Recherche avancée

Sur d’autres sites (9963)

• What is a Cohort Report ? A Beginner’s Guide to Cohort Analysis

  3 janvier 2024, par Erin

  Handling your user data as a single mass of numbers is rarely conducive to figuring out meaningful patterns you can use to improve your marketing campaigns.

  A cohort report (or cohort analysis) can help you quickly break down that larger audience into sequential segments and contrast and compare based on various metrics. As such, it is a great tool for unlocking more granular trends and insights — for example, identifying patterns in engagement and conversions based on the date users first interacted with your site.

  In this guide, we explain the basics of the cohort report and the best way to set one up to get the most out of it.

  What is a cohort report ?

  In a cohort report, you divide a data set into groups based on certain criteria — typically a time-based cohort metric like first purchase date — and then analyse the data across those segments, looking for patterns.

  Date-based cohort analysis is the most common approach, often creating cohorts based on the day a user completed a particular action — signed up, purchased something or visited your website. Depending on the metric you choose to measure (like return visits), the cohort report might look something like this :

  

  Note that this is not a universal benchmark or anything of the sort. The above is a theoretical cohort analysis based on app users who downloaded the app, tracking and comparing the retention rates as the days go by. 

  The benchmarks will be drastically different depending on the metric you’re measuring and the basis for your cohorts. For example, if you’re measuring returning visitor rates among first-time visitors to your website, expect single-digit percentages even on the second day.

  Your industry will also greatly affect what you consider positive in a cohort report. For example, if you’re a subscription SaaS, you’d expect high continued usage rates over the first week. If you sell office supplies to companies, much less so.

  What is an example of a cohort ?

  As we just mentioned, a typical cohort analysis separates users or customers by the date they first interacted with your business — in this case, they downloaded your app. Within that larger analysis, the users who downloaded it on May 3 represent a single cohort.

  

  In this case, we’ve chosen behaviour and time — the app download day — to separate the user base into cohorts. That means every specific day denotes a specific cohort within the analysis.

  Diving deeper into an individual cohort may be a good idea for important holidays or promotional events like Black Friday.

  Of course, cohorts don’t have to be based on specific behaviour within certain periods. You can also create cohorts based on other dimensions :

  • Transactional data — revenue per user
  • Churn data — date of churn
  • Behavioural cohort — based on actions taken on your website, app or e-commerce store, like the number of sessions per user or specific product pages visited
  • Acquisition cohort — which channel referred the user or customer

  For more information on different cohort types, read our in-depth guide on cohort analysis.

  How to create a cohort report (and make sense of it)

  Matomo makes it easy to view and analyse different cohorts (without the privacy and legal implications of using Google Analytics).

  Here are a few different ways to set up a cohort report in Matomo, starting with our built-in cohorts report.

  Cohort reports

  With Matomo, cohort reports are automatically compiled based on the first visit date. The default metric is the percentage of returning visitors.

  

  Changing the settings allows you to create multiple variations of cohort analysis reports.

  Break down cohorts by different metrics

  The percentage of returning visits can be valuable if you’re trying to improve early engagement in a SaaS app onboarding process. But it’s far from your only option.

  You can also compare performance by conversion, revenue, bounce rate, actions per visit, average session duration or other metrics.

  

  Change the time and scope of your cohort analysis

  Splitting up cohorts by single days may be useless if you don’t have a high volume of users or visitors. If the average cohort size is only a few users, you won’t be able to identify reliable patterns. 

  Matomo lets you set any time period to create your cohort analysis report. Instead of the most recent days, you can create cohorts by week, month, year or custom date ranges. 

  

  Cohort sizes will depend on your customer base. Make sure each cohort is large enough to encapsulate all the customers in that cohort and not so small that you have insignificant cohorts of only a few customers. Choose a date range that gives you that without scaling it too far so you can’t identify any seasonal trends.

  Cohort analysis can be a great tool if you’ve recently changed your marketing, product offering or onboarding. Set the data range to weekly and look for any impact in conversions and revenue after the changes.

  Using the “compare to” feature, you can also do month-over-month, quarter-over-quarter or any custom date range comparisons. This approach can help you get a rough overview of your campaign’s long-term progress without doing any in-depth analysis.

  You can also use the same approach to compare different holiday seasons against each other.

  If you want to combine time cohorts with segmentation, you can run cohort reports for different subsets of visitors instead of all visitors. This can lead to actionable insights like adjusting weekend or specific seasonal promotions to improve conversion rates.

  Try Matomo for Free

  Get the web insights you need, without compromising data accuracy.

  Start for free

  No credit card required

  Easily create custom cohort reports beyond the time dimension

  If you want to split your audience into cohorts by focusing on something other than time, you will need to create a custom report and choose another dimension. In Matomo, you can choose from a wide range of cohort metrics, including referrers, e-commerce signals like viewed product or product category, form submissions and more.

  

  Then, you can create a simple table-based report with all the insights you need by choosing the metrics you want to see. For example, you could choose average visit duration, bounce rate and other usage metrics.

  

  If you want more revenue-focused insights, add metrics like conversions, add-to-cart and other e-commerce events.

  Custom reports make it easy to create cohort reports for almost any dimension. You can use any metric within demographic and behavioural analytics to create a cohort. (You can explore the complete list of our possible segmentation metrics.)

  We cover different types of custom reports (and ideas for specific marketing campaigns) in our guide on custom segmentation.

  Create your first cohort report and gain better insights into your visitors

  Cohort reports can help you identify trends and the impact of short-term marketing efforts like events and promotions.

  With Matomo cohort reports you have the power to create complex custom reports for various cohorts and segments. 

  If you’re looking for a powerful, easy-to-use web analytics solution that gives you 100% accurate data without compromising your users’ privacy, Matomo is a great fit. Get started with a 21-day free trial today. No credit card required. 

  Try Matomo for Free

  21 day free trial. No credit card required.

  
  

  Your email address
  

  Your website address
  

  Your Analytics subdomain will be .matomo.cloud (change)

  Choose your analytics subdomain
  

  .matomo.cloud

  I accept the terms and conditions and data processing agreement (DPA)

  
  Your information will be used to create an account on our cloud service. For more information please consult our privacy policy.
  
  
  

  Start improving your websites now

  

• Announcement : Piwik to focus on Reliability, Performance and Security

  7 octobre 2014, par Matthieu Aubry — About, Community

  To our valued team and community,

  Well, we have moved fast and achieved so much during the past few months. Relentlessly releasing major version after major version… We got a lot done including several major new features !

  The speed of adding new features was a great showcase of how agile our small teams and the larger community are. And I’m so proud to see automated testing becoming common practice among everyone hacking on Piwik !

  For the next few months until the new year we will focus on making what we have better. We will fix those rare but longstanding critical bugs, and aim to solve all Major issues and other must-have performance and general improvements. The core team and Piwik PRO will have the vision of making the existing Piwik and all plugins very stable and risk free. This includes edge cases, general bugs but also specific performance issues for high traffic or issues with edge case data payloads.

  We’ll be more pro-active and take Piwik platform to the next level of Performance, Security, Privacy & Reliability ! We will prove to the world that Free/Libre Web software can be of the highest standard of quality. By focusing on quality we will make Piwik even easier to maintain and improve in the future. We are building the best open platform that will let every user liberate their data and keep full control of it.

  If you have any feedback or questions get in touch or let’s continue the discussion in the forum.

  Thank you for your trust and for liberating your data with Piwik,

  Matthieu Aubry
  Piwik founder

  More information

  • How long is each release of Piwik maintained for ? (FAQ)
  • When is the next release of Piwik ? What is the release schedule ? (FAQ)
  • Pulse of Piwik : our monthly pulse report shows 27 different people contributed to Piwik core in the last month !

  

  This is an amazing testament of the power of free/libre software and yet we think this is just the beginning. We hope more developers will join and contribute to the Piwik project !

• Executing ffmpeg command using Popen

  21 septembre 2014, par dragonator

  I have a strange problem trying to execute ffmpeg command using Popen.
  I have the following piece of code, which I use for executing an external commands in Python :

  from subprocess import Popen, PIPE
  
  from datetime import datetime
  
  
  
  
  
  class Executor(object):
  
  
  
      @classmethod
  
      def execute(cls, command):
  
          """
  
          Executing a given command and
  
          writing into a log file in cases where errors arise.
  
          """
  
          p = Popen(command, stdin=PIPE, stdout=PIPE, stderr=PIPE)
  
          output, err = p.communicate()
  
          if p.returncode:
  
              with open("failed_commands.log", 'a') as log:
  
                  now = datetime.now()
  
                  log.write('{}/{}/{} , {}:{}:{}\n\n'.format(now.day, now.month,
  
                                                             now.year, now.hour,
  
                                                             now.minute,
  
                                                             now.second))
  
  
  
                  log.write("COMMAND:\n{}\n\n".format(" ".join(command)))
  
                  log.write("OUTPUT:\n{}\n\n".format(output.decode("utf-8")))
  
                  log.write("ERRORS:\n{}\n".format(err.decode("utf-8")))
  
                  log.write('-'*40)
  
                  log.write('\n')
  
  
  
              return ''
  
  
  
          if not output:
  
              output += ' '
  
  
  
          return output

  I’ve tested it with others commands, but when I try to execute ffmpeg command - it fails.
  I’m trying to convert some audio format to mp3 format.
  Here is an example of my command :

  ffmpeg -i "/path/old_song.m4a" "/path/new_song.mp3"

  ...simple as that.When I run it in terminal it works fine, but when I try to execute it using the above function it fails.
  Here is the exact error :

  ----------------------------------------
  
  21/9/2014 , 19:48:50
  
  
  
  COMMAND:
  
  ffmpeg -i "/path/old_song.m4a" "/path/new_song.mp3"
  
  
  
  OUTPUT:
  
  
  
  
  
  ERRORS:
  
  ffmpeg version 2.2.3 Copyright (c) 2000-2014 the FFmpeg developers
  
    built on Jun  9 2014 08:01:43 with gcc 4.9.0 (GCC) 20140521 (prerelease)
  
    configuration: --prefix=/usr --disable-debug --disable-static --enable-avisynth --enable-avresample --enable-dxva2 --enable-fontconfig --enable-gnutls --enable-gpl --enable-libass --enable-libbluray --enable-libfreetype --enable-libgsm --enable-libmodplug --enable-libmp3lame --enable-libopencore_amrnb --enable-libopencore_amrwb --enable-libopenjpeg --enable-libopus --enable-libpulse --enable-librtmp --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libv4l2 --enable-libvorbis --enable-libvpx --enable-libx264 --enable-libx265 --enable-libxvid --enable-pic --enable-postproc --enable-runtime-cpudetect --enable-shared --enable-swresample --enable-vdpau --enable-version3 --enable-x11grab
  
    libavutil      52. 66.100 / 52. 66.100
  
    libavcodec     55. 52.102 / 55. 52.102
  
    libavformat    55. 33.100 / 55. 33.100
  
    libavdevice    55. 10.100 / 55. 10.100
  
    libavfilter     4.  2.100 /  4.  2.100
  
    libavresample   1.  2.  0 /  1.  2.  0
  
    libswscale      2.  5.102 /  2.  5.102
  
    libswresample   0. 18.100 /  0. 18.100
  
    libpostproc    52.  3.100 / 52.  3.100
  
  "/path/old_song.m4a": No such file or directory
  
  Conversion failed!
  
  
  
  ----------------------------------------

  ...and as you can think of - the file exists.

  I think there is something in passing the command to Popen.communicate but I don’t know exactly.

  Kind regards,

  Teodor D.
  PS : I’m passing the command to Executor.execute as Python list.

  PSS : Calling the Executor.execute :

  def process_conversion(self):
  
      for song in self.files_to_convert:
  
          current_format = song.rsplit('.', 1)[-1]
  
  
  
          old_file = '"{}{}{}"'.format(self.target_dir, os.sep, song)
  
          new_file = '"{}{}{}"'.format(self.target_dir, os.sep,
  
                                       song.replace(current_format, 'mp3'))
  
  
  
          command = ["ffmpeg", "-i", old_file, new_file]
  
          Executor.execute(command)