Igwe ojii nwere ọgụgụ isi kacha mma: Otu esi eji injin stats gaa nyocha A/B mara nke ọma, yana ngwa ngwa

Injin Stats kacha mma na atụmatụ nnwale A/B

Ọ bụrụ na ị na -achọ ịme mmemme nnwale iji nyere ule azụmahịa gị aka & mụta, enwere ike ị na -eji ya Igwe ojii ọgụgụ isi kacha mma - ma ọ bụ na ọ dịkarịa ala elele ya. Optimizely bụ otu n'ime ngwa kachasị ike n'egwuregwu ahụ, mana dịka ụdị ngwa ọrụ ọ bụla, ị nwere ike iji ya mee ihe na -adịghị mma ma ọ bụrụ na ị ghọtaghị etu o si arụ ọrụ. 

What makes Optimizely so powerful? At the core of its feature set lies the most informed and intuitive statistics engine in a third-party tool, allowing you to focus more on getting important tests live – without needing to worry that you’re misinterpreting your results. 

Dị ka ọmụmụ ọdịnala kpuru ìsì na nkà mmụta ọgwụ, A / B ule ga -egosi enweghị usoro agwọ ọrịa nke saịtị gị na ndị ọrụ dị iche iche wee tụọ ịdị irè ọgwụgwọ ọ bụla. 

Ndekọ ọnụ na -enyere anyị aka ime ntụnyere gbasara etu ọgwụgwọ ahụ nwere ike isi dị ogologo oge. 

Most A/B testing tools rely on one of two types of statistical inference: Frequentist or Bayesian stats. Each school has various pros and cons – Frequentist statistics require a sample size to be fixed in advance of running an experiment, and Bayesian statistics mainly care about making good directional decisions rather than specifying any single figure for impact, to name two examples. Optimizely’s superpower is that it’s the only tool on the market today to take a kacha mma n'ụwa abụọ bia.

Ọgwụgwụ pụta? Optimizely na -enyere ndị ọrụ aka ịme nnwale ngwa ngwa, ntụkwasị obi karịa, yana nghọta.

In order to take full advantage of that, though, it’s important to understand what’s happening behind the scenes. Here are 5 insights and strategies that will get you using Optimizely’s capabilities like a pro.

Usoro #1: Ghọta na ọ bụghị metrik niile ka ahapụrụ

N'ọtụtụ ngwa nnwale, ihe a na -elegharakarị anya bụ na ka ị na -agbakwunye metrik na soro dị ka akụkụ nke ule gị, o yikarịrị ka ị ga -ahụ nkwubi okwu ụfọdụ na -ezighi ezi n'ihi ohere enweghị atụ (na ọnụ ọgụgụ, a na -akpọ nke a “nsogbu ule ọtụtụ. ”). Iji mee ka nsonaazụ ya bụrụ nke a pụrụ ịdabere na ya, Optimizely na -eji usoro njikwa na mgbazi iji mee ka ohere nke ihe ahụ na -eme dị ka o kwere mee. 

Njikwa na ndozi ahụ nwere ihe abụọ ọ pụtara mgbe ị na -aga melite ule n'Ọdịmma. Nke mbụ, metric ị họpụtara dị ka nke gị Metric nke mbụ ga -eru ihe ndekọ ọnụ ọgụgụ ngwa ngwa, ihe ndị ọzọ niile na -agbanwe agbanwe. Nke abụọ, ka metrik na -agbakwunye na nnwale, ogologo oge metrik gị ga -adị ogologo iji nweta ihe ndekọ ọnụ ọgụgụ.

Mgbe ị na -eme atụmatụ nnwale, make sure you know which metric will be your True North in your decision-making process, make that your Primary Metric. Then, keep the rest of your metrics list lean by removing anything that’s too superfluous or tangential.

Usoro #2: Zụlite Àgwà nke Gị

Optimizely is great at giving you several interesting and helpful ways to segment your experiment results. For example, you can examine whether certain treatments perform better on desktop vs. mobile, or observe differences across traffic sources. As your experimentation program matures though, you’ll quickly wish for new segments – these may be specific to your use case, like segments for one-time vs. subscription purchases, or as general as “new vs. returning visitors” (which, frankly, we still can’t figure out why that isn’t provided out of the box).

The good news is that via Optimizely’s Project Javascript field, engineers familiar with Optimizely can build any number of interesting custom attributes that visitors can be assigned to and segmented by. At Cro Metrics, we’ve built a number of stock modules (like “new vs. returning visitors”) that we install for all of our clients via their Project Javascript. Leveraging this ability is a key differentiator between mature teams who have the right technical resources to help them execute, and teams who struggle to realize the full potential of experimentation.

Usoro #3: Explore Optimizely’s Stats Accelerator

One often-overhyped testing tool feature is the ability to use “multi-armed bandits”, a type of machine learning algorithm that dynamically changes where your traffic is allocated over the course of an experiment, to send as many visitors to the “winning” variation as possible. The issue with multi-armed bandits is that their results aren’t reliable indicators of long-term performance, so the use case for these types of experiments are limited to time-sensitive cases like sales promotions.

Otú ọ dị, ọ kacha mma nwere ụdị algọridim dị iche iche dịịrị ndị ọrụ nwere atụmatụ dị elu - Stats Accelerator (nke a maara ugbu a dị ka nhọrọ “Mee ngwangwa mmụta” n'ime ndị ohi). Ntọala a, kama ịnwa ike ikesa oke okporo ụzọ ka ọ bụrụ nke kacha arụ ọrụ, Optimizely dynamically dynamite traffic to the variants to reach statistical important quickestst. N'ụzọ dị otu a, ị nwere ike mụta ngwa ngwa, ma jigide mmegharị nke nsonaazụ nnwale A/B ọdịnala.

Usoro #4: Tinye Emojis na aha Metric gị

Na nlele mbu, echiche a nwere ike ọ gaghị adị mma, ọbụlagodi inane. Agbanyeghị, otu akụkụ dị mkpa n'ịhụ na ị na -agụ nsonaazụ nnwale ziri ezi na -amalite site n'ịhụ na ndị na -ege gị ntị nwere ike ịghọta ajụjụ ahụ. 

Mgbe ụfọdụ n'agbanyeghị mbọ anyị kacha mma, aha metric nwere ike bụrụ ihe mgbagwoju anya (chere - ọ bụ metric ahụ ọkụ mgbe anabatara iwu ahụ, ma ọ bụ mgbe onye ọrụ kụrụ ibe ekele?), Ma ọ bụ nnwale nwere ọtụtụ metrik na -agbadata ma na -agbadata nsonaazụ ya. Peeji na -eduga n'ịba oke oke echiche.

Ịtinye emojis na aha igwe metrik gị (ebumnobi, akara akara akwụkwọ ndụ akwụkwọ ndụ, ọbụlagodi nnukwu akpa ego nwere ike ịrụ ọrụ) nwere ike ibute ibe ndị a na -enyocha. 

Tụkwasa anyị obi - ịgụpụta nsonaazụ ga -adị mfe.

Usoro #5: Cheba echiche maka ọkwa ihe ndekọ ọnụ ọgụgụ gị

Results are deemed conclusive in the context of an Optimizely experiment when they’ve reached ihe ndekọ ọnụ ọgụgụ. Ihe ndekọ ọnụ ọgụgụ bụ okwu mgbakọ na mwepụ siri ike, mana nke bụ na ọ nwere ike bụrụ ihe nlele gị sitere na ezigbo ọdịiche dị n'etiti mmadụ abụọ, ọ bụghịkwa ohere nkịtị. 

Optimizely’s reported statistical significance levels are “always valid” thanks to a mathematical concept called nyocha usoro - nke a na -eme ka ha bụrụ ndị a pụrụ ịtụkwasị obi karịa nke ngwaọrụ nnwale ndị ọzọ, nke nwere ike ibute ụdị nsogbu '' nlele '' ma ọ bụrụ na ị gụọ ha ngwa ngwa.

It’s worth considering what level of statistical significance you deem important to your testing program. While 95% is the convention in the scientific community, we’re testing website changes, not vaccines. Another common choice in the experimental world: 90%.  But are you willing to accept a little more uncertainty in order to run experiments faster and test more ideas? Could you be using 85% or even 80% statistical significance? Being intentional about your risk-reward balance can pay exponential dividends over time, so think this through carefully.

Gụkwuo gbasara igwe ojii nwere ọgụgụ isi kacha mma

Ụkpụrụ ise ndị a ngwa ngwa na nghọta ga -enyere gị aka nke ukwuu iburu n'uche mgbe ị na -eji Optimizely. Dị ka ọ dị na ngwa ọ bụla, ọ na-agbadata iji jide n'aka na ị nwere ezi nghọta maka nhazi nhazi ihe nkiri niile, yabụ ị nwere ike ijide n'aka na ị na-eji ngwa ahụ nke ọma na n'ụzọ dị irè. Site na nghọta ndị a, ị nwere ike nweta nsonaazụ ntụkwasị obi ị na -achọ, mgbe ịchọrọ ha. 

Kedu ihe ị chere?

Ebe a na-eji Akismet iji belata spam. Mụta otu esi etinye usoro data gị.