre­sen­ted in this mat­ter by Wal­der Wyss, the law firm whe­re the aut­hor of this artic­le also works.

Initi­al situation

In the cau­se célèb­re “Money­hou­se”, the Fede­ral Admi­ni­stra­ti­ve Court has his judgment plea­ses Money­hou­se is a busi­ness infor­ma­ti­on ser­vice that offers limi­t­ed infor­ma­ti­on on indi­vi­du­als and com­pa­nies free of char­ge for regi­stered users. More detail­ed infor­ma­ti­on, inclu­ding details of the resi­den­ti­al situa­ti­on and neigh­bors, cre­dit­wort­hi­ness infor­ma­ti­on and, depen­ding on the appli­ca­ble can­to­nal regu­la­ti­ons, land regi­ster infor­ma­ti­on and tax records, is only acce­s­si­ble to paying pre­mi­um mem­bers. As far as per­sons ente­red in the com­mer­cial regi­ster are con­cer­ned, Money­hou­se ent­ries are also inde­xed by search engi­nes, i.e. they are also acce­s­si­ble via Goog­le, for example.

The action under­ly­ing the judgment filed by the FDPIC against Money­hou­se goes back to the second recom­men­da­ti­on of the FDPIC to Money­hou­se (respec­tively at that time the Itonex AG) of Novem­ber 6, 2014, after an initi­al fact-fin­ding exer­cise had been com­ple­ted with the recom­men­da­ti­on (adopted by Itonex) of Novem­ber 14, 2012 (Recom­men­da­ti­on, PDF). In the second recom­men­da­ti­on, the FDPIC took the posi­ti­on that Money­hou­se was pro­ce­s­sing per­so­nal data in a dis­pro­por­tio­na­te man­ner and vio­la­ting the prin­ci­ples of pro­por­tio­na­li­ty, pur­po­se limi­ta­ti­on, trans­pa­ren­cy and data accuracy.

In addi­ti­on, Money­hou­se dis­c­lo­ses per­so­na­li­ty pro­files to third par­ties. The­re is no justi­fi­ca­ti­on for this, which is why the dis­clo­sure is unlawful (Art. 12 Para. 2 lit. c FDPA).

Money­hou­se has con­firm­ed the recom­men­da­ti­on of the FDPIC only par­ti­al­ly accept­ed, whereu­pon the FDPIC by legal action (pur­su­ant to Art. 29 para. 4 FDPA) rea­ched the Fede­ral Admi­ni­stra­ti­ve Court.

Con­side­ra­ti­ons of the FAC

Pro­ce­du­ral points

The FAC first had to cla­ri­fy pro­ce­du­ral issues. The first of the­se que­sti­ons con­cer­ned the Admis­si­bi­li­ty of the request for a decla­ra­to­ry judgment of the FDPIC. The FDPIC had deman­ded a fin­ding that Money­hou­se had crea­ted per­so­na­li­ty pro­files wit­hout justi­fi­ca­ti­on and had thus vio­la­ted the per­so­na­li­ty of many per­sons. In this case, the FAC denies the right under Art. 25 BZP neces­sa­ry inte­rest in a decla­ra­to­ry judgment, becau­se the effects of any unlawful pro­ce­s­sing of data would be FDPIC would be eliminated.

The requi­red amount of the Defi­ni­ten­ess of the fur­ther legal claims of the plain­ti­ffThe BVG accepts this legal request, e.g. the legal request that Money­hou­se be obli­ged to dele­te “all links on Money­hou­se that enable the crea­ti­on of per­so­na­li­ty pro­files of per­sons who have not con­sen­ted to this in accordance with the law”. Alt­hough the con­tent of the legal terms used is not clear and part­ly dis­pu­ted, the FAC accepts this legal cla­im (wit­hout refe­rence to the cor­re­spon­ding refe­ren­ces in the Street View decis­i­on of the BVGer), becau­se it is suf­fi­ci­ent­ly clear with refe­rence to the state­ment of the grounds for the action and can be ele­va­ted to a judgment in the event of an affir­ma­ti­on – if neces­sa­ry with judi­cial clarifications:

What is to be under­s­tood by legal­ly suf­fi­ci­ent con­sent in con­nec­tion with the crea­ti­on of per­so­na­li­ty pro­files then fol­lows from the legal basis (cf. Art. 4 para. 5 FDPA). Regar­ding the qua­li­fi­ca­ti­on of data pro­ce­s­sing as the crea­ti­on of a per­so­na­li­ty pro­fi­le in a spe­ci­fic indi­vi­du­al case, the lite­ra­tu­re, case law and mate­ri­als can be refer­red to in the con­text of the sub­stan­ti­ve examination.

Sub­stan­ti­ve points

Preli­mi­na­ry note

From a sub­stan­ti­ve point of view, the FAC exami­nes abo­ve all the Dis­clo­sure of per­so­na­li­ty pro­files to third par­tieswhich is inad­mis­si­ble wit­hout a justi­fi­ca­ti­on (Art. 12 para. 2 lit. c FDPA). Sin­ce it affirm­ed the­se facts, it refrai­ned from exami­ning the fur­ther cri­ti­cisms of the FDPIC and the­r­e­fo­re lea­ves open whe­ther Money­hou­se com­plies with the gene­ral data pro­ce­s­sing prin­ci­ples of pro­por­tio­na­li­ty and pur­po­se limi­ta­ti­on. Howe­ver, the fol­lo­wing had to be exami­ned in more detail Search engi­ne index­ing, the rea­sonable mea­su­res taken to ensu­re the Data cor­rect­ness (Art. 5 FDPA) and for data queries the checkwhe­ther the queries are real­ly made for the pur­po­se of checking cre­dit­wort­hi­ness (Art. 13 Par. 2 lit. c FDPA).

About the con­cept of per­so­na­li­ty profile

Initi­al­ly, it was que­stionable whe­ther Money­hou­se dis­c­lo­sed per­so­na­li­ty pro­files to regi­stered paying cus­to­mers (Art. 12 para. 2 lit. c FDPA). This invol­ves the fol­lo­wing data points:

  • First and last name
  • Resi­dence, postal code
  • Date of birth and age
  • cur­rent pro­fes­si­on and pro­fes­sio­nal career
  • Hou­se­hold mem­bers and living situa­ti­on (with link to Goog­le Street View images as well as neigh­bors and old addresses/residences).

As a result, the FAC affirms that per­so­na­li­ty pro­files are pre­sent here:

The Defen­dant shall pro­vi­de regi­stered and paying users, in addi­ti­on to the infor­ma­ti­on requi­red for iden­ti­fi­ca­ti­on, such as sur­na­me, first name, cur­rent address and, if appli­ca­ble, date of birth – pro­vi­ded that the rele­vant data is available – with Syste­ma­ti­cal­ly lin­ked infor­ma­ti­on on the pri­va­te housing and living situa­ti­on natu­ral per­sons con­cer­ned, i.e. con­cer­ning their hou­se­hold mem­bers and neigh­bors, and thus to an essen­ti­al aspect of their personality. […] 

In the case of per­sons recor­ded in the Com­mer­cial Regi­ster, the addi­tio­nal Natio­na­li­ty announ­ced as well as their pro­fes­sio­nal care­er and your pro­fes­sio­nal net­work, wher­eby a fur­ther sub­di­vi­si­on of the per­so­na­li­ty is affected.

One must pro­ba­b­ly even under­stand the BVGer here in such a way that two pro­files distin­gu­is­ha­ble from each other are pre­sent, name­ly on the one hand the data com­plex “resi­den­ti­al and living situa­ti­on” and on the other hand the com­plex “care­er and pro­fes­sio­nal network”.

The rela­ti­on­ship bet­ween per­so­na­li­ty pro­fi­le and cre­dit assessment

In doing so, the FAC right­ly took into account that the justi­fi­ca­ti­on of the cre­dit­wort­hi­ness check of Art. 13 para. 2 lit. c FDPA express­ly exclu­des the pro­ce­s­sing of per­so­na­li­ty pro­files, and that, con­ver­se­ly Infor­ma­ti­on neces­sa­ry for checking cre­dit­wort­hi­ness, nor­mal­ly not a per­so­na­li­ty pro­fi­le can repre­sent. This rai­ses the que­sti­on of which data is still meaningful­ly requi­red for checking cre­dit­wort­hi­ness and the­r­e­fo­re does not con­sti­tu­te a per­so­na­li­ty pro­fi­le. The ans­wer to this que­sti­on can­not be unam­bi­guous, howe­ver, becau­se checking cre­dit­wort­hi­ness is not an abso­lu­te mea­su­re, but a pro­cess that can only be made more pre­cise by using lar­ger amounts of data. As a result, it is the­r­e­fo­re neces­sa­ry to weigh up the public and pri­va­te Inte­rest in the relia­ble veri­fi­ca­ti­on of cre­dit­wort­hi­ness on the one hand and the Per­so­na­li­ty pro­tec­tion on the other hand. In this balan­cing pro­cess, the prin­ci­ple of the Data cor­rect­ness (Art. 5 FDPA), i.e., in the con­text of cre­dit assess­ment, the con­cern that the iden­ti­ty of a (poten­ti­al) debtor is estab­lished and that his cre­dit­wort­hi­ness is cor­rect­ly assessed.

Howe­ver, the FAC does not take such a syste­ma­tic approach and does not con­sider the prin­ci­ple of data accu­ra­cy. The BVGer’s rea­so­ning on this point the­r­e­fo­re remains strikin­gly vague and con­ta­ins state­ments that can hard­ly be gene­ra­li­zed. The FAC even seems to imply that cre­dit­wort­hi­ness may only be asses­sed on the basis of data that each in its­elf alre­a­dy con­ta­ins unam­bi­guous state­ments on creditworthiness:

It is true that con­clu­si­ons can be drawn from such infor­ma­ti­on [sc. living situa­ti­on] with regard to the finan­cial cir­cum­stances of a natu­ral per­son and for this very rea­son an essen­ti­al aspect of the per­so­na­li­ty is illu­mi­na­ted. Howe­ver, the infor­ma­ti­on can also lead to incor­rect assump­ti­ons and thus do not relia­bly pro­ve the cre­dit­wort­hi­ness of a natu­ral per­son.

This can­not be cor­rect. It is true that the housing situa­ti­on can of cour­se lead to fal­se assump­ti­ons about cre­dit­wort­hi­ness (a mil­lion­aire can also live in a one-room apart­ment). But this is also true for all other data pro­ce­s­sed by Money­hou­se, inclu­ding such essen­ti­al para­me­ters as remin­ders (even a mil­lion­aire can be an undi­sci­plined payer). A broad data base ser­ves pre­cis­e­ly to com­pen­sa­te for unavo­ida­ble defi­ci­en­ci­es in the infor­ma­ti­ve value of indi­vi­du­al data points by aggre­ga­ti­on (which is why the Ger­man BDSG in Sec­tion 28b(3) cre­dit ratings based on address only).

A simi­lar com­ment by the FAC in this context:

In addi­ti­on, the lin­king of the dis­c­lo­sed data may make it pos­si­ble to Infe­ren­ces about per­so­nal data requi­ring spe­cial pro­tec­tion within the mea­ning of Art. 3 let. c FDPA in par­ti­cu­lar on the sexu­al ori­en­ta­ti­on. By means of infor­ma­ti­on on resi­den­ti­al part­ners and their age, it is pos­si­ble to draw con­clu­si­ons about the sexu­al ori­en­ta­ti­on of the per­sons con­cer­ned in gene­ral – not only with regard to same-sex cou­ples, to which the plain­ti­ff refers – or, alter­na­tively, it is pos­si­ble to deter­mi­ne the sexu­al ori­en­ta­ti­on of the per­sons con­cer­ned. fal­se assump­ti­ons made, so when fel­low stu­dents or good fri­ends live together.

Infor­ma­ti­on about the living situa­ti­on cer­tain­ly allo­ws spe­cu­la­ti­on about inti­ma­te rela­ti­on­ships. Howe­ver, to regard them as par­ti­cu­lar­ly wort­hy of pro­tec­tion would be going much too far.

(Also) decisi­ve for the­se state­ments of the BVGer was cer­tain­ly also the alre­a­dy in the Street View ruling of the Fede­ral Supre­me Court that tech­ni­cal pro­gress gene­ral­ly requi­res a strict appli­ca­ti­on of data pro­tec­tion law:

As a result of tech­no­lo­gi­cal deve­lo­p­ments in recent years, the sto­rage capa­ci­ty, per­mea­bi­li­ty and net­wor­king of infor­ma­ti­on have increa­sed enorm­ously […]. Sin­ce elec­tro­nic data pro­ce­s­sing can be used to store, link and repro­du­ce per­so­nal infor­ma­ti­on on any sca­le, even harm­less infor­ma­ti­on that could easi­ly be attri­bu­ted to the public sphe­re can be tur­ned into per­so­na­li­ty pro­files that are actual­ly wort­hy of pro­tec­tion. con­den­se […]. The­se sto­rage and eva­lua­ti­on pos­si­bi­li­ties of auto­ma­tic data pro­ce­s­sing and the lin­king of auto­ma­ted data stocks have made the Per­so­na­li­ty pro­fil­ing easier and more fre­quent­ly beco­me […]. The inter­lin­ked per­so­nal data reach rela­tively quick­ly an infor­ma­ti­on den­si­tythat reve­al beha­vi­oral pat­terns and per­so­na­li­ty pro­files […]. The per­sons con­cer­ned often have no know­ledge of the exi­stence of a pro­fi­le and thus can­not con­trol its accu­ra­cy and use.

It pro­ba­b­ly also had an influence on the fact that Money­hou­se had expan­ded the pre­mi­um offer with que­sti­ons such as “Who does the per­son you are loo­king for live with? Does she live alo­ne? And who are her neigh­bors? Who owns the pro­per­ty she lives in?”, and pro­ba­b­ly also the fact that this is not Moneyhouse’s first pro­ce­du­re. As a result, the ruling can the­r­e­fo­re hard­ly be generalized.

Final­ly, the FAC sta­tes that the con­cept of a per­so­na­li­ty pro­fi­le is inde­pen­dent of whe­ther the indi­vi­du­al data used for this pur­po­se is publicly available:

For the qua­li­fi­ca­ti­on of a com­pi­la­ti­on of per­so­nal data as a per­so­na­li­ty pro­fi­le, it must be agreed with the plain­ti­ff that the ori­gin of the data or the type of data source is irrele­vant. […] In the case of data from public sources such as the com­mer­cial or tax regi­ster, from offi­ci­al gazet­tes or the land regi­ster, the per­sons con­cer­ned could not, moreo­ver, deter­mi­ne their natu­re and scope due to the cor­re­spon­ding legal obli­ga­ti­ons to dis­c­lo­se data. The only rele­vant que­sti­on in this case is whe­ther the lin­king of infor­ma­ti­on – inclu­ding infor­ma­ti­on that is alre­a­dy acce­s­si­ble to the public or that is not par­ti­cu­lar­ly wort­hy of pro­tec­tion within the mea­ning of the FDPA The infor­ma­ti­on pro­vi­ded by the test results is based on the data of the test per­son and the data of the test subject.

Justi­fi­ca­ti­on of data sharing?

Sin­ce the FAC thus assu­med that per­so­na­li­ty pro­files were dis­c­lo­sed to third par­ties, grounds for justi­fi­ca­ti­on (Art. 13 Par. 1 FDPA) to be exami­ned. A justi­fi­ca­ti­on by Law was obvious­ly out of the question.

A Con­sent was also not pre­sent, in par­ti­cu­lar becau­se Money­hou­se acqui­res data from other com­pa­nies and obta­ins it from public sources, so that the data sub­jects have no know­ledge of the pro­ce­s­sing by Moneyhouse:

This argu­ment is the­r­e­fo­re just as unhel­pful as the one accor­ding to which an expli­cit con­sent of the per­sons con­cer­ned had been given, in that they had given their con­sent to Post AG had con­sen­ted to the for­war­ding of addres­ses for cre­dit report­ing agen­ci­es, among others, with refe­rence to a for­war­ding order/change of resi­dence. This means that the per­sons con­cer­ned have did not con­sent to the crea­ti­on of a per­so­na­li­ty pro­fi­le vis-à-vis the defen­dant.

The FAC should real­ly have asked its­elf here whe­ther con­sent real­ly repres­ents a decla­ra­ti­on of intent that must be recei­ved and is only effec­ti­ve vis-à-vis its addres­see, or whe­ther it does not have to be effec­ti­ve vis-à-vis all pro­ces­sors when it is given. In view of the cur­rent divi­si­on of labor in pro­ce­s­sing pro­ce­du­res, but also in view of the pur­po­se of con­sent under data pro­tec­tion law, the lat­ter will have to be assu­med. Effec­ti­ve con­sent must rela­te in each case to the para­me­ters rele­vant under data pro­tec­tion law i.e. tho­se points which must also be reco­gnizable. Fre­quent­ly (and pro­ba­b­ly also here) it is the­r­e­fo­re suf­fi­ci­ent if the con­sent refers to Edi­tors cate­go­ries refers. Con­sent expli­ci­t­ly rela­ted to Money­hou­se was the­r­e­fo­re hard­ly neces­sa­ry here.

In the end, howe­ver, it was rather decisi­ve that Money­hou­se obta­ins data from other com­pa­nies and from public sources and that the per­sons con­cer­ned the­r­e­fo­re not neces­s­a­ri­ly know­ledge of the fact that data about them is pro­ce­s­sed for the pur­po­se of cre­dit checks. For this rea­son also Art. 12 par. 3 FDPA not fulfilled.

Within the frame­work of a over­all weig­hing of inte­rests (Art. 13 para. 1 FDPA) then show­ed that the inte­rest of the data sub­jects in pre­ven­ting Money­hou­se from pro­ce­s­sing their data out­weig­hed this inte­rest. The FAC reco­gnized a public inte­rest in the Money­hou­se plat­form. Howe­ver, if data pro­ce­s­sing is not rele­vant to cre­dit­wort­hi­ness, which is the case for per­so­na­li­ty pro­files (see abo­ve), the­re is no such inte­rest. It the­r­e­fo­re remains the purely finan­cial inte­rest of Money­hou­se. This can­not out­weigh the inte­rest of the data sub­jects, espe­ci­al­ly sin­ce the ope­ra­ti­on of the plat­form is also pos­si­ble in com­pli­ance with data pro­tec­tion law.

Index­ing by search engines

The FDPIC had fur­ther deman­ded that Money­hou­se must Search engi­ne index­ing of infor­ma­ti­on about per­sons ente­red in the com­mer­cial regi­ster. Such ent­ries from Money­hou­se could only be dis­play­ed on Goog­le if not only the name of the per­son in que­sti­on was sear­ched for, but also the term “money­hou­se”, as is the case today with Zefix. The BVG rejects this legal request: First­ly, Money­hou­se has no or only limi­t­ed pos­si­bi­li­ties to influence the publi­ca­ti­on of search results. The imple­men­ta­ti­on of the request would the­r­e­fo­re alre­a­dy be dif­fi­cult. On the other hand, the data pro­ce­s­sing with the imple­men­ta­ti­on of the appro­ved legal requests is legal­ly com­pli­ant. It the­r­e­fo­re makes no dif­fe­rence in terms of data pro­tec­tion law if lawful ent­ries can also be found via search engines.

On the contrary:

[…] the Findabili­ty of data via search engi­nes is posi­ti­ve from a data pro­tec­tion point of view in that is to be con­side­red as Trans­pa­ren­cy regar­ding data pro­ce­s­sing is crea­ted, which enables the effec­ti­ve exer­cise of the rights of access, rec­ti­fi­ca­ti­on and dele­ti­on of one’s own data.

The FAC thus reco­gnizes search engi­nes as pro­mo­ting trans­pa­ren­cy, which is likely to be signi­fi­cant bey­ond this case. Howe­ver, this result is not enti­re­ly sur­pri­sing: Even the FDPIC has regard­ed infor­ma­ti­on on Inter­net pages as reco­gnizable within the mea­ning of Art. 4 para. FDPA reco­gnized (of all things, in its first recom­men­da­ti­on in Money­hou­se, Novem­ber 15, 2012; PDF; para. 8).

Final­ly, Money­hou­se is not obli­ga­ted to sup­port affec­ted per­sons with dele­ti­on requests in order to have the cor­re­spon­ding results remo­ved more quick­ly from search engi­nes. The dele­ti­on in its own domain is sufficient.

Veri­fi­ca­ti­on of data correctness

The FDPIC had final­ly deman­ded that Money­hou­se veri­fy the accu­ra­cy of its data­ba­se in an appro­pria­te per­cen­ta­ge of the queries made, to be deter­mi­ned by the court. The FAC fol­lo­wed this request. Art. 5 para. 1 FDPA requi­res pro­ces­sors to take appro­pria­te mea­su­res to ensu­re the accu­ra­cy of the per­so­nal data they process.

Accor­ding to the BVGer, the neces­sa­ry mea­su­res must be deter­mi­ned in a pro­por­tio­na­te man­ner and with a view to the data pro­tec­tion risks. As a result, the data should be review­ed in the ratio of 5% of the queries made:

In view of the lar­ge num­ber of per­sons affec­ted by the data pro­ce­s­sing of the defen­dant and the importance of the accu­ra­cy, com­ple­ten­ess and time­liness of data both in the con­text of a per­mis­si­bly con­duc­ted after con­sent of the data sub­jects, sen­si­ti­ve pro­ce­s­sing of per­so­na­li­ty pro­files as well as in the area of Cre­dit reports appears appro­pria­te to review Defendant’s data set for accu­ra­cy in rela­ti­on to 5% queries made on its platform.

Howe­ver, the ruling does not explain why the limit should be 5%. This check is, howe­ver, an orga­nizatio­nal mea­su­re of data secu­ri­ty within the mea­ning of Art. 7 FDPA. Even if the Fede­ral Supre­me Court does not argue so, Art. 8 para. 2 FDPOaccor­ding to which secu­ri­ty mea­su­res depend on the pur­po­se of the data pro­ce­s­sing, the type and scope of the data pro­ce­s­sing, the risk assess­ment and the sta­te of the art. From this, it must be deter­mi­ned in each indi­vi­du­al case in an over­all assess­ment what spe­ci­fi­cal­ly appears to be appro­pria­te, wher­eby the­re is room for dis­creti­on for the data pro­ces­sor. For this rea­son alo­ne, the 5% limit of the FAC can­not be applied unseen to other cases.

Exami­na­ti­on of the pur­po­se of pro­ce­s­sing queries (Art. 13 para. 2 lit. c FDPA)

When dis­clo­sing cre­dit­wort­hi­ness infor­ma­ti­on to third par­ties, the pro­ces­sor must check whe­ther the requi­re­ments for dis­clo­sure are met, i.e. whe­ther the query is made for the pur­po­se of checking cre­dit­wort­hi­ness. The FDPIC requi­red a cor­re­spon­ding check at 5% of the queries. The FAC con­siders an exami­na­ti­on of 3% to be appro­pria­te:

Inten­si­fy­ing manu­al checks does lead to addi­tio­nal work on the part of the defen­dant, but ulti­m­ate­ly it is also in its own inte­rest that the per­so­nal data it pro­ce­s­ses is not used by third par­ties for inap­pro­pria­te pur­po­ses. In addi­ti­on, accor­ding to the infor­ma­ti­on pro­vi­ded by the defen­dant, an auto­ma­ted veri­fi­ca­ti­on system has been in the plan­ning stage for some time, which should mini­mi­ze the con­trol effort. The­se facts and the The inte­rest of the lar­ge num­ber of data sub­jects affec­ted by the defendant’s data pro­ce­s­sing, which is to be weigh­ted high­ly. in the pro­tec­tion of their data, in par­ti­cu­lar also with regard to any lawful pro­ce­s­sing of data that may take place. sen­si­ti­ve per­so­na­li­ty pro­files make a regu­lar check of the pro­ofs of inte­rest at the time of the cre­dit­wort­hi­ness query in the ratio of 3 % to the queries made on the defendant’s plat­form appear reasonable.

In the end, it remains open how the BGer justi­fi­es the 3% requi­re­ment. To a cer­tain ext­ent, this may be in the natu­re of things – limits such as the­se are always arbi­tra­ry to a cer­tain ext­ent. What is striking, howe­ver, is the major dif­fe­rence bet­ween this and the legal situa­ti­on in Ger­ma­ny, which is not sub­stan­tia­ted in the Fede­ral Supre­me Court’s ruling. Accor­ding to § 29 para. 2 BDSG has “the trans­mit­ting agen­cy […] to con­duct sam­pling pro­ce­du­res in accordance with Sec­tion 10 (4) sen­tence 3 and, in doing so, also to deter­mi­ne and veri­fy the exi­stence of a legi­ti­ma­te inte­rest on a case-by-case basis.” Howe­ver, the law does not spe­ci­fy the fre­quen­cy of sam­pling. Accor­ding to the lite­ra­tu­re (Hei­ne­mann, ZD 2014, 294), the fol­lo­wing applies:

The law lacks rigid spe­ci­fi­ca­ti­ons for a mini­mum quan­ti­ty of the sub­stances to be trea­ted in accordance with § 10 Para. 4 Sen­tence 3. BDSG samples to be taken is the­r­e­fo­re justi­fi­ed. The deter­mi­na­ti­on of the con­trol quo­ta is instead a Que­sti­on of case-by-case assess­ment. First of all, the design of the respec­ti­ve retrie­val pro­ce­du­re with its indi­vi­du­al risk situa­ti­on must be taken into account. Data wort­hy of litt­le pro­tec­tion justi­fy litt­le effort in sam­pling – and vice ver­sa. Fur­ther­mo­re, the qua­li­ty of the sta­tis­ti­cal pro­ce­du­re used must be taken into account. If the method is cor­rect, a few samples can be enough to get an accu­ra­te pic­tu­re of the who­le. Final­ly, con­side­ra­ti­ons of rea­son­ab­leness for the obli­ga­ted body can­not be dis­re­gard­ed. Par­ti­cu­lar­ly in the case of mass pro­ce­e­dings, even ran­dom sam­pling obli­ga­ti­ons can in the ran­ge of low per mil­le values lead to con­sidera­ble orga­nizatio­nal and finan­cial chal­lenges […]. In view of the fact that this is “only” a secon­da­ry obli­ga­ti­on under data pro­tec­tion law and not the core busi­ness of the com­pa­ny, it must be ensu­red that the acting enti­ty is not cripp­led by the sam­pling effort.

Even if the sam­pling fre­quen­cy is a que­sti­on of risk assess­ment and rea­son­ab­leness in indi­vi­du­al cases, the value of 3% appears to be very high. The Düs­sel­dor­fer Kreis, a body of Ger­man data pro­tec­tion super­vi­so­ry aut­ho­ri­ties, seems to con­sider a review rate of 2‰ to be suf­fi­ci­ent, i.e., around 150 times less than the rate requi­red by the BVGer. The BVGer’s ruling is the­r­e­fo­re unli­kely to be gene­ra­lizable on this point as well. In par­ti­cu­lar, the FAC did not set a gene­ral review quo­ta of 3% that would app­ly bey­ond the case under review (see abo­ve regar­ding the review of correctness).