Loading ...
Sorry, an error occurred while loading the content.

19152Re: [webanalytics] Re: Google Analytics conversion discrepancies

Expand Messages
  • Tim Leighton-Boyce
    Sep 3, 2008
    • 0 Attachment
      Hi Hugh,
      I now have several days worth of data to consider.

      First I have confirmed for myself that there certainly is a latency issue,
      as well all knew already. I saw a 38% increase in goals for 31 August
      between reports run on 01 and 02 September and a 36% increase in
      transactions. The counts did not increase in the reports run on the 3rd day
      (today)

      The figures for 01 September showed slightly smaller increases between day 1
      and day 2 -- 29% for goals and 31% for transactions.

      Returning to your main point now: once the figures had stabilised my overall
      group showed that number of transactions was slightly more than goals: 1.2%
      for the 30th and 0.47% for the 31st.

      The difference for September 1st was 1.29% -- of course I cannot yet confirm
      that figure will not change by tomorrow, but I suspect that it will not. I
      mention the September 1st figure only because it does contain a single
      instance where one of the sites did 'settle' showing fewer transactions that
      goals. About 1% fewer.

      It looks to me as if I am confirming what others have suggested would be the
      case and not what you're experiencing yourself. So it's of no help. But it
      was certainly an interesting exercise and I'm glad to have done it since it
      confirms, in my opinion, that the system is performing very well given the
      inherent limitations of the technology. I'm always pleasantly surprised by
      how accurate it is. Of course these figures have not been cross-checked
      against the back end systems for these sites. It's been a while since I went
      down that painful road, but when I did we used to run with a discrepancy of
      less than 3%, I think.

      Tim



      2008/9/2 Hugh <hugh.gage@...>

      > Hi Tim,
      >
      > Firstly at least it's not just me, so F.U.D. slightly relieved.
      >
      > On the scheduling issue, I am certain that GA has a latency issue with
      > goals and I have stopped taking the previous day's goal data seriously
      > until at least a full 18 hrs has elapsed since the end of the
      > reporting day in question.
      >
      > Thinking about the analysis you are doing I have just downloaded
      > weekly goal and e-commerce conversion data on one of my clients from
      > w/c 6th July to w/e 30th Aug so hopefully all scheduling / latency
      > issues will have been eliminated. Looking at the data in each week
      > over the course of that period, goal conversion is higher than
      > e-commerce conversion by an average of 2.6% (not percentage points).
      >
      > As mentioned by Brian Clifton in his post
      > (
      > http://www.advanced-web-metrics.com/blog/2007/10/23/transactions-v-goal-conversions/
      > )
      > I would expect if anything e-commerce to be higher than goals.
      >
      > It's interesting that your test on the 30th showed more transactions
      > than goals in all cases. I wonder if there is some other tracking
      > issue between the two perhaps related to the tagging process or if the
      > issues just vary from site to site.
      >
      > I'd be interested in anything else you unearth.
      >
      > Hugh
      >
      >
      > --- In webanalytics@yahoogroups.com <webanalytics%40yahoogroups.com>,
      > "tim.lboyce" <tim.lboyce@...> wrote:
      > >
      > > Hi Hugh,
      > >
      > > I realise that I risk creating Fear Anxiety and Doubt in raising this,
      > > but here goes.
      > >
      > > I wonder whether different processing schedules may apply to
      > > e-commerce data and page views and if this could have a bearing on
      > > your question?
      > >
      > > For quite a while now I have noticed how the e-commerce conversion
      > > rate always looks alarming when I do my daily scoot around various
      > > profiles first thing in the morning (UK time). In fact, I'd stopped
      > > looking at it and had formed the impression that it tended to change
      > > during the course of the day.
      > >
      > > But, by coincidence, this morning before even reading your post I had
      > > noticed that the figure looked more like the truth on at least one
      > > site and had lodged a vague thought that things might have improved.
      > >
      > > See what I mean about F.U.D? I feel very wary about even mentioning
      > > this in case it fuels yet more doubt. Especially when my own first
      > > attempts to investigate further suggest that I am talking nonsense!
      > >
      > > By the way: my normal analysis is done on weekly figures, compiled on
      > > a Tuesday, so my impression that there might be a timing issue did not
      > > cause any problems in my real life. That's why I've never bothered to
      > > even try to check whether there was any basis for my impression.
      > >
      > > However, as a form of penance for all the confusion I risk causing by
      > > this post, I've just started a quick bit of investigation.
      > >
      > > I've compared transactions and goals across a random group of half a
      > > dozen sites, running on 3 different e-commerce platforms (partly to
      > > make the sample more general and partly to ensure that the data cannot
      > > be identified). Between them they've done about 1,300 transactions per
      > > day on the last couple of days. None of these are super-busy, but
      > > they're not tiny sites either.
      > >
      > > On the 30th all of them reported more e-commerce transactions than
      > > goals. The overall difference came to 1.2% more. There were variations
      > > within the group. One was spot on. The worst reported 2% more
      > > transactions than goals.
      > >
      > > The situation for yesterday (report run just now, 13:30 -- which may
      > > become relevant) was more varied. The average was 1.7% more
      > > transactions than goals. Which might appear to knock my theory on the
      > > head.
      > >
      > > But the variation was much more extreme. In fact some sites DID report
      > > fewer transactions than goals (2% fewer in one case) and at the other
      > > extreme one site reported 6% more transactions.
      > >
      > > Of course the real test will be to see whether the numbers have
      > > changed tomorrow. I will also try to check the data for today earlier
      > > in the morning and then again in the middle of the day.
      > >
      > > At the moment it looks as if there's nothing here to help throw light
      > > on your particular case, but I'll report back anyway.
      > >
      > > Tim
      > >
      > >
      > > --- In webanalytics@yahoogroups.com <webanalytics%40yahoogroups.com>,
      > "Hugh" <hugh.gage@> wrote:
      > > >
      > > > Hi Feras
      > > >
      > > > Thanks for your response. What I should have included in my original
      > > > post is that the total volume of transactions in the Goal is higher
      > > > then that shown in the e-commerce tracking. I think it should at least
      > > > be the other way around.
      > > >
      > > > Hugh
      > > >
      > > > --- In webanalytics@yahoogroups.com <webanalytics%40yahoogroups.com>,
      > "Feras Alhlou" <feras@> wrote:
      > > > >
      > > > > Hi Hugh,
      > > > >
      > > > > It working as designed :). In Google Analytics' goal tracking
      > > reports, a
      > > > > visitor can convert only once, i.e. let say your goal is a
      > > > successful form
      > > > > submission or ecommerce/sales transaction, and a visitor submitted
      > > > the form
      > > > > twice or made two transactions (i.e. converted twice), the goal
      > > tracking
      > > > > would count it as one conversion while ecommerce would count it
      > as two
      > > > > transactions. The logic is that you really can only "convert" a
      > > > visitor into
      > > > > a lead/customer once.
      > > > >
      > > > > To get more details about this topic, check out the blog post by
      > Brian
      > > > > Clifton
      > > > http://www.advanced-web-metrics.com/blog/2007/10/23/transactions-v-
      > > > > goal-conversions/, this is especially helpful if you want to use the
      > > > > eCommerce feature in GA for track non-ecommerce transaction (a
      > > very cool
      > > > > thing providing very rich data!).
      > > > >
      > > > > Also, I believe Justin Cutroni's eBook,
      > > > > http://oreilly.com/catalog/9780596514969/, covers some of these
      > > > techniques
      > > > > as well.
      > > > >
      > > > > Hope you find the above helpful.
      > > > >
      > > > > That is it for the week :), for those of you in the US and Canada
      > > > and other
      > > > > places that celebrate labor day at this time of the year, happy
      > labor
      > > > > day/long weekend!
      > > > >
      > > > > ________________________________
      > > > > Feras Alhlou
      > > > > website: www.e-nor.com
      > > > > blog: www.e-nor.com/blog
      > > > > ________________________________
      > > > >
      > > > >
      > > > >
      > > > > On Fri, Aug 29, 2008 at 2:53 AM, Hugh <hugh.gage@> wrote:
      > > > >
      > > > > > Hi,
      > > > > >
      > > > > > For those who are using GA on e-commerce sites and have GA
      > > e-commerce
      > > > > > tracking implemented, has anybody noticed if they are getting
      > > > > > different conversion rates on the e-commerce report compared with
      > > > > > their funnel goals where the goal page for both is exactly the
      > same?
      > > > > >
      > > > > > I have put in a query with GA support but it's gone all quite.
      > > > > >
      > > > > > Hugh
      > > > > >
      > > > > >
      > > > > >
      > > > >
      > > > >
      > > > > [Non-text portions of this message have been removed]
      > > > >
      > > >
      > >
      >
      >
      >


      [Non-text portions of this message have been removed]
    • Show all 12 messages in this topic