In my last blog, I men­tioned Insight as an ana­lyt­i­cal tool for big data and dif­fer­en­ti­ated Insight as an ana­lyt­i­cal tool, not a report­ing tool. The key ques­tion is, what is the dif­fer­ence between report­ing and analy­sis? This is an impor­tant dis­tinc­tion for dig­i­tal mar­keters. We, as dig­i­tal mar­ket­ing con­sul­tants at Adobe, get asked this ques­tion all the time. It is a dif­fi­cult ques­tion to answer, because there is quite a bit of over­lap between the two from one per­spec­tive and sig­nif­i­cant dif­fer­ences from another viewpoint.

What is com­mon between an ana­lyt­i­cal tool and a report­ing tool?

Both tools can store large data. Both tools have defined met­rics (Key Per­for­mance Indi­ca­tors – KPI’s) and dimen­sions (attrib­utes of data). You can make deci­sions by look­ing at the data in both tools. The decision-making capa­bil­ity of the one, I would argue, is what dif­fers most.

For exam­ple, let’s look at a report. This is a some­what sim­ple exam­ple, but gets the point across very eas­ily. Here’s a report on cam­paign performance.

Num­ber of Impressions % Con­ver­sion Num­ber of Conversions
Cam­paign A

1000000

2.10%

21000

Cam­paign B

1000000

2.00%

20000

 

Age Num­ber of Impressions % Con­ver­sion Num­ber of Conversions
Less 35

1000000

2.10%

21000

Greater 35

1000000

2.00%

20000

 

Gen­der Num­ber of Impressions % Con­ver­sion Num­ber of Conversions
Male

1000000

2.10%

21000

Female

1000000

2.00%

20000

Look­ing at the report, it is obvi­ous that noth­ing much sep­a­rates cam­paign A from Cam­paign B. The con­clu­sion would be to con­tinue the ad-spend as is, with­out mak­ing any change. Let’s stick to that deci­sion, since that is the best choice we can make with the data we have.

What sep­a­rates an ana­lyt­i­cal tool from a report­ing tool?

If I am using an ana­lyt­i­cal tool, I would quickly try to iso­late the data for each cam­paign and attempt to see any trends. Let’s select “Cam­paign A” and exam­ine its behav­ior for each age segment.

Age Num­ber of Impressions % Con­ver­sion Num­ber of Conversions
Less 35

500000

0%

0

Greater 35

500000

4.00%

20000



Can we stick with the same deci­sion now? Well, you may claim this is a set-up, that the report was badly designed, and that I would have designed the report bet­ter with cross-tab with age as the sec­ond dimen­sion on the top. Well, I will grant that. Now let’s intro­duce Gen­der into the mix.

Gen­der Num­ber of Impressions % Con­ver­sion Num­ber of Conversions
Male

500000

8%

20000

Female

500000

0%

0

Well, now a cross tab does not work. We are at three dimen­sions and report­ing fails. Let’s assume there is another dimension—region.

Region Num­ber of Impressions % Con­ver­sion Num­ber of Conversions
North

250000

16%

10000

South

250000

0%

0

East

250000

0%

0

West

250000

16%

10000

You get the point. It is a los­ing bat­tle to ana­lyze this data using these reports. In the real world, the num­ber of dimen­sions avail­able to ana­lyze a sim­ple cam­paign is on the order of 10’s of dimen­sions.  Other attrib­utes could include time of the day, day of the week, income cat­e­gory, type of cre­ative, etc. This com­plex­ity is due to the multi-dimensionality of the data.  There is no way to know up-front which dimen­sions impact a par­tic­u­lar cam­paign. Maybe for Cam­paign B it is Income Cat­e­gory and Time of Day. There is no way I can guess which dimen­sion will impact the campaign.

It is not only the num­ber of dimen­sions that com­pli­cate things. Imag­ine the com­plex­ity when con­sid­er­ing the num­ber of cam­paigns, num­ber of age cat­e­gories, and num­ber of regions. This com­plex­ity is due to the car­di­nal­ity of the dimen­sions (num­ber of items in each dimen­sion). In the real world, to run hun­dreds of cam­paigns and cre­ate reports for every com­bi­na­tion of dimen­sions is just not possible.

So, going back to the exam­ple, what would you do? I would run Cam­paign A only for males whose age is greater than 35 in the North and the West. Using the same mar­ket­ing spend and real­lo­cat­ing the mar­ket­ing spend to only the rel­e­vant seg­ments, my analy­sis would have resulted in a 100% increase in con­ver­sions for the same mar­ket­ing spend.

The goal of this blog is not to say that report­ing does not have a role, but rather to make the point that report­ing by itself is just num­bers. With­out an under­stand­ing of the big pic­ture, you can’t rely on indi­vid­ual reports to make your deci­sions.  So, in sum­mary, I would encour­age all of you to start ana­lyz­ing your data using the most pow­er­ful analy­sis tools avail­able. You’ll begin to find nuggets of gold hid­den in the data that you can use to increase your profitability.

Note: In my pre­vi­ous blog, I promised to write a blog post about cross-sell for finan­cial ser­vices. I got great feed­back on the pre­vi­ous post and decided to stay on generic top­ics of data analy­sis. I antic­i­pate that it will be another 3–5 posts before I return to the topic of cross-sell. In the mean­time, I appre­ci­ate your feed­back, so let me know what you think.

0 comments