Pinterest Analytics - From Strategic Planning to Tactical Measurement Part III

(Welcome to the last part of the series of "Pinterest Analytics - From Strategic Planning to Tactical Measurement" ! What? you haven't read the previous ones? Why standing here? Go read it now!! XD)


The Last Stop of the Journey

When we're measuring an external platform, one should never forget the multi-channel mindset. The conversion cycle might look like a "waterfall model", yet indeed, it's so much more complicated than we all think - People visits and leaves different channels before converted at the end. Pinerest might be a very good "visual catalog", yet it doesn't imply that visitor will be converted right away after they've seen the Pins. Particularly, people usually visit one channel after the other, says viewing your Pin, then join your FanPage, then get exposed to your SEM 10%-off campaign, and finally be converted on your site. Basic waterfall-style analysis (like the above engagement and click-through) wouldn't be enough for handling this situation. Luckily, we have Multi-Channel Funnel in Google Analytics. :)

If you have been following my blog, you should have read about how we could leverage the Multi-Channel Funnel feature for better ROI benchmarking and optimization across different channels using an advanced model. Sure we could apply the same framework for this section, still analyze ROI on Pins-level maybe a bit too much. How about simplified it?

The idea is simple: we want to identify how each Pin could contribute, in terms of quantity and dollars, to the final conversion in both Assisted (as a middle step across the whole funnel before conversion) and Direct way (as last step before converted). With the "bridges" we have configured for each Pin in last section (the costumed URLs), indeed, half of the work is done. :)

Thanks to the campaign tagging, everything can be done within a few clicks!

Visiting the "Assisted Conversion Report" and select "Campaign" as the major dimension. TaDa! All the Pins details are here! And let's export all these to the spreadsheet we've prepared so far.... (make sure the period for the data is since the first day of management... i will explain more later)

Indeed, we would only be interested in how much each Pin contribute rather than how it contribute, so i use "Total Contribution" and "Total Contribution Value" to demonstrate each of their achievement. (If you've been using Omniture, you will know that this is actually one of their features call "Participation Value", quite useful if you ask me.). Sounds fair so far, right?

But again, don't get satisfied yet.... Remember we have talked about the true value of a Pin is because of its timeless nature? If you still remember the ADI (Average Daily Interactions) concept in the last section, in fact, how much a Pin have contributed in lifetime should follow similar calculation to reflect its true value !! For the last time, let's do some magic for these data !!

Abstracting the Lifetime efficiency in creating value by ADC

Using the same logic as before, the Lifetime value of each Pin has finally reflected in dollar signs !

We have finally reached the end of our adventure, let's claim the rewards after all these hard work we have done!!



Unlocking The Treasure Chest... (by putting everything together!)

Our effort has finally lead us to the last step (okay i swear this is really the last one....). The Treasure Chest is now right in front of us, but need some tools to lock-pick it. They're Pivot Table and Relational Map !


1. Pivot Table


So far we have put each Pin into different Groups and Categories, plus have them placed in a well-structured spreadsheet, all because we want to leverage the powerful feature, the Best Friend Forever (BFF) of any analyst, the Pivot Table !!

No worry, i am not gonna to spend another lengthy paragraph to go through the steps in using this feature (please please master this function if you haven't....). The idea in using pivot table is to let us know how each "Category" or "Group" are performing. Yes, Pin level analysis is already very obvious based on the demonstration in last few sections, yet where the Gold nuggets hiding are in fact on the abstraction layer: Would a with-models photo or plain-product snapshot provide more dollars contribution? If my objective is for-engagement, will a stylish-photoshopped pin be better than a random Instagram snapping?

Side-note: only RAW data works in grouping using pivot table.... that's why i mentioned about exporting RAW data only..... XD

Pivot Table answers these well. Personally, i would be very interested in how each Pin or Group performing in terms of on-Pinterest Engagement and Effectiveness in Lifetime Contribution, thus I would recommend calculating the corresponding ADI (Avg. Daily Interaction for engagement, i would suggesting include RePin # for this ADI for simplicity) and ADC (Avg. Daily Contribution Value for conversions). To ensure the data are more indicative, i would also be interested in Avg. Contribution Value generated by each item (Pin or Group or Category)...

 "Average Contribution Value" is indeed avg. Goal value gained.

As a result, we should have a Pivot Table like the following (let's assume we analyze for Category along with Sub-Category).

Things're getting interesting....

Side-note: For the Days Passed that required for ADI and ADC, try creating a stand-alone column for each Pin such that it indicates the actual number of day passed. In Pivot Table, we could simply sum them all for the calculation, since all we need is the Averages, an Abstraction of the problem itself.

Time for some drawing !!


2. Relational Map

A pretty straight forward step, just map everything to a relational map and visualize the data.

The "Treasure Map": x = ADI (Engagement); y = ADC (Lifetime $); size = (Unit $$)

Do you see anything special?

Do you see that among all "Style" item, only Print-Ad+Style has relatively low Avg. Contribution Value? Maybe increasing its contribution, says, publishing more Pins with more High-end products using the approach under that group would help improving it performance?

Do you see that Print-Ad+Celebrity has surprisingly high engagement? Does this due to the celebrity effect? What if we leverage that celebrity for our next promotional Pin, shall we expect it will contribute more to our revenue if we target for High-end products?

Do you see that the more generic presentation from Catalog+Style almost has the least ADC? Is it because those images are getting too old? Is there a way to refresh its content? Otherwise, how shall we prepare our next Catalog+Style Pin? Maybe Instagram snapshot style would help?

Questions, questions, questions. My dearest Analysts, the rest will be your show time. :)



Last Thoughts

First of all, if you're reading this sentence, thank you so much for the company with me so far. This's my first time to write such a long post in a single run.... (16 hours in total i guess.... Orz). Do drop me a line if you think this post is "Wow this is very informative and thoughtful!" or "Damn not another technical lecture...". Any comment is always welcome. :)

Back to the topic, in fact i think Pinterest is currently toooooo hot for marketers to engage, especially for those without a sounded plan and strategy. I know that some startups are working hard on the analytical field for Pinterest, like Pintics, but personally i still believe that analytics have to walk hand-in-hand with strategy in order to maximize the effect. That's why, i have decided to take a step forward, by putting one of my prototype framework, which help strategizing anlaytics for any platforms, into a test. Pinterest is thus my first test-subject, and this series of Pinterest Analytics is my result.

I hope you enjoy this series. :)

the post or tweet it out to share this post with your friends. 


And remember to subscribe me if you want to stay tune with me. :)



See you next time.

Cheers,
Dickson W.


Stuff that you will be interested:

Advanced Multi-Channel Funnel Analysis using Google Analytics 

16 Pinterest-Clones in China (and More's Coming) !! 

4 Approaches to Estimate Cost for Multi-Channel in Google Analytics