Segmentacija kupcev

Kako ustvariti segment kupcev z veliko verjetnostjo odhoda

Posted on July 19, 2026 by

Segment kupcev z veliko verjetnostjo odhoda je seznam za zgodnje opozarjanje: kupci, ki se tiho ohlajajo, a še niso povsem odšli. Zgradiš ga iz upadajočih signalov — odpiranja padajo v primerjavi z njihovim preteklim vedenjem, ponovno naročilo zamuja glede na njihov običajni cikel, v zadnjih nekaj pošiljkah ni klikov, obiski strani so se ustavili. Bistvo je, da jih ujameš, dokler je odnos še mogoče rešiti, ne šele potem, ko so pol leta molčali in se e-mail za ponovno pridobivanje zdi kot tapkanje tujca po rami. Zaznaj oddaljevanje zgodaj, pošlji relevantno sporočilo za ponovno vključitev in obdržiš kupce, za katerih ponovno pridobitev bi sicer plačal.

Signal, ki ga bereš, ni “odšli so.” Je “odhajajo.” Velika razlika.

Težava: ugotoviš prepozno

Večina trgovin opazi osip v vzvratnem ogledalu. Zaženeš poročilo, vidiš kup kupcev, ki že leto niso naročili, in pomisliš: hm, kdaj se je to zgodilo? Zgodilo se je počasi, v mesecih upadajoče vključenosti, ki je nihče ni spremljal.

Do takrat je popravek drag. Kupec, ki ga ni že leto dni, je šel naprej — morda h konkurentu, morda le izven navade tvoje znamke. Zdaj hladno pišeš nekomu, ki te je nekoč imel rad, in upaš, da bo popust nekaj prebudil. Včasih deluje. Precej manj pogosto kot lovljenje iste osebe, medtem ko se je zgolj oddaljevala.

Različica te zgodbe za lastnika trgovine: prihodek je videti v redu, ker novi kupci nenehno nadomeščajo tiste, ki iztekajo zadaj. Priliv skrije puščanje. Dokler stroški pridobivanja ne narastejo, priliv ne upočasni in naenkrat je osip, ki si ga prezrl, celotna težava.

Zakaj “zaženi kampanjo za ponovno pridobivanje” ni celoten odgovor

Tokovi za ponovno pridobivanje so dobri. Uporabljam jih. A kampanja za ponovno pridobivanje se sproži pri ljudeh, ki so že odpadli — to je po definiciji pozno posredovanje. Poskušaš znova zagnati hladen motor.

Cenejši potez je, da motorja nikoli ne pustiš povsem ohladiti. Če ukrepaš šele, ko nekdo prečka mejo “neaktiven X dni,” si preskočil celotno okno, ko je še odpiral tvoje e-maile, bil še napol zainteresiran, še dosegljiv z nečim lahkotnim. Prijazno, relevantno sporočilo kupcu, ki se oddaljuje, te stane skoraj nič in ne nosi obupa “prosim, vrni se,” ki ga ima pravo ponovno pridobivanje.

Tudi več popustov tega ne popravi. Če daš pavšalni popust celotni bazi, da bi podprl vključenost, s tem le naučiš vse, naj čakajo na naslednjo razprodajo, in tiho podariš maržo ljudem, ki so tako ali tako ostajali. Segment po občutljivosti na popuste je boljše orodje za odločanje o tem, kdo dejansko potrebuje cenovni sunek.

Kje prihodek odteka

Zadržati kupca, ki ga že imaš, je dramatično ceneje kot pridobiti novega — strošek pridobitve si že plačal, ponovni kupci pa sčasoma običajno porabijo več. Vsak kupec, ki iz “aktivnega” zdrsne v “odšlega”, ne da bi opazil, je drug strošek pridobivanja, ki ga boš na koncu moral plačati, da ga nadomestiš.

Groba ilustracija. Recimo, da je v zadnjem letu pri tebi naročilo 1.000 kupcev in 15 % (ilustrativno) jih letno zdrsne v neaktivnost brez posredovanja. To je 150 kupcev. Če pravočasen opomnik za zgodnje opozarjanje reši vsaj tretjino, si obdržal 50 kupcev, ki bi jih sicer ponovno kupil prek oglasov. Ob življenjski vrednosti, recimo, 180 € (ilustrativno) na kupca je to 9.000 € zadržane vrednosti od tega, da si oddaljevanje ujel zgodaj namesto pozno.

Sklop rast življenjske vrednosti kupca razloži, zakaj se zadržani kupci obrestujejo na načine, kot se novi ne.

Kako zgraditi segment

Zvijača pri signalih osipa je, da so relativni. Ni univerzalnega “ni naročil 60 dni = odhaja,” ker je 60 dni normalno za naročnika na kavo in zaskrbljujoče za nekoga, ki naroča vsak teden. Vsakega kupca meriš glede na njegovo lastno preteklo vedenje.

Signali, ki jih kombiniraš:

Razmik do ponovnega naročila se razteza čez njihov običajni cikel. Če je nekdo zanesljivo ponovno naročil vsakih 30 dni in je zdaj 50. dan, je to oddaljevanje. To je najkoristnejši posamezni signal za porabne in obnovljive izdelke.
Vključenost v e-maile pada glede na njihovo lastno izhodišče. Nekdo, ki je nekoč odpiral večino tvojih pošiljk in zdaj ne odpre nobene, je jasnejše opozorilo kot nekdo, ki že od začetka ni odpiral ničesar.
Ni obiskov strani v obdobju, ki je zanje neobičajno.
Kliki padajo na nič, tudi če se odpiranja nadaljujejo — zanimanje se tanjša.
Upadajoča pogostost naročil čez zadnjih nekaj nakupov.

Kombiniraj dva ali tri. Eno zamujeno ponovno naročilo je morda le založena shramba. Zamujeno ponovno naročilo plus trije neodprti e-maili plus nič obiskov strani so vzorec.

Dva gradnika sta podrobneje obravnavana v segmentiranje po času od zadnjega nakupa in segmentiranje po pogostosti nakupov — segment osipa sta v resnici prav ta dva, brana glede na normalo vsakega kupca in navzkrižno preverjena z ravnjo vključenosti.

Kaj avtomatizirati

Sprožilec: stik vstopi v segment z veliko verjetnostjo odhoda — upadajoči signali prečkajo tvoj prag.

Segment: oddaljujoč se, a ne odšel. Izključi vsakogar, ki je že povsem odpadel (ti spadajo v pravi tok za ponovno pridobivanje), in vsakogar, ki je pravkar naročil.

Čas: takoj ko je oddaljevanje zaznano, ne po fiksnem koledarju. Če je njihov cikel ponovnega naročila 30 dni, naj se sporočilo sproži, ko dosežejo, recimo, 40. dan — ne, ko slučajno gre ven mesečna kampanja.

Kanal: e-mail za večino. SMS za kupce z višjo vrednostjo, kjer njihova izguba resnično zaboli.

Vsebina: prvo sporočilo ni “pogrešamo te” in ni popust. Začni z nečim koristnim ali relevantnim — opomnikom za obnovitev zaloge, če gre za porabno stvar, “tukaj je novo v kategoriji, ki jo kupuješ,” če ne gre, resničnim vprašanjem, kako gre. Opominjaš jih, da obstajaš in da si relevanten. Kakršno koli spodbudo zadrži za drugo sporočilo, in le če se še vedno ne odzovejo. Če odpreš s popustom, svoje najboljše kupce naučiš, da oddaljevanje prinese ponudbo.

Cilj: ponovno jih vključi, preden so izgubljeni — klik, obisk, idealno naročilo — brez zapravljanja marže na ljudeh, ki bi tako ali tako ostali.

To je natančno ogledalo segmenta z veliko verjetnostjo nakupa, ki bere naraščajočo namero in potiska sunek k nakupu. Enaka vedenjska disciplina, nasprotna smer. Eden ujame ljudi, ki se nagibajo noter. Ta ujame ljudi, ki se tiho nagibajo ven.

Kako to izmeriti

Stopnja rešitve: od kupcev, ki so vstopili v segment, koliko jih je oddalo naročilo ali se ponovno vključilo v določenem oknu.
Prihodek od reaktivacije, pripisan toku.
Koliko jih ni nikoli potrebovalo celotnega ponovnega pridobivanja — to je sistem zgodnjega opozarjanja, ki poceni opravlja svoje delo.
Lažni pozitivi: kupci, označeni kot oddaljujoči se, ki vseeno naročajo normalno. Nekaj je v redu; poplava pomeni, da so tvoji pragovi preveč nervozni.
Odvisnost od popustov: naj ostane nizka. Če večina rešitev potrebuje kodo, posreduješ prepozno.

Kako se v to vklaplja Omnisend

To zmore vsaka zmogljiva platforma. V svojih trgovinah uporabljam Omnisend, zato bom govoril o njem.

Pred izbiro sem ga primerjal s Klaviyom — Omnisend je bilo lažje voditi brez najema specialista, cenovno je bil ugodnejši pri mojem obsegu in združeval je e-mail, SMS in push, tako da sem sporočila lahko prihranil za kupce, ki so jih dejansko vredni. Za segment osipa konkretno so pomembni deli segmentacija, ki se lahko sklicuje na zgodovino nakupov in vključenost vsakega kupca skozi čas, ter avtomatizacije, ki se sprožijo ob vstopu v segment, tako da sunek pade točno takrat, ko se pojavi oddaljevanje, in ne po fiksnem urniku.

Razkritje: povezave do Omnisenda na Shopimationu so partnerske — prijavi se prek njih in Shopimation lahko prejme provizijo brez stroškov zate. Poštena omejitev: napovedovati osip iz redke zgodovine je ugibanje. Če je kupec naročil samo enkrat, ni “normalnega cikla,” ob katerem bi merili oddaljevanje, tako da ta segment deluje veliko bolje za ponovne kupce kot za enkratne. Za slednje je pretvorba drugega nakupa zgodnejša, pomembnejša naloga.

Tvoj naslednji korak

Izberi svoj izdelek z največ ponovnimi nakupi in ugotovi njegov običajni razmik do ponovnega naročila. Zgradi en segment: kupci, ki so ta razmik presegli z jasno rezervo in nehali odpirati tvoje e-maile. Ta teden jim pošlji eno resnično koristno sporočilo brez popusta. Spremljaj, koliko se jih vrne. Ta en tok ti bo povedal, ali se splača zgodnje zaznavanje osipa razširiti čez celoten katalog.

Customer Segmentation

How to Create a Likely-to-Churn Segment

Posted on July 17, 2026July 17, 2026 by

A likely-to-churn segment is an early-warning list: customers who are quietly cooling off but haven’t fully left yet. You build it from declining signals — opens dropping compared to their own past behavior, a reorder that’s running late against their normal cycle, no clicks in the last several sends, site visits that stopped. The point is to catch them while there’s still a relationship to save, not after they’ve been silent for six months and a win-back email feels like a stranger tapping their shoulder. Spot the drift early, send a relevant re-engagement message, and you keep customers you’d otherwise pay to reacquire.

The signal you’re reading isn’t “they left.” It’s “they’re leaving.” Big difference.

The problem: you find out too late

Most stores notice churn in the rear-view mirror. You run a report, see a chunk of customers who haven’t ordered in a year, and think, huh, when did that happen? It happened slowly, over months of fading engagement that nobody was watching.

By then the fix is expensive. A customer who’s been gone a year has moved on — maybe to a competitor, maybe just out of the habit of your brand. You’re now cold-emailing someone who used to like you, hoping a discount reawakens something. It sometimes works. It works far less often than catching the same person while they were merely drifting.

The store owner’s version of this: revenue looks fine because new customers keep replacing the ones bleeding out the back. The leak is hidden by the inflow. Until acquisition costs rise, the inflow slows, and suddenly the churn you were ignoring is the whole problem.

Why “run a win-back campaign” isn’t the whole answer

Win-back flows are good. I use them. But a win-back campaign fires on people who are already lapsed — that’s a late intervention by definition. You’re trying to restart a cold engine.

The cheaper move is to never let the engine go fully cold. If you only act once someone crosses “inactive for X days,” you’ve skipped the entire window where they were still opening your emails, still half-interested, still reachable with something light. A friendly, relevant message to a drifting customer costs you almost nothing and doesn’t carry the “please come back” desperation of a true win-back.

More discounts don’t fix this either. Blanket-discounting your whole base to prop up engagement just trains everyone to wait for the next sale and quietly hands margin to people who were staying anyway. The discount-sensitivity segment is a better tool for deciding who actually needs a price nudge.

Where the revenue leaks

Retaining a customer you already have is dramatically cheaper than winning a new one — you’ve already paid the acquisition cost, and repeat buyers tend to spend more over time. Every customer who slips from “active” to “gone” without you noticing is a second acquisition cost you’ll eventually have to pay to replace them.

Rough illustration. Say 1,000 customers ordered from you in the last year, and 15% (Illustrative) drift into inactivity annually without intervention. That’s 150 customers. If a timely early-warning nudge saves even a third of them, you’ve kept 50 buyers you’d otherwise have re-bought through ads. At a lifetime value of, say, €180 (Illustrative) each, that’s €9,000 in retained value from catching drift early instead of late.

The customer lifetime value growth cluster gets into why retained customers compound in ways new ones don’t.

How to build the segment

The trick with churn signals is that they’re relative. There’s no universal “hasn’t ordered in 60 days = churning,” because 60 days is normal for a coffee subscriber and alarming for someone who reorders weekly. You measure each customer against their own past behavior.

Signals to combine:

Reorder gap stretching past their normal cycle. If someone reliably reordered every 30 days and it’s now day 50, that’s drift. This is the single most useful signal for consumable and replenishable products.
Email engagement falling versus their own baseline. Someone who used to open most of your sends and now opens none is a clearer warning than someone who never opened anything to begin with.
No site visits in a stretch that’s unusual for them.
Clicks going to zero even if opens continue — interest is thinning.
Declining order frequency over the last few purchases.

Combine two or three. A single missed reorder might just be a stocked pantry. A missed reorder plus three unopened emails plus no site visits is a pattern.

The two building blocks here are covered in more depth in segmenting by time since last purchase and segmenting by purchase frequency — the churn segment is really those two, read against each customer’s own normal, and cross-checked with engagement level.

What to automate

Trigger: a contact enters the likely-to-churn segment — declining signals cross your threshold.

Segment: drifting-but-not-gone. Exclude anyone who’s already fully lapsed (they belong in a proper win-back flow) and anyone who just ordered.

Timing: the moment the drift is detected, not on a fixed calendar. If their reorder cycle is 30 days, the message should fire when they hit, say, day 40 — not when a monthly campaign happens to go out.

Channel: email for most. SMS for higher-value customers where losing them genuinely stings.

Content: message one is not “we miss you” and not a discount. Lead with something useful or relevant — a replenishment reminder if it’s a consumable, a “here’s what’s new in the category you buy” if it’s not, a genuine check-in. You’re reminding them you exist and you’re relevant. Hold any incentive for a second message, and only if they still don’t respond. Opening with a discount teaches your best customers that drifting earns them a deal.

Goal: re-engage before they’re lost — a click, a visit, ideally an order — without burning margin on people who’d have stayed anyway.

This is the exact mirror of a likely-to-buy segment, which reads rising intent and pushes a purchase nudge. Same behavioral discipline, opposite direction. One catches people leaning in. This one catches people quietly leaning out.

How to measure it

Save rate: of customers who entered the segment, how many placed an order or re-engaged within a set window.
Reactivation revenue attributed to the flow.
How many never needed the full win-back — that’s the early-warning system doing its job cheaply.
False positives: customers flagged as drifting who order normally anyway. Some is fine; a flood means your thresholds are too jumpy.
Discount reliance: keep it low. If most saves need a code, you’re intervening too late.

How Omnisend fits

Any capable platform can do this. I use Omnisend in my own stores, so that’s what I’ll speak to.

I compared it with Klaviyo before choosing — Omnisend was easier to run without hiring a specialist, priced better at my volume, and combined email, SMS, and push so I could reserve texts for the customers actually worth them. For a churn segment specifically, the parts that matter are segmentation that can reference each customer’s purchase history and engagement over time, and automations that trigger on segment entry so the nudge fires exactly when drift appears rather than on a fixed schedule.

Disclosure: Omnisend links on Shopimation are affiliate links — sign up through them and Shopimation may earn a commission at no cost to you. The honest limitation: predicting churn from a thin history is guesswork. If a customer has only ordered once, there’s no “normal cycle” to measure drift against, so this segment works far better for repeat buyers than for one-timers. For those, converting the second purchase is the earlier, more important job.

Your next step

Pick your most repeat-heavy product and figure out its normal reorder gap. Build one segment: customers who’ve blown past that gap by a clear margin and stopped opening your emails. Send them one genuinely useful, non-discount message this week. Track how many come back. That single flow will tell you whether early-warning churn detection is worth expanding across your catalog.