如何有效分析付費用戶(一)

作者:Josh Bycer

付費用戶是指那些爲你的產品花錢的人。我們必須瞭解他們的行爲的所有細微差別:他們付錢的內容是什麼,支付速度多快,以及付了多少錢。如此我們便能夠更好地摸清楚他們的感受以及他們是否能夠從對你的產品的投資中獲得滿足。實際上,甚至在免費在線遊戲中,玩家所做出的每筆消費都是一種投資:一開始他們會先花錢,而最後他們會獲得一些投資回報(ROI)—-通常是以現金或他們的體驗情感呈現出來。因此你應該將這些用戶當成投資者,即使他們只是中小投資者。

爲了更好地理解這種行爲,我們可以找到有關付費用戶的分銷與需求的特定分析方法與報告。但今天我們將把RFM分析作爲了解你的付費用戶結構的一種基本方法。

RFM表示:

R–時間,即多久前進行了最後一次的購買;

F–頻率,即多長時間進行一次購買;

M–消費金額,即總的購買費用。

你需要提供三個與這些付費用戶參數相關的標記。通常情況下,在理論材料中,用戶評估是按照三點式進行(好,一般,差),但實際上在RFM分析中我們甚至會使用五點式或十點式規模進行用戶評估。爲了更簡單地說明,讓我們着眼於一個三點式系統例子:

R==1,自從用戶上次花錢已經過去很長時間了;

R==2,上次消費距離不是很久;

R==3,用戶最近剛剛消費;

F==1,用戶很少消費;

F==2,用戶會基於一定頻率消費;

F==3,用戶經常消費;

M==1,總的消費數額較小;

M==2,用戶爲項目支付了一定的數額;

M==3,用戶支付了很多錢。

當然了,這時候問題便出現了:在這種情況下該如何理解很久以前/最近,經常/很少以及很多/較少。我們可能會基於兩種方式去回答這一問題:

1.專家評估。沒人知道你的項目比你優秀。因此,你可以自己定義很久以前和最近,以及很多和較少代表什麼。讓我們假設很久以前指的是一個多月以前,很少是指一個月一次或更少,較少是在整個消費期間投入的錢不到100盧布。

2.分位數和四分位數。讓我們回想一下數理統計。根據其中的一個參數分配你的用戶(例如在一段時期內的支付總數),選擇所有用戶中的前5%,並假設這些用戶是花費較多的用戶。恭喜你,你已經擁有5%的用戶樣本分位數。你還可以使用四分位數(遊戲邦注:四分位數==25%的分位數–級別),並將第一四分位數當成較多,最後四分位數當成較少,而它們之間的數值便是平均的付費金額。而即使如此,當你使用分位數和四分位數時,你也不能漏掉主觀評估,因此你需要在這時候再次着眼於第1種方法。

不管怎樣,你需要花些時間在Excel表格(或其它工具)上按照時間,頻率和消費金額去標記每一個付費用戶。

現在是最有趣的部分。

你可以觀察這些標記在你的付費用戶間的分佈,並判斷哪些用戶的數量最多。這能夠幫助你更好地劃分付費用戶並規劃市場營銷行動,以獲取更高利益。

一個簡單的例子:

最近剛購買,但是購買頻率很低(或者只購買了一次)—-全新付費用戶。你該如何面對他們?表達出你的感激之情!你的目標是激勵他們不斷進行購買。就像許多研究表明的那樣,用戶的不斷購買以及購買頻率和數額都能夠提高一款應用賺取百萬美元收益的機率。

最近剛購買,且購買頻率很高—-忠實用戶。他們不需要額外的激勵,但是你也應該想辦法表達對於他們的忠誠的感謝(遊戲邦注:如意外的獎勵,驚喜等等)。

頻繁購買,但上一次購買已經過去很久—-處於離開邊緣的忠實用戶。換句話說,這些用戶的錢正逐漸從你的指縫間流走。所以你的目標便是提醒他們你的存在。也許一封簡單的推送郵件便足以。或者你應該詢問他們發生了什麼改變以及爲什麼他們會離開。

很少購買,且上一次購買已經過去很久—-已經離開的用戶。他們不可能成爲你的忠實用戶,這是在過去出現的某些內容所導致的。你可以提供給他們(不只是他們)一個行動建議(即使這對於你來說不一定是有益的),這可能會激勵他們再次購買並回到產品中。即使不行的話你也能夠明確他們不喜歡什麼內容,並基於反饋去完善你的產品。

讓我們想象以下情況:

1.項目X想要提高收益;

2.他們進行了RFM分析,結果如下:

1)忠實用戶的流失率非常高;

2)許多用戶只進行了一次購買。

3.他們使用了一些觸發內容去判斷用戶何時處於“上一次購買已經過了很久”或者用戶在停止花錢前屬於忠實用戶等狀態。而他們也在這些時刻提供給用戶“他們難以拒絕的內容”(遊戲邦注:如特殊行動,巨大的折扣,登錄時來自推送通知或彈出窗口的信息);

4.重複購買的比例上升了,更多忠實用戶留在了產品中;

5.利潤增加。

上述提到的例子都只使用兩個參數:時間和頻率。

而添加消費金額參數到報告中將讓你能夠使用每個用戶的支付金額。

除此之外,這樣的分析也可以是基於用戶數量或你從他們那賺到的錢。

我們還能夠將消費金額–時間(即用戶花了多少錢以及他們上次花錢是在多久前)和消費金額–頻率(用戶花了多少錢以及他們花錢的頻率)結合在一起。

而基於一個參數框架去分析付費用戶的最簡單的方法便是根據時間,頻率和消費金額去決定用戶和他們的消費的分佈。

根據付費用戶在免費遊戲中的消費規模而對他們進行的分析經常使用一些海洋生物作比喻:

鯨魚—-帶來巨大收益的用戶;

海豚—-帶來平均收益的用戶;

小魚—-帶來較少收益的用戶。

在這裏我們並不是在談論一次付費的總數,而是用戶在整個付費期間所支付的總體金額。再一次地,這裏的巨大,平均和較少的總額也是基於專家的評估。

通過分析每個部分的用戶數以及你從每個部分的用戶中賺取的錢數,你便能夠判斷該採取怎樣的行動去提高收益。降低價格?提高價格?專注於“鯨魚用戶”的留存率?

data(from gamasutra)

data(from gamasutra)

在我們的devtodev.com中,我們根據用戶的消費金額將其劃分成五個部分,即多了“大鯨魚”和“大海豚”。從討論例子中我們可以看出收入的主要部分是來自“鯨魚”和“大海豚”用戶,因此市場營銷規劃應該主要專注於這類型用戶。

而我們提到的這些內容只是分析付費用戶的衆多方法之一。還有很多其它問題和方法能夠幫助你更好地定製自己的項目盈利方法。例如:

你的用戶轉換成付費用戶的速度?是在第一次購買,第二次購買還是在第十次購買的時候?

用戶願意花錢買什麼?爲什麼他們會成爲付費用戶?

你在用戶的第一次購買時能賺到多少錢?在用戶重複消費時又能賺到多少錢?

新手能帶給你多少錢?而資深用戶又能帶給你多少錢?

而我們也會在之後的文章中告訴你所有這些問題的答案。

本文爲遊戲邦/gamerboom.com編譯,拒絕任何不保留版權的轉發,如需轉載請聯繫:遊戲邦

How to analyze paying users? Part 1, RFM-analysis

by Vasiliy Sabirov

So, paying users are those guys that bring money to your product. It is very important to know all the nuances of their behavior: what do they pay for, how fast and how much. It is important to know what they feel by doing this, do they get satisfaction from the investments they made into your product. In fact, even in the case of f2p online-game every payment made by gamers is their investment: at the beginning they pay, at the end they get some ROI, which can be denominated in a currency or in the emotions they experience. Therefore, you should perceive users as investors even if they are minority investors.

In order to better understand the behavior, distribution and needs of paying users there exist special analytical methods and reports. Today we will talk about RFM-analysis as one of the basic methods of understanding the structure of your paying audience.

RFM stands for:

R – Recency – how long ago the last purchase was made;
F – Frequency – how often purchases were made;
M – Monetary – the volume of purchases during all the time.

You give three marks that correspond to those parameters to every paying user. As a rule, in theoretical materials a user is assessed with a three-point scale (relatively speaking: good, normal, bad), however in practice we have also faced five- or even ten-point scales in RFM-analysis. To make it easier, let’s look at the example of the three-point system:

R = 1, it has been a long time since a user payed last time;
R = 2. last payment was made relatively short time ago;
R = 3, a user has payed recently;

F = 1, a user pays very rarely;
F = 2, a user pays with moderate regularity;
F = 3, a user pays often;

M= 1, the sum of all payments is small;
M = 2, a user have payed moderate amount of money to the project;
M = 3, a user have payed much money.

Of course, the question arises: how to understand in this case long time ago/recently, often/rarely and much/little. It is possible to answer this question in two ways:

1.Expert assessment. No one knows your project better than you. Therefore, define for yourself what is long time ago and recently, what is much and little. Let’s say, long time ago is a month and more ago, rarely is once a month or more rarely, little is not more that hundred rubles during the whole payment history.

2.Quantiles and quartiles. Let’s recall mathematical statistics. Arrange your users according to one of the parameters (for example, to the sum of payments made during all the time), take, for instance, top-5% of all the users and say that these users payed much. Congratulations, you have already got five-percent quantile of your users’ sample. You can also take quartiles (quartile = quantile of 25% -level), and assess first quartile as much, last as little, and what is between them as an average size of a payment. Be that as it may, even when you use quantile and quartile you cannot do without subjective assessment, therefore, look at point 1 again.

Anyway, having spent some time in Excel (or somewhere else), you will give a mark to every paying user for recency, frequency and amount of payments.

Now is the most interesting part.

You can see how those marks are distributed among your paying users and what are the majority of users. This will allow you to segment your paying audience and plan your marketing actions aimed at raising profit.

A simple example:

Purchased recently, but rarely (or only one payment) – new paying users. What should you do with them? Express your gratitude! Your aim is to stimulate them to make repeated purchases. As many researches show, repeated purchases, their regularity and size raise the app’s chances to earn a million of dollars.

Purchased recently and purchase often – loyal users. They do not need additional stimulation, however you can find a way and thank them for their loyalty (unexpected bonus, surprise, just “thanks” – all these work).

Purchased often, but long time ago – loyal users on the verge of leaving. In other words, these are the money that right now slip through your fingers. Your aim is to remind them about yourself. Maybe, a simple push-notification will be enough. Maybe, it is worth asking them what has changed and why they are leaving.

Purchased rarely and long time ago – outflow of users. They didn’t become loyal, something in the past prevented them from doing it. You can propose them (and not only them) an action – even if it is not profitable for you – which would stimulate them to make a repeated purchase and to return to the product. Otherwise, at least you can try to find out what they didn’t like, and adjust the product based on the feedback.

Imagine the following case.

1.Project X wants to raise its income;

2.They conduct RFM-analysis, and it shows that:

1)the outflow of loyal users is very high;

2)many users make only one purchase.

3.They introduce some triggers to the project that allow to recognize the moment when the stay of a user in a status “one purchase” is too long or when a user that was loyal before stops paying. At these moments users get “an offer that they can’t refuse” (special action, big discount, information is delivered by push-notification or pop-up window when logging);

4.The percentage of repeated purchases rises, more loyal users stay with the product;

5.Profit.

Both of discussed examples operate only two parameters: Recency, Frequency.

Adding the Monetary parameter to the report will allow to use volumes of every user’s payments in addition.

Besides that, the analysis can be conducted based on the quantity of users or based on money that you get from them.

In addition, it is possible to look at the combination Monetary-Recency (how much do users pay and how long ago did they pay), Monetary-Frequency (how much and how often do users pay).

The easiest way is to analyze paying users in the frames of one parameter is to get distribution of users and their payments depending on time (long time ago – recently), on frequency (often – sometimes – rarely), on amount (much – average – little).

In particular, analysis of paying users according to the size of their payment in f2p-games is usually described with the help of the inhabitants of sea depths:

Whales – users that bring big sums;

Dolphins – users that bring average sums;

Minnows – users that bring small sums.

Here we don’t speak about the sums of one payment but rather about general sums accumulated during the whole payment history of a user. The differentiation on big, average and small sums is made based on expert assessment again.

By analyzing the amount of users in every segment and the amount of money that you get from every segment you will be able to understand, which actions are better for raising the profit. To lower prices? To raise prices? To focus on the retention of “whales”?

At our service devtodev.com we have divided users depending on the volume of their payment into five segments additionally defining “grand whales” and “grand dolphins”. In particular, the discussed example shows that the main part of the income is brought by “whales” and “grand dolphins”, therefore, the focus of the marketing forces should be on them.

This is only a part of methods that can be used to analyze paying users. There are many more questions, answers to which will help you to customize your project’s monetization better. There are only some of them:

How fast your users are converted into paying? During the first, the second or the tenth purchase?

What do users pay for? Why do they become paying after all?

How much money do you make on the first payments of gamers, how much – on repeated?

How much money do beginners bring to you, how much do “oldies”?

We will surely tell you about every method in detail in our future articles.

If you don’t want to wait, we invite you to the free webinar, which will be held on October, 27th 2015 at 12 p.m CDT.

During the webinar we will tell you about all the methods of paying users’ analysis, describe cases devoted to how analytics of paying users helps to raise the income. (source:Gamasutra