Most retailer marketers are sitting on a mine of unused big data. This kicks off a series of how-to-guides for constructing agnostic strategies around big data for the purpose of improving conversion.

Big Data has saturated the news cycle in 2012. But what exactly is big data, who is using it and how can your brand apply it?

Lately, you may have been wondering “How do I get big data”? 

Hold on, full stop please. Your brand is big data rich already.

Don’t believe me? Then ask yourself:

  • What were December top selling products in Nebraska vs. Florida?
  • Did we sell more products designed for men or women?
  • What is our average order value (AOV)?
  • What is the average time period between return visits?

If you don’t know the answers immediately, what would you do, where would you look? To your analytics package of course. Regardless of your brand’s product of choice, SiteCatalyst, Webtrends, Google Analytics or any other, the data already exists (while perhaps untapped).

The only question remaining: Will you let the data sit docile in 2013?

During my tenure with Adobe/Omniture I had a unique vantage point to monitor the health of all optimization engagements.  I observed that most marketers struggle with translating heaps of data into actionable to-dos.  

I realized that, retailers especially, would leave valuable mounds of data untapped. To overcome data-paralysis, I aimed to instill a culture of data-artists.

A data-artist is similar to a data scientist, but more of a creative thinker. Combining curiosity and tact, the purpose of a data-artist is to construct new strategies that exploit insights leading to profitable outcomes. 

The first principle: ask great questions to isolate great insights. For example, the questions above generated big data, before asking them only uncategorized and unexplored data sets existed.

Secondly a data-artist should strive to deliver relevant experiences.  Increased relevancy drives ROI for the retail marketer.  I encourage you to seek-out the micro-segment, and then cater to what they are in-market for. 

Turn data into cash

For onsite traffic, the retail marketer could experiment with targeting visitors based on their wallet. Marketing merchandize based on their spending habits is 1 approach to deliver a more intimate experience, reflective of the shoppers’ budgets.

A. Set up fundamentals to track as the shopper checks out

Step 1. Store these, upon checkout

Store the total transaction amount within the shopper’s cookie. If possible do a real-time lookup to your analytics package to call the average order value for the last 14 days, store this in the shopper’s cookie as a separate variable.  

Step 2. Distance to AOV

Run an equation to calculate the margin of difference between the shopper’s total transaction value and the 14 day AOV. Store the outcome in the cookie as well. 

Step 3. Define buckets for different budgets

Are there three common total transaction values which your shoppers cluster around? Perhaps a Low, Mid, High? The quantity of budget buckets you define will vary on the nature of your product offerings and their price points. 

For example if your AOV is $80 perhaps setup three buckets such as:

  • DistanceToAOV = "Ontarget" // if shopper’s total transaction is between $65-95.
  • DistanceToAOV = "BelowAOV" //  if shopper’s total transaction is less than $65.
  • DistanceToAOV =  "AboveAOV" //  if shopper’s total transaction is above $95.

Early on, keep it simple, you can always add more buckets later to segment further.

B. Target the shopper with relevant experiences during future visits

Step 4. Track category affinity

Using a propensity score, identify which product category a unique visitor is interested in.

Perhaps the simplest propensity setup should merely total which category the visitor generates the greatest amount of page views. Last category viewed is also a worthwhile approch to begin with, ie: Handbags. 

Step 5. Display relevant merchandize, within shopper’s budget

Combining what is being tracked between step three and four, on the homepage you could promote (inject) relevant merchandize from the category of greatest interest. 

Based on the DistanceToAOV variable, promote merchandize on the higher/lower end of the pricing spectrum, whichever is appropriate to the unique visitor. 

With tools like Monetate or Adobe’s Test and Target, if you were Kate Spade’s ecommerce manager, you could predefine creative for each bucket, ie a handbag on sale, a moderately priced handbag, and a premium handbag. 

Based on the setup described above, the visitor would be targeted with handbags reflective of their spending history.

The outline above is simply a starting point to inspire creativity. Please elaborate on it. I’ve intentionally painted with broad strokes to maintain a vendor agnostic prescription.  

Yes! There are an indefinite amount of profiles and methods to target visitors with This is not 100% fool-proof.  I would recommend you work with a landing page optimization consultant to transcribe the above concept into a suitable solution for your business.

The biggest takeaways are:

  • Relevancy lifts conversion rates. 
  • Retail marketers have a gold mine of unstructured big data, use it.

In the continuation of this series I will discuss how to drive social influence and audience modeling by big data for the purpose of generating conversions.

Jacob Ajwani

Published 2 January, 2013 by Jacob Ajwani

Jacob Ajwani is a serial startup executive, scaling adoption & utilization of Offermatica, Omniture, Adobe and Cognitive Match.  You can connect with him on LinkedIn or Google+.

3 more posts from this author

You might be interested in

Comments (7)

Save or Cancel
Jacob Ajwani

Jacob Ajwani, VP of Strategy at

I took this a bit deep into the weeds. If you need me to clarify any point, leave a question below.

Cheers and best of luck in 2013,

over 5 years ago

James Gurd

James Gurd, Owner at Digital JugglerSmall Business Multi-user

Hi Jacob,

Thanks for sharing your thoughts on how to mine value from big data.

I'm always unsure about how accurate category affinity is because people have different motivations for site visits that could lead to inaccurate matching. For example, if I use ASOS to shop for a present for my brother, I don't want to by hit with products my brother would like next time I visit.

I think this type of modelling can work well but only when you have a decent history of visits from which to discern patterns.

What's your take?


over 5 years ago

Jacob Ajwani

Jacob Ajwani, VP of Strategy at

James, awesome question.

The answer is to time-cap the affinity developed or create a trigger mechanism. Upon conversion you could clear the learning.

But really, wouldn't you rather mistarget because of out-of-date information rather than a mistarget made on no information. You have a GREATER chance of success using what you may know rather than just taking random shots.

The setup will learn over time...and yes make some mistakes, but those will be out-weighed by targeted content that really hits the nail on the head.

Jacob Ajwani.

over 5 years ago


Alex Kellehe

Great post, Jake.

over 5 years ago

Jacob Ajwani

Jacob Ajwani, VP of Strategy at

Installments #2 - predictive prospecting (DCO) and #3 - social influence are coming. And they will be blazing hot.

over 5 years ago


Eric Shinn

Great post, Jacob! While I don't hold nearly as deep a knowledge of data mining and analytics as you, I have noticed that director or c-level people tend to have a strong interest in learning how to use this treasure trove of information but are reluctant either because "BIG DATA" (apply echo effect) kind of intimidates them or are not sure of how to best approach implementing the key performance indicators, or KPIs, (i.e. analysis paralysis). Do you find that as well?

Also, James Gurd, brings a very interesting point. I'll give a real-world example; My fiancee asked me to send her a couple of things from Victoria's Secret (as she's outside of the U.S.) and ever since then, Victoria's Secret has been asking me to try on their latest undergarments to feel more like a woman via direct mail - direct indeed.

To add emphasis on your point of being "creative" in solving problems or finding new ways to add to the value and usefulness of data, the issue above could be addressed in a couple different ways that I can see: Option 1. data can be stored for up to a couple of years to see if it's an annual or seasonal purchase - like a gift or an atypical segment. Granted this would only work for Victoria's Secret if I bought something from the retailer every year without fail for her birthday (and a heads-up reminder of the up-coming occasion would be nice as well). or 2. More accurately, a cashier could have simply asked if it was a gift and entered that into the purchase order to be skimmed by an automated system. As a matter of fact, they didn't have to ask. I stated right away I was interested in purchasing their wares for someone other than myself - it's just not how I get my kicks you see.

over 5 years ago

Jacob Ajwani

Jacob Ajwani, VP of Strategy at

Thanks for sharing Eric. Tapping "big data" is a KBI in 2013, like you said. But lots of conversations are just talk. Taking a bite out of the apple has been a bit ambiguous and intimidating for some. Hopefully this article demystifies it a bit.

over 5 years ago

Save or Cancel

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Digital Pulse newsletter. You will receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.