A/B-testiranje in optimizacija

Kako testirati preproste tekstovne emaile v primerjavi z oblikovanimi

Posted on July 18, 2026 by

Testiranje preprostih tekstovnih emailov proti oblikovanim je test formata, ne običajen A/B-test — različici se hkrati razlikujeta v ducatih stvari, zato boste izvedeli, kateri format zmaga za to vrsto emaila, ne pa zakaj. To je v redu, dokler s tem ravnate v skladu s tem: test izvedite ločeno za vsako vrsto emaila (sporočilo za ponovno pridobitev in predstavitev izdelka sta različni nalogi), presojajte po naročilih in odgovorih, ne po odprtjih, ter vedite, da je možnost “preprosto tekstovno” v vaši emailni platformi v resnici minimalno oblikovan HTML, kar je ravno to, kar želite — pravi tekstovni email izgubi sledenje odprtjem in ne more napolniti dinamičnih blokov izdelkov. Pričakujte, da preprosto zmaga tam, kjer naj email deluje osebno, in oblikovanje tam, kjer je treba izdelek videti. Ta članek vam pokaže, kako ugotoviti, kateri od vaših emailov je kateri.

Pravo vprašanje za razpravo o formatu

Videli ste oba tabora. Eden pravi, da oblikovani emaili izgledajo profesionalno in prodajajo izdelke vizualno. Drugi pravi, da se preprosto sporočilce od ustanovitelja zdi človeško, pogosteje pristane v primarnem nabiralniku in pretvarja bolje kot bleščeča predloga.

Oba tabora imata prav glede nekaterih emailov, zato se razprava nikoli ne razreši. Objava bliskovite razprodaje za modno trgovino ima drugačno nalogo kot sporočilo “je bilo z vašim naročilom vse v redu?”. Prvo potrebuje fotografijo; nihče ne kupi obleke iz odstavka. Drugo z blagovno glavo in tabelo izdelkov postane slabše, ker oblikovanje signalizira “množična promocija” in ubije edino, kar je sporočilo imelo — občutek, da ga je napisal človek.

Torej uporabno vprašanje nikoli ni bilo “preprosto ali oblikovano?”. Je “kateri od mojih emailov prodajajo vizualno in kateri govorijo enemu človeku?”. Na to odgovarja test.

Zakaj to ni čist test z eno spremenljivko

Običajna disciplina testiranja pravi, naj spremenite eno stvar naenkrat, in ta disciplina obstaja z dobrim razlogom — kako testirati po eno spremenljivko naenkrat jo obravnava. Test preprosto proti oblikovano to pravilo namenoma prekrši. Med različicama spreminjate postavitev, slike, barve, dolžino besedila, verjetno ton in kako se email prikaže v predogledu nabiralnika. Vse naenkrat.

Pošteno sprejmite kompromis. Ko zmaga preprosto, ne boste vedeli, ali je zmagalo zaradi intimnega tona, hitrejšega nalaganja, umestitve v primarni zavihek ali odsotnosti motečega grida izdelkov. Vedeli boste le, da je format zmagal, za to vrsto emaila, za vaš seznam. To je še vedno dragocen, uporaben rezultat — le da ni diagnoza. Če želite vedeti, kateri element vašega oblikovanja pomaga ali škodi, je to drugačen program manjših testov: A/B-testiranje oblikovanja emailov brez zmede v rezultatih je vodnik za to. Jasno začrtajte mejo: najprej test formata za izbiro smeri, nato testi elementov za optimizacijo znotraj nje.

Ena sosednja past: ne dovolite, da bi preprosta različica hkrati postala dramatično krajša ali daljša od oblikovane, sicer ste nehote izvedli še test dolžine. Dolžina je svoj poskus — glejte kako testirati kratke in dolge emaile spletne trgovine.

Tehnična podrobnost, ki jo večina vodnikov preskoči

Ko vaša emailna platforma ponudi “videz preprostega besedila”, ne pošilja dejanskega preprostega besedila. Pošilja HTML, oblikovan tako, da izgleda kot osebno sporočilce — sistemska pisava, brez slik, brez stolpcev. Dobro. To uporabite.

Pristno preprost tekstovni email (surova tekstovna oblika MIME) sploh nima HTML-ja, kar pomeni brez sledilnega piksla — torej se odprtja ne zabeležijo — in brez oblikovanih gumbov ali dinamične vsebine. Vaše poročilo o testu bi pokazalo preprosto različico s katastrofalno stopnjo odpiranja, ki je v resnici merilni artefakt, ne vedenje bralca. Sledenje klikom običajno preživi, ker platforme povezave ovijejo tako ali tako, a polovica vaše nadzorne plošče potemni. Ostanite pri HTML-ju, ki izgleda preprosto, in primerjava ostane poštena.

Druga praktična posledica: preprosto izgledajoč email ne more nositi vašega dinamičnega bloka priporočil ali kartic izdelkov. Če se email iz toka opira na samodejne izbore izdelkov, mora preprosta različica te nadomestiti s pisano vrstico (“keramični počasni prelivnik, ki ste ga gledali, je spet na zalogi”) in tekstovno povezavo. Včasih je prav ta zamenjava razlog, zakaj preprosto zmaga. Včasih je razlog, zakaj izgubi.

Kje si vsak format ponavadi prisluži svoje

Preden testirate na slepo, prednapolnite kolone s smiselno domnevo za vsako vrsto emaila:

Verjetno območje za preprosto: sporočila za ponovno pridobitev, ponaročilne poizvedbe, prošnje za oceno, opomniki za dopolnitev zaloge, dobrodošlica z glasom ustanovitelja. Ti delujejo, ko se berejo, kot da en človek opazi drugega.
Verjetno območje za oblikovanje: predstavitve novih kolekcij, sezonske promocije, kurirane zbirke najboljših prodajnih artiklov, karkoli, kjer je videz izdelka argument.
Resnično negotovo, najprej testirajte: emaili opuščene košarice (slika izdelka opomni; osebno sporočilce razoroži) in obvestila o vrnitvi na zalogo.

Te obravnavajte kot izhodiščne ugibe, ne sklepe. V mojih trgovinah tok za ponovno pridobitev teče preprosto izgledajoče, koledar kampanj pa oblikovano — a ta delitev je nastala iz testiranja, in trgovina, ki oblikovno občutljivemu občinstvu prodaja vizualne izdelke, bi lahko pristala drugje.

Kako pravilno izvesti test

Izberite en email, ne celotnega programa. Začnite tam, kjer je domneva o osebnem sporočilu najmočnejša in obstaja obseg — prvi email opuščene košarice ali vaš uvod za ponovno pridobitev sta pogosta izbira.
Preprosto različico napišite na novo. Ne slecite slik z oblikovanega emaila in mislite, da je opravljeno; oskubljena predloga se bere kot pokvarjen email. Napišite jo kot sporočilce: pozdrav, dva ali trije kratki odstavki, ena tekstovna povezava ali en tih gumb, človeški podpis.
Ohranite ponudbo, čas pošiljanja in cilj enake. Isti popust (ali brez), ista pristajalna stran, isti sprožilec in zamik. Format je edina namerna razlika.
Razdelite 50/50 in počakajte na naročila. Odprtja so tukaj neuporabna — formata različno vplivata na dostavljivost in besedilo predogleda, zato so razlike v stopnji odpiranja deloma stvar umestitve v nabiralnik, ne prepričevanja. Prihodek na prejemnika je pošten semafor, s stopnjo klikov kot podpornim signalom.
Opazujte tudi odgovore in odjave. Preprosti emaili pogosto pritegnejo prave odgovore, kar se oblikovanim skoraj nikoli ne zgodi. Odgovori nekaj veljajo — hranijo dostavljivost in občasno rešijo prodajo — zato jih štejte.
Dajte mu pravi obseg. V toku prejemniki kapljajo dnevno, zato berljiv vzorec traja tedne. Za grob občutek za številke: pri 30 sprožitvah na dan vam mesec da približno 450 prejemnikov na krak, pri 3-odstotni stopnji naročil pa je to približno 13 naročil proti 13 (ilustrativno) — še vedno tanko. Ali vaš seznam sploh zmore ta test, se splača preveriti vnaprej: kako velik mora biti emailni seznam za A/B-testiranje.

Primer trgovine (ilustrativno)

Email za ponovno pridobitev pri trgovini s prehranskimi dopolnili je bil oblikovana predloga: glava z logotipom, tri kartice izdelkov, 10-odstotna koda, oranžen gumb. Testirali so ga proti preprosto izgledajočemu sporočilu od ustanovitelja — “Že nekaj mesecev je od vašega zadnjega naročila magnezijevega kompleksa. Če vam ni ustrezal, bi to iskreno rad izvedel. Če ste le pozabili, tukaj je vaša povezava.” Ista 10-odstotna koda kot tekstovna povezava.

V šestih tednih (ilustrativne številke): oblikovana različica je dobila več klikov, 4,1 % proti 3,2 %. Preprosta različica je dobila več naročil — 2,4 % proti 1,7 % — plus enajst pravih odgovorov, dva od katerih sta razkrila pritožbo glede dostave, za katero trgovina ni vedela. Kliki in naročila, ki kažejo v nasprotni smeri, so pri testih formata pogosti in zato odloči metrika naročil. Kaj storiti, ko se vaše metrike ne strinjajo, je svoja tema: izbira zmagovalnega emailnega testa.

Nastavitev tega v Omnisendu

Tovrstni test izvajam v Omnisendu in mehanika je na dober način neizumetničena. Za tokove korak razdelitve A/B pošlje polovico stikov po vsaki veji; oblikovani email zgradite v urejevalniku povleci-in-spusti, preprosto različico pa kot na besedilo osredotočeno postavitev — minimalno oblikovanje, brez slik — in veji primerjajte po prihodku v poročilu toka. Za kampanje razdelitev opravi vgrajeni A/B-test, čeprav bi morali prihodek na različico prebrati sami, namesto da pusti izbrati zmagovalca po odprtjih. Omejitev, ki jo velja poznati: preprosta različica ne more nositi bloka priporočevalnika izdelkov, zato je vsaka primerjava, ki vključuje dinamična priporočila, po naravi format-plus-vsebina, ne zgolj format. Tega ne popravi nobeno orodje; to je narava testa.

Vaš naslednji korak

Izberite en email v svojem programu, ki mu bo najverjetneje koristilo, da zveni človeško — za večino trgovin je to uvod za ponovno pridobitev ali ponaročilna poizvedba. Preprosto različico napišite ta teden, razdelite jo 50/50 in mesec dni ne glejte rezultatov. Nato presojajte po prihodku na prejemnika in odgovorih ter pustite, da si vsaka vrsta emaila v vašem programu prisluži svoj format, namesto da ga podeduje od predloge.

A/B Testing and Optimization

How to Test Plain-Text vs Designed Emails

Posted on July 17, 2026 by

Testing plain-text against designed emails is a format test, not a normal A/B test — the two variants differ in dozens of ways at once, so you’ll learn which format wins for this type of email, not why. That’s fine, as long as you treat it accordingly: run the test separately for each email type (a win-back note and a product launch are different jobs), judge on orders and replies rather than opens, and know that the “plain-text” option in your email platform is really minimally-styled HTML, which is what you want — a true plain-text email loses open tracking and can’t populate dynamic product blocks. Expect plain to win where the email should feel personal, and design to win where the product needs to be seen. This article shows you how to find out which of your emails is which.

The real question behind the format debate

You’ve seen both camps. One says designed emails look professional and sell products visually. The other says a plain note from the founder feels human, lands in the primary inbox more often, and out-converts the glossy template.

Both camps are right about some emails, which is why the debate never resolves. A flash-sale announcement for a fashion store has a different job than a “was everything okay with your order?” message. The first needs photography; nobody buys a dress from a paragraph. The second gets worse with a branded header and a product grid, because the design signals “mass promotion” and kills the one thing the message had going for it — the sense that a person wrote it.

So the useful question was never “plain or designed?” It’s “which of my emails are selling visually, and which are talking to one person?” That’s what the test answers.

Why this isn’t a clean single-variable test

Normal testing discipline says change one thing at a time, and that discipline exists for good reason — how to test one variable at a time covers it. A plain vs designed test breaks the rule on purpose. Between the two variants you’re changing layout, imagery, colors, copy length, probably tone, and how the email renders in the inbox preview. All at once.

Accept the trade honestly. When plain wins, you won’t know whether it won because of the intimate tone, the faster load, the primary-tab placement, or the absence of a distracting product grid. You’ll only know the format won, for this email type, for your list. That’s still a valuable, actionable result — it just isn’t a diagnosis. If you want to know which element of your design helps or hurts, that’s a different program of smaller tests: A/B testing email design without confusing the results is the guide for that. Draw the line clearly: format test first to pick a lane, element tests afterward to optimize within it.

One adjacent trap: don’t let the plain variant also become dramatically shorter or longer than the designed one, or you’ve accidentally run a length test too. Length is its own experiment — see how to test short vs long ecommerce emails.

The technical detail most guides skip

When your email platform offers a “plain text look,” it’s not sending actual plain text. It’s sending HTML styled to look like a personal note — system font, no images, no columns. Good. Use that.

A genuinely plain-text email (the raw MIME text format) has no HTML at all, which means no tracking pixel — so opens don’t register — and no styled buttons or dynamic content. Your test report would show the plain variant with a catastrophic open rate that’s actually a measurement artifact, not reader behavior. Click tracking usually survives because platforms wrap the links either way, but half your dashboard goes dark. Stick with plain-looking HTML and the comparison stays fair.

The other practical consequence: a plain-looking email can’t carry your dynamic recommendation block or product cards. If a flow email leans on automated product picks, the plain variant has to replace them with a written line (“the ceramic pour-over you looked at is back in stock”) and a text link. Sometimes that swap is exactly why plain wins. Sometimes it’s why it loses.

Where each format tends to earn its keep

Before testing blind, stack the deck with a sensible hypothesis per email type:

Likely plain territory: win-back messages, post-purchase check-ins, review requests, replenishment reminders, the founder-voice welcome email. These work when they read like one person noticing another.
Likely design territory: new-collection launches, seasonal promotions, curated bestseller roundups, anything where the product’s appearance is the argument.
Genuinely uncertain, test first: abandoned-cart emails (the product image reminds; the personal note disarms) and back-in-stock alerts.

Treat these as starting guesses, not conclusions. In my own stores the win-back flow runs plain-looking and the campaign calendar runs designed — but that split came from testing, and a store selling visual products to a design-conscious audience could land elsewhere.

How to run the test properly

Pick one email, not your whole program. Start where the personal-note hypothesis is strongest and volume exists — the first abandoned-cart email or your win-back opener are common choices.
Write the plain variant fresh. Don’t strip images from the designed email and call it done; a gutted template reads like a broken email. Write it as a note: greeting, two or three short paragraphs, one text link or single quiet button, a human sign-off.
Keep the offer, the send time, and the destination identical. Same discount (or none), same landing page, same trigger and delay. The format is the only intended difference.
Split 50/50 and wait for orders. Opens are useless here — the two formats affect deliverability and preview text differently, so open rate differences are partly about inbox placement, not persuasion. Revenue per recipient is the honest scoreboard, with click rate as a supporting signal.
Watch replies and unsubscribes too. Plain emails often pull actual replies, which designed emails almost never do. Replies are worth something — they feed deliverability and occasionally rescue a sale — so count them.
Give it real volume. In a flow, recipients trickle in daily, so a readable sample takes weeks. As a rough feel for the numbers: at 30 triggers a day, a month gives you about 450 recipients per arm, and at a 3% order rate that’s roughly 13 orders against 13 (illustrative) — still thin. Whether your list can support the test at all is worth checking first: how large does an email list need to be for A/B testing.

A store example (illustrative)

A supplement store’s win-back email was a designed template: logo header, three product cards, a 10% code, an orange button. They tested it against a plain-looking note from the founder — “It’s been a few months since your last order of the magnesium complex. If it didn’t work for you, I’d honestly like to know. If you just forgot, here’s your link.” Same 10% code as a text link.

Over six weeks (illustrative figures): the designed version got more clicks, 4.1% against 3.2%. The plain version got more orders — 2.4% against 1.7% — plus eleven actual replies, two of which surfaced a shipping complaint the store didn’t know about. Clicks and orders pointing in opposite directions is common in format tests, and it’s why the order metric decides. What to do when your metrics disagree is its own topic: choosing a winning email test.

Setting this up in Omnisend

I run this kind of test in Omnisend, and the mechanics are unglamorous in a good way. For flows, an A/B split step sends half the contacts down each branch; build the designed email in the drag-and-drop editor and the plain variant as a text-focused layout — minimal styling, no images — and compare the branches on revenue in the flow report. For campaigns, the built-in A/B test handles the split, though you should read revenue per variant yourself rather than letting it pick a winner on opens. The limitation worth knowing: the plain variant can’t carry the product recommender block, so any comparison involving dynamic recommendations is inherently format-plus-content, not format alone. No tool fixes that; it’s the nature of the test.

Your next step

Pick the one email in your program most likely to benefit from sounding human — for most stores that’s the win-back opener or the post-purchase check-in. Write the plain version this week, split it 50/50, and don’t look at the results for a month. Then judge on revenue per recipient and replies, and let each email type in your program earn its own format instead of inheriting one from your template.