{"id":7086,"date":"2026-06-25T08:24:29","date_gmt":"2026-06-25T06:24:29","guid":{"rendered":"https:\/\/www.insightsofa.com\/?p=7086"},"modified":"2026-06-29T15:09:48","modified_gmt":"2026-06-29T13:09:48","slug":"kdyz-se-z-metriky-stane-cil","status":"publish","type":"post","link":"https:\/\/www.insightsofa.com\/cs\/kdyz-se-z-metriky-stane-cil\/","title":{"rendered":"Kdy\u017e se z metriky stane c\u00edl"},"content":{"rendered":"

Existuje paradox, kter\u00fd tr\u00e1p\u00ed z\u00e1kaznickou zku\u0161enost v\u00edc ne\u017e \u0161patn\u00e1 data nebo nedostatek rozpo\u010dtu. \u0158\u00edk\u00e1me mu Goodhart\u016fv z\u00e1kon<\/strong>. Zformuloval ho britsk\u00fd ekonom Charles Goodhart v roce 1975 v kontextu m\u011bnov\u00e9 politiky, ale jeho platnost sah\u00e1 daleko za hranice centr\u00e1ln\u00edho bankovnictv\u00ed: jakmile se z metriky stane c\u00edl, p\u0159est\u00e1v\u00e1 b\u00fdt dobrou metrikou.<\/p>\n

V CX (customer experience, z\u00e1kaznick\u00e1 zku\u0161enost) se tento z\u00e1kon projevuje s mimo\u0159\u00e1dnou p\u0159esnost\u00ed a s d\u016fsledky, kter\u00e9 jsou pro firmy fat\u00e1ln\u00ed, p\u0159esto\u017ee je na prvn\u00ed pohled v\u016fbec nevid\u00ed.<\/p>\n

Sk\u00f3re roste. Z\u00e1kazn\u00edci odch\u00e1zej\u00ed.<\/h2>\n

P\u0159edstavte si firmu, jej\u00ed\u017e NPS (Net Promoter Score, sk\u00f3re doporu\u010den\u00ed z\u00e1kazn\u00edk\u016f) stoup\u00e1 ka\u017ed\u00e9 \u010dtvrtlet\u00ed. Management je spokojen\u00fd. Bonusy jsou vyplaceny. A pak p\u0159ijdou ro\u010dn\u00ed data a uk\u00e1\u017ee se, \u017ee churn \u00a0(m\u00edra odchodu z\u00e1kazn\u00edk\u016f) z\u016fst\u00e1v\u00e1 stejn\u00fd nebo se dokonce zhor\u0161uje.<\/p>\n

Jak je to mo\u017en\u00e9?<\/strong><\/p>\n

Odpov\u011b\u010f nen\u00ed v z\u00e1kazn\u00edc\u00edch. Je v syst\u00e9mu odm\u011b\u0148ov\u00e1n\u00ed, kter\u00fd se mezit\u00edm nau\u010dil optimalizovat metriku, nikoli zku\u0161enost. Zam\u011bstnanci kontaktn\u00edho centra, jejich\u017e bonusy z\u00e1vis\u00ed na NPS, p\u0159irozen\u011b za\u010dnou d\u011blat v\u011bci, kter\u00e9 sk\u00f3re zv\u00fd\u0161\u00ed, ale z\u00e1kaznickou zku\u0161enost nutn\u011b nezlep\u0161\u00ed. Volaj\u00ed z\u00e1kazn\u00edk\u016fm p\u0159ed odesl\u00e1n\u00edm pr\u016fzkumu. \u017d\u00e1daj\u00ed je o vysok\u00e9 hodnocen\u00ed. Odkl\u00e1daj\u00ed pr\u016fzkum na okam\u017eik, kdy je z\u00e1kazn\u00edk v dobr\u00e9 n\u00e1lad\u011b. Nebo se vyhnou jeho odesl\u00e1n\u00ed \u00fapln\u011b, pokud tu\u0161\u00ed, \u017ee interakce prob\u011bhla \u0161patn\u011b.<\/p>\n

Ipsos tento jev pojmenoval jako survey gaming – manipulaci s pr\u016fzkumy ze strany zam\u011bstnanc\u016f nebo intern\u00edch stakeholder\u016f s c\u00edlem um\u011ble nav\u00fd\u0161it sk\u00f3re. Podle jejich anal\u00fdzy to m\u016f\u017ee m\u00edt formy od p\u0159\u00edm\u00e9ho prosen\u00ed z\u00e1kazn\u00edka o konkr\u00e9tn\u00ed hodnocen\u00ed a\u017e po klik\u00e1n\u00ed na spokojenostn\u00ed termin\u00e1ly v provozovn\u00e1ch bez jak\u00e9koliv kontroly, kdo a kolikr\u00e1t tla\u010d\u00edtko zm\u00e1\u010dkne (Ipsos, Gaming the Score: When Feedback Becomes Fiction, 2026).<\/p>\n

Data jsou v tomto ohledu konzistentn\u00ed. Studie z roku 2024 publikovan\u00e1 v International Journal of Market Research potvrdila, \u017ee samotn\u00e1 metodologie v\u00fdpo\u010dtu NPS vn\u00e1\u0161\u00ed do v\u00fdsledk\u016f statistick\u00fd \u0161um, kter\u00fd zt\u011b\u017euje detekci skute\u010dn\u00fdch zm\u011bn v sentimentu z\u00e1kazn\u00edk\u016f, a to je\u0161t\u011b p\u0159edt\u00edm, ne\u017e se kdokoliv pokus\u00ed v\u00fdsledky z\u00e1m\u011brn\u011b ovlivnit (Why NPS Is Lying to You About Customer Experience, CMSWire, 2026).<\/p>\n

Goodhart\u016fv z\u00e1kon v praxi: od NHS po Wells Fargo<\/h2>\n

Aby bylo jasn\u00e9, \u017ee nejde o fenom\u00e9n specifick\u00fd pro CX, stoj\u00ed za to p\u0159ipomenout dva klasick\u00e9 p\u0159\u00edklady.<\/p>\n

Britsk\u00e9 nemocnice dostaly v 80. letech c\u00edl: \u017e\u00e1dn\u00fd pacient nesm\u00ed \u010dekat na o\u0161et\u0159en\u00ed d\u00e9le ne\u017e 18 t\u00fddn\u016f. C\u00edl byl spln\u011bn. \u010cekac\u00ed doby klesly. P\u0159i bli\u017e\u0161\u00edm pohledu se ale uk\u00e1zalo, \u017ee l\u00e9ka\u0159i z\u00e1m\u011brn\u011b odkl\u00e1dali form\u00e1ln\u00ed referral pacient\u016f, aby nespustili odpo\u010d\u00edt\u00e1v\u00e1n\u00ed. Slo\u017eit\u00fdm p\u0159\u00edpad\u016fm se vyh\u00fdbali, proto\u017ee sni\u017eovaly statistiku. Jednoduch\u00fdch p\u0159\u00edpad\u016f p\u0159ib\u00fdvalo na \u00fakor t\u011bch, kter\u00e9 skute\u010dn\u011b pot\u0159ebovaly rychlou p\u00e9\u010di. Metrika byla spln\u011bna. P\u00e9\u010de se nezlep\u0161ila.<\/p>\n

Wells Fargo je v kontextu firemn\u00edho sv\u011bta asi nejcitovan\u011bj\u0161\u00edm p\u0159\u00edkladem destrukce metriky. Management za\u010dal v po\u010d\u00e1tku 2000. let pe\u010dliv\u011b sledovat po\u010det nov\u011b otev\u0159en\u00fdch \u00fa\u010dt\u016f v ka\u017ed\u00e9 pobo\u010dce a tento ukazatel propojil s pr\u00e9mii, kari\u00e9rn\u00edm postupem i udr\u017een\u00edm zam\u011bstn\u00e1n\u00ed. V\u00fdsledek byl p\u0159edv\u00eddateln\u00fd: mana\u017ee\u0159i za\u010dali otv\u00edrat \u00fa\u010dty bez v\u011bdom\u00ed z\u00e1kazn\u00edk\u016f. V roce 2017 banka p\u0159iznala, \u017ee \u0161lo o 3,5 milionu fale\u0161n\u00fdch \u00fa\u010dt\u016f. Pokuta regul\u00e1tor\u016f dos\u00e1hla 185 milion\u016f dolar\u016f, celkov\u00e9 vyrovn\u00e1n\u00ed s americk\u00fdm ministerstvem spravedlnosti a SEC pak 3 miliardy dolar\u016f. P\u011bt tis\u00edc t\u0159i sta zam\u011bstnanc\u016f bylo propu\u0161t\u011bno. Nikdo z nich nevymyslel podvod z vlastn\u00ed iniciativy, ale pouze reagoval na syst\u00e9m pob\u00eddek, kter\u00fd byl nastaven \u0161patn\u011b (Ethics Unwrapped, University of Texas, 2020).<\/p>\n

Kl\u00ed\u010dov\u00e1 podm\u00ednka v\u0161ak z\u016fst\u00e1v\u00e1 ve v\u0161ech t\u011bchto p\u0159\u00edpadech opomenuta: nikdo v\u011bdom\u011b nezam\u00fd\u0161lel zkazit metriku<\/strong>. Syst\u00e9m s\u00e1m ji zkazil t\u00edm, \u017ee ji pov\u00fd\u0161il na c\u00edl.<\/p>\n

Probl\u00e9m s NPS, CSAT a CES, jakmile jsou na nich bonusy<\/h2>\n

T\u0159i nejroz\u0161\u00ed\u0159en\u011bj\u0161\u00ed CX metriky NPS, CSAT (Customer Satisfaction Score, sk\u00f3re spokojenosti z\u00e1kazn\u00edk\u016f) a CES (Customer Effort Score, sk\u00f3re vynalo\u017een\u00e9ho \u00fasil\u00ed) sd\u00edl\u00ed jeden struktur\u00e1ln\u00ed rys: jsou to proxy metriky. Nem\u011b\u0159\u00ed p\u0159\u00edmo loajalitu, spokojenost nebo n\u00edzk\u00e9 t\u0159en\u00ed. M\u011b\u0159\u00ed zpr\u00e1vu z\u00e1kazn\u00edka o tom, jak loaj\u00e1ln\u00ed, spokojen\u00fd nebo nenam\u00e1han\u00fd se c\u00edt\u00ed. To je d\u016fle\u017eit\u00fd rozd\u00edl.<\/p>\n

Proxy metrika je ze sv\u00e9 podstaty zraniteln\u00e1. Jakmile existuje tlak finan\u010dn\u00ed nebo kari\u00e9rn\u00ed na jej\u00ed v\u00fdsledek, lid\u00e9 optimalizuj\u00ed proxy, ne realitu, kterou m\u00e1 reprezentovat.<\/p>\n

Fred Reichheld, zakladatel NPS a partner Bain & Company, s\u00e1m tuto dynamiku pojmenoval jako nejv\u011bt\u0161\u00ed syst\u00e9mov\u00e9 selh\u00e1n\u00ed v implementaci NPS: „Bad things happen when you just focus on it as a score. When scores are the objective, they no longer help people. It doesn’t inspire them to learn. All they want is a 10.“ (Bain & Company, Refocusing NPS for Earned Growth, 2022). A dodal je\u0161t\u011b explicitn\u011bji: programy, kter\u00e9 propojuj\u00ed NPS s bonusy frontliniov\u00fdch zam\u011bstnanc\u016f, typicky zkolabuj\u00ed do jednoho a\u017e dvou let.<\/p>\n

Ve sv\u00e9m \u010dl\u00e1nku Net Promoter 3.0 v Harvard Business Review (2021) Reichheld spolu s kolegy z Bain & Company identifikoval linking Net Promoter Scores to bonuses for frontline employees jako jeden z hlavn\u00edch zp\u016fsob\u016f zneu\u017eit\u00ed, kter\u00fd podryl kredibilitu cel\u00e9ho syst\u00e9mu. Jako reakci na toto selh\u00e1n\u00ed navrhl dopl\u0148uj\u00edc\u00ed metriku Earned Growth Rate (m\u00edru zaslou\u017een\u00e9ho r\u016fstu) kter\u00e1 m\u011b\u0159\u00ed pod\u00edl p\u0159\u00edjm\u016f poch\u00e1zej\u00edc\u00edch od vracej\u00edc\u00edch se z\u00e1kazn\u00edk\u016f a z\u00e1kazn\u00edk\u016f, kte\u0159\u00ed p\u0159i\u0161li na doporu\u010den\u00ed. Na rozd\u00edl od NPS jde o tvrdou \u00fa\u010detn\u00ed metriku, kter\u00e1 se ned\u00e1 snadno „vyhr\u00e1t“ soci\u00e1ln\u00edm in\u017een\u00fdrstv\u00edm.<\/p>\n

Situace s CSAT a CES je analogick\u00e1. CSAT, typicky m\u011b\u0159en\u00fd jako pr\u016fm\u011br hodnocen\u00ed na \u0161k\u00e1le 1\u20135 nebo 1\u201310 bezprost\u0159edn\u011b po interakci, je vysoce citliv\u00fd na na\u010dasov\u00e1n\u00ed. Pr\u016fzkum odeslan\u00fd ihned po vy\u0159e\u0161en\u00ed probl\u00e9mu generuje jin\u00e9 sk\u00f3re ne\u017e pr\u016fzkum odeslan\u00fd o t\u00fdden pozd\u011bji. T\u00fdmy, jejich\u017e bonusy z\u00e1vis\u00ed na CSAT, toto v\u011bd\u00ed a pr\u016fzkumy na\u010dasov\u00e1vaj\u00ed podle toho.<\/p>\n

CES, navr\u017een\u00fd t\u00fdmem CEB (nyn\u00ed Gartner) v roce 2010 jako prediktor v\u011brnosti z\u00e1kazn\u00edk\u016f skrze n\u00edzk\u00e9 t\u0159en\u00ed p\u0159i interakci, je mo\u017en\u00e1 nejodoln\u011bj\u0161\u00ed ze t\u0159\u00ed metrik, ale ani on nen\u00ed imunn\u00ed. Agenti se mohou nau\u010dit kl\u00e1st pr\u016fzkumn\u00e9 ot\u00e1zky zp\u016fsobem, kter\u00fd sugeruje snadnost („Douf\u00e1m, \u017ee jsme v\u00e1m s t\u00edm pomohli bez zbyte\u010dn\u00e9 n\u00e1mahy?“<\/em>), nebo aktivn\u011b pou\u010dovat z\u00e1kazn\u00edka, jak odpov\u011bd\u011bt.<\/p>\n

Pro\u010d se to d\u011bje: psychologie a syst\u00e9my<\/h2>\n

Goodhart\u016fv z\u00e1kon nen\u00ed v\u00fdsledkem \u0161patn\u00fdch lid\u00ed. Je v\u00fdsledkem racion\u00e1ln\u00edch lid\u00ed reaguj\u00edc\u00edch na \u0161patn\u011b nastaven\u00e9 syst\u00e9my.<\/p>\n

Psychologov\u00e9 mluv\u00ed v t\u00e9to souvislosti o fenom\u00e9nu surrogation stavu, kdy lid\u00e9 za\u010dnou zam\u011b\u0148ovat c\u00edl (lep\u0161\u00ed z\u00e1kaznick\u00e1 zku\u0161enost) s jeho proxy (vy\u0161\u0161\u00ed sk\u00f3re). Studie Michaela Harrise a Billa Taylera v Harvard Business Review (Don’t Let Metrics Undermine Your Business, 2019) uk\u00e1zala, \u017ee surrogation nast\u00e1v\u00e1 prakticky automaticky ve chv\u00edli, kdy jsou metriky propojeny s odm\u011b\u0148ov\u00e1n\u00edm, a to i u lid\u00ed, kte\u0159\u00ed p\u016fvodn\u011b sv\u016fj c\u00edl ch\u00e1pali spr\u00e1vn\u011b.<\/p>\n

V\u00fdsledek je dob\u0159e popsan\u00fd: organizace optimalizuje \u010d\u00edslo m\u00edsto reality<\/strong>. \u010c\u00edslo roste. Realita stagnuje nebo se zhor\u0161uje. Veden\u00ed je spokojen\u00e9. Z\u00e1kazn\u00edci ne.<\/p>\n

Tento stav je nav\u00edc sebeposiluj\u00edc\u00ed. Jakmile jsou metriky „vyhr\u00e1v\u00e1ny“<\/em>, p\u0159est\u00e1vaj\u00ed m\u00edt diagnostickou hodnotu. Mana\u017ee\u0159i nemaj\u00ed dobr\u00e1 data pro rozhodov\u00e1n\u00ed. D\u011blaj\u00ed tedy rozhodnut\u00ed na z\u00e1klad\u011b \u010d\u00edsel, kter\u00e1 neodr\u00e1\u017e\u00ed realitu a cyklus pokra\u010duje.<\/p>\n

Co s t\u00edm: od m\u011b\u0159en\u00ed k u\u010den\u00ed<\/h2>\n

Skute\u010dn\u00e1 ot\u00e1zka proto nezn\u00ed: „Jak chr\u00e1n\u00edme na\u0161e metriky p\u0159ed zneu\u017eit\u00edm?“<\/em> Ale: „K \u010demu metriky vlastn\u011b slou\u017e\u00ed?“<\/em><\/p>\n

Reichheld i Bain & Company konzistentn\u011b argumentuj\u00ed, \u017ee NPS a plat\u00ed to i pro CSAT i CES byly navr\u017eeny jako n\u00e1stroj u\u010den\u00ed, ne jako n\u00e1stroj odm\u011b\u0148ov\u00e1n\u00ed<\/strong>. Je-li v\u00fdsledek pr\u016fzkumu vstupem do t\u00fdmov\u00e9 diskuze o tom, co z\u00e1kazn\u00edk\u016fm vad\u00ed a co funguje, metrika si zachov\u00e1v\u00e1 hodnotu. Je-li v\u00fdsledkem jednozna\u010dn\u00e9ho v\u00fdpo\u010dtu bonusu, ztr\u00e1c\u00ed ji.<\/p>\n

Z pohledu designu syst\u00e9m\u016f odm\u011b\u0148ov\u00e1n\u00ed z toho plyne n\u011bkolik praktick\u00fdch z\u00e1v\u011br\u016f.<\/p>\n

Metriky v\u00e1zan\u00e9 na bonusy by m\u011bly b\u00fdt v\u00fdstupov\u00e9, ne pr\u016fzkumov\u00e9. M\u00edsto NPS jako KPI (Key Performance Indicator, kl\u00ed\u010dov\u00fd ukazatel v\u00fdkonnosti) propojen\u00fd s pr\u00e9mi\u00ed d\u00e1v\u00e1 v\u011bt\u0161\u00ed smysl sledovat m\u00edru retence, m\u00edru opakovan\u00fdch n\u00e1kup\u016f nebo pr\u00e1v\u011b Earned Growth Rate tedy chov\u00e1n\u00ed z\u00e1kazn\u00edk\u016f, nikoli jejich deklarace.<\/p>\n

Ka\u017ed\u00e1 metrika pot\u0159ebuje sv\u016fj „st\u00edn“<\/em>\u00a0protimetriku, kter\u00e1 odhal\u00ed p\u0159esun probl\u00e9mu jinam. Pokud sni\u017euji pr\u016fm\u011brnou dobu zpracov\u00e1n\u00ed hovoru (AHT, Average Handle Time), mus\u00edm sledovat m\u00edru opakovan\u00fdch kontakt\u016f. Pokud optimalizuji CSAT bezprost\u0159edn\u011b po interakci, mus\u00edm sledovat NPS s \u010dasov\u00fdm odstupem. Bez st\u00ednov\u00e9 metriky je Goodhart\u016fv z\u00e1kon nevyhnuteln\u00fd.<\/p>\n

Pr\u016fzkumov\u00e1 data by m\u011bla b\u00fdt odd\u011blena od mana\u017eersk\u00e9 odpov\u011bdnosti za sk\u00f3re. Zam\u011bstnanec by m\u011bl b\u00fdt hodnocen za to, zda z\u00e1kaznick\u00e9 podn\u011bty z pr\u016fzkum\u016f vedly k akci ne za to, jak\u00e9 \u010d\u00edslo pr\u016fzkum vygeneroval. To je z\u00e1sadn\u00ed struktur\u00e1ln\u00ed rozd\u00edl.<\/p>\n

Z\u00e1v\u011brem: metriky p\u0159est\u00e1vaj\u00ed b\u00fdt dobr\u00e9 ve chv\u00edli, kdy se st\u00e1vaj\u00ed motivy. To nen\u00ed argument proti m\u011b\u0159en\u00ed, je to argument pro v\u011bt\u0161\u00ed sofistikovanost v tom, pro\u010d a jak m\u011b\u0159\u00edme. Goodhart\u016fv z\u00e1kon nelze obej\u00edt p\u0159id\u00e1v\u00e1n\u00edm dal\u0161\u00edch metrik. Lze ho zm\u00edrnit t\u00edm, \u017ee rozli\u0161ujeme metriky pro u\u010den\u00ed od metrik pro odpov\u011bdnost a nepou\u017e\u00edv\u00e1me jedny tam, kde pat\u0159\u00ed druh\u00e9.<\/p>\n","protected":false},"excerpt":{"rendered":"

Existuje paradox, kter\u00fd tr\u00e1p\u00ed z\u00e1kaznickou zku\u0161enost v\u00edc ne\u017e \u0161patn\u00e1 data […]<\/p>\n","protected":false},"author":4,"featured_media":7087,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[52,72],"tags":[],"class_list":["post-7086","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-cx-metriky","category-vsechny-clanky"],"_links":{"self":[{"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/posts\/7086","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/comments?post=7086"}],"version-history":[{"count":2,"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/posts\/7086\/revisions"}],"predecessor-version":[{"id":7089,"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/posts\/7086\/revisions\/7089"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/media\/7087"}],"wp:attachment":[{"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/media?parent=7086"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/categories?post=7086"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.insightsofa.com\/cs\/wp-json\/wp\/v2\/tags?post=7086"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}