DATA STORIES | CONTENT MARKETING | KNIME ANALYTICS PLATFORM
When the Demanding Boss Told You to SPEED UP ! (Marketers’ Nightmare)
You’re struggling to be the nimble marketer the big man wants you to be — but the marketplace is a brutal battlefield. Can a soldier win a fight without the proper equipment?
The fast-changing nature of a mature market demands not only a quick marketing response, but also one that’s specific to the geographic area where your target audience is— otherwise, you’re wasting resources.
This is why a proper support from the data analytics team is crucial to back up marketers with the right data, and to do it timely.
But for smaller-sized companies, what’s the alternative?
In this article, we’ll dig in on a case study behind the use of a multipurpose tool that you as marketers can use hands-on.
Revealing both temporal and spatial marketing insights, the tool comes at no cost. Yes, you’ve heard it right! The tool is totally free.
The Case of Walmart & Target
Recently, I came across a study concerning the two general stores; two amongst the major retail players in the US.
Year-round, Walmart receives a much larger share of total estimated visitors, compared to Target. August is Walmart’s biggest month for visitors, possibly due to back-to-school shopping. December isn’t as big of a month for Walmart, which could indicate that its customers rely more on Walmart for daily needs rather than for holiday shopping. Target saw a larger percentage of visitors during October, November, and December 2021 and January 2022. During those months, consumers are shopping for the holiday season, which might explain the higher share of visitors at Target stores.
That’s interesting, isn’t it? You wouldn’t expect that for some reason, consumers prefer one over the other when it comes to holiday shopping.
This shows the importance of reacting fast to any change in the market, before it affects one’s brand or company significantly.
But wait, there’s more! The study also says:
Year-over-year, Target has seen a significantly greater increase in foot traffic, compared to Walmart. Despite posting lower profits, this suggests that Target’s cost-cutting strategies not only kept their current customers coming back, but also attracted additional shoppers as well. This data could also suggest that Walmart shoppers were more affected by inflation than Target customers and shopped less in-person as a result. Last quarter, Walmart’s e-commerce earnings saw a 12% increase year-over-year. It’s possible that the majority of Walmart’s customers are shopping online rather than in-person, which could explain why Walmart’s foot traffic isn’t as high in comparison to Target’s traffic.
The interplay between the locations, the pricing strategy and consumer behavior in different regions, on top of the previously mentioned temporal factor, have made things more complex — yet more interesting.
This isn’t the first time comparison reports for the two retailers were made.
A 2020 academic journal article also did a comparative study for the two stores, where they found Walmart to dominate “most market structures and is not impacted significantly by Target” and suggested for Target to “operate in a comfort zone to benefit from the spillover effects of Wal-Mart’s presence and avoid cannibalizing its own stores.”
Reading those reports provided me with an impetus to understand the visitors of these two retail stores better.
I’m interested to see what’s the recentmost group of customers of these two companies look like demographically and psychographically.
Since not all Walmart stores share the same regional location as Target, I had to exclude all non-overlapping regions.
The first thing I did was to prepare the list of geo-coordinates where the two companies overlap in their local regions. This was done manually in an Excel file, using the publicly available information on the websites of both companies alongside Google Map as reference.
Following the creation of the complete list of coordinates, I then proceeded to pick one representative coordinate for every overlapping area where the stores of both companies co-exist. Otherwise, the list would have been too many to analyze with (i.e. consumes too much time).
Finding the representative coordinates was done in KNIME, a low-code data analytics platform. With some help from the official KNIME Forum, I was able to combine the use of the Palladian’s Extension in KNIME and KNIME’s own native nodes to look for the representative coordinates.
The Magic Tool for the Case Study
While KNIME itself is a powerful platform, its potential stretches as far as where your skills and knowledge can take you.
The skyline where you can reach out to, using this open source app, is almost always untouchable — as they say, the vast sky is the limit — the more knowledge you accumulate, the more you’re humbled with its unknown potential.
For the specific purpose of case studies like this, I combined some of the skills I had from the past and that which I developed recently as well as along the way, towards developing a custom tool — a customer avatar research tool with data analytics as the backbone.
I’ve distributed it openly via the official KNIME Community Hub repository and made sure its interface be designed as intuitive as I can make it be, such that new users unfamiliar with KNIME are (hopefully) able to adopt its use, in alignment with the spirit of open sourcing.
While the methods to use the tool is beyond the scope of this article, I would recommend my other article that serves as the introduction to the customer avatar research tool I developed. It covers everything technical on the front-end aspects, whilst the back-end aspects will be covered in the near future — *coughs* — is this the part where I tell you to subscribe to my Medium to get notified? 🤭
Now, without getting too technical, here’s some context before we dive straight to the findings:
- ~12 miles radial distance from the representative locations were covered; it’s the equivalent of ~19 km, which is, according to this study, the average distance traveled by consumers to retail stores in the US
- 238,653 tweets queried in total, 78,250 tweets used in the first round to extract raw entities
- 160,403 tweets queried in the second round, 702 tweets identified as containing the key entities
- 17,806 raw entities extracted, 4,734 workable entities identified, 100 key entities identified, 4 made it to the final round
- 3,116 total Twitter users identified, 147 passes the initial clustering process, 8 passes the manual removal process, 3 made it to the final stage & thus upgraded to avatars
- 18 initial clusters formed, 4 passed the programmatic quality checking, 3 made it to the final round following manual selection
The Results
Querying the Twitter API for tweets in the above-mentioned 520 locations mentioning either @Target or @Walmart, 3 avatars representing their respective clusters were identified by my tool.
Avatar 1 is from California, Avatar 2 from New Jersey while Avatar 3 is from Georgia.
But before we meet them, I would highly recommend you to visit the link to the introductory article I provided above. Over there, I talked about the principles behind the tool, one of them revolving around entities. Without appreciating the role that entities play in a customer avatar research like this, it might be difficult for you to follow along the rest of this writeup.
Now, let’s meet the 3 avatars!
i — Adam the Friendly Neighbor (California, US)
The entity associated with Adam, the first avatar, is ‘Long Beach’ — an area in California. Adam the Friendly Neighbor makes his living as a real estate agent. As a realtor, he engages a lot with his audience on Twitter. He’s very conversational and appears nice and authoritative about the local area. His professional goal is only one; that is to sell his property listings. His pain points are not obvious; the fact that only some of his tweets are sales tweets could mean that he isn’t burdened with much listings at hand. But it could also mean that he simply focuses on building a non-salesy self-image as part of his branding and considers being pushy in sales as not part of his overall personal brand. He seems to be carefree and laidback about what he does. Based on these, it is highly likely that financial issues are the least of his worries. His psychological motivation is driven by freedom, and in general, people in this psychological group will feel the pain when their freedom to express their self-identity or self-image and to do what they want to do are being constrained. One thing about Adam’s spending attitude is that he is loyal to his favorite local stores, as exemplified by his almost daily visits to the local bakery café where he’d go to have his routine breakfast and lunch, sipping coffee happily. Based on this, in addition to the loyalty factor mentioned earlier, it can be postulated that his spending behaviors are non-impulsive, but recurrent. There’s no info whatsoever about his personal lives; even his profile descriptions are non-telling in terms of family status, social activities or any interests outside of work, except for his love for coffee. It appears that he respects the boundary between professional and personal lives. Due to his laidback-ness and being carefree, it is highly unlikely that money, time and fear could be the barriers between him and a brand. The only possible rejection when selling to Adam is likely to come from him being loyal to his favorite products or services.
ii — Bobby the Ballsy Investor (New Jersey, US)
Bobby is a podcast speaker and is an investor of Target stocks and crypto assets. Like Adam (Avatar 1), Bobby’s psychological motivation centers around freedom to do whatever he wants. This is evident from his affinity towards dipping his toes into something he knows is risky. In general, he’ll feel the pain when he’s restricted from doing what he desires. Whilst any obvious pain points are non-observant, Bobby recently suffered loss from the FTX scandal. Despite so, he did not express anger, but he did feel cheated. That however, has nothing to do with pain points, rather simply an after-effect of taking financial risks. As an active investor in financial assets, he keeps himself updated with relevant news affecting his portfolio. His goal is to accumulate wealth. It’s highly likely that he has his financial portfolio properly diversified, hinted by the fact that the cluster he belongs to has more than one entity — ‘Target stocks’ and ‘FTX’ — both of which are financial assets. This diversification possibly means that financial issues will not be a problem in Bobby’s life, at least not for the time being. Meanwhile, his spending behavior and lifestyle are non-observable neither from his tweets nor from his profile description. If there’s any barrier between Bobby and a brand, surely money and fear are not one of them.
iii — Renée the Persistent Realtor (Georgia, US)
The entity associated with Renée is ‘real estate.’ The major role she plays in her life is as a real estate agent. As a realtor, similar to Adam (Avatar 1), she too engages a lot with her audience on Twitter. But unlike Adam, Renée is very salesy. Her ultimate goal is only one; that is to continuously sell off her property listings. Due to her strong sales character, coupled with the fact that she always notifies her Twitter audience of available listings, it can be inferred that her pain points are the lack of support or resources for efficient sales. Renée is also persistent in marketing — either in the form of encouragement to buy or sell houses, or in the form of addressing possible objections from potential clients such as myths about real estate, stress and fear factors around the subject. She’s possibly an author, hinted by the fact that she supports other authors and celebrates Author’s Day. Renée sees Halloween as an opportunity to be a funny salesperson figure. In most days, however, she generally tries to appear as a trustworthy, informative and hardworking realtor. Her main motivation centers around self-empowerment and social status. This is evident from her belief that owning houses is the most important indicator of success when it comes to personal finances, and she influences her Twitter audience to believe likewise. It appears that Renée will do whatever it takes to achieve her goals and ambitions. Like the other two avatars, she keeps her personal lives private. It’s difficult to postulate her lifestyle and spending attitudes. What could be a barrier between Renée and a brand would obviously be time, since she’s constantly preoccupied in her role in real estate sales.
Highlights:
- Both Avatar 1 & Avatar 3 are real estate agents, but their attitudes in life are totally different. While the former’s life is built around being carefree in their activities in the local area and being nice to others, the latter is driven by the importance of careers, ambitions and social status.
- All 3 avatars’ goals are to accumulate wealth in one way or another. Therefore, money is definitely not a possible objection factor when selling to them.
- All 3 avatars keep their personal lives private. I find this to be interesting. To provide some background, I’ve pulled up around 3 million tweets during the prototyping tests of the tool, and never have I observed all avatars from the same batch to show this attribute all at the same time. Many profiles will at least reveal some things about their personal values, lifestyles or family status, be it in their profile descriptions or in their tweets. But obviously not this time around. This shows a really strong pattern coming from this batch, and keeping a keen eye on uncommon observations like this and acting upon it quickly can make or break one’s marketing differentiation.
There you go! The Walmart & Target’s buyer personas from the 520 representative locations. I have so far demonstrated how the tool can provide insights from a location analytics point of view.
The Time Factor (Temporal)
Earlier on, I’ve also made a point about gaining temporal insights — how important being timely is for marketers to be nimble and responsive against the dynamics of the market.
One cool thing about the tool I developed is that the built-in setting had already taken into account the time factor. The API calls which my tool perform are all bounded by the 7-day rule.
What that means is that only Tweets with maximum age of 7 days old are pulled in from the system. If that still doesn’t click, here’s what that translates to:
- The data reflects the recent market movement. If you think the villain in the marketing space is your stuck-up, demanding boss — no matter how much you want to believe that to be true — it’s actually not. Old data is. Relying on old data prevents timely response to the ebbs and flows of the market. With fresh data, you can focus your efforts on the relevant groups of people, hook them in before being swapped away by the tides.
- It limits the amount of personas analyzed in every batch, thus also limiting the number of entities found. It’s impossible to study everything at once. The tool is never designed to be a comprehensive, all-encompassing tool; the avatars discovered from every batch will still be needing data enrichment from elsewhere. Find ways to incorporate your current customer data to enrich the avatars in order to further segment them. After all, avatar research is meant to guide your decisions, by informing which distinct group of people to go to at the right time, but it doesn’t specify the exact identity of the person within that group. That’s something you’ll have to decide yourself based on the current resources you have at hand.
- To the point above, limitation doesn’t equal restriction. The customer avatar research tool is designed to extract all entities identified by the Twitter’s built-in entity recognition system, before subjecting those to a quality check I’ve put in place. Every time you run a query, you’ll get different raw (unprocessed) entities based on what the users tweet about in their first 100 tweets within 7 days time from the day you use the tool. This ensures variety in terms of clusters and avatars outcome. In my past usage of the tool, running the exact query on the same coordinates but on different days gave me different results; sometimes I received as many as 14 clusters, sometimes only 1. Using entities as the basis of clustering allows the users of the tool to distinguish one group from another naturally, since entities represent either the constructs, objects or events they are interested in — that tells a lot about who they are both individually and collectively. The results reflect the latest tweet content obtained from the query.
- Tactics-wise, this is what it means: Targeting curated ads to the relevant people until you intend to change the target audience, or until you’ve discovered new targets upon running the next batch. The same also applies on content creation. Bottom-line: The tool is not a one-timer. It’s supposed to be used periodically. Frequent use of the tool (within the allowed monthly Twitter API limits) will sharpen your ability in discovering trends and reading patterns of consumer behaviors, hopefully keeping you a few steps ahead of the others.
What Would I Do if I’m a Walmart/Target Marketer?
In this case, it’s a strong sign that the recent-most groups of customers picked up by the tool do not belong to those who’re struggling to meet basic needs. They might or might not be cost-conscious still, but at least they’re wealthy enough to potentially spend more.
To fuel sales growth, I would therefore look at the ways to entice them with non-monetary values. Instead of offering these groups of people price promotions, educate them on other values of the products instead.
A recent global consumer report by DynaData which studied just over 11,000 consumers in 11 countries including that of the US, discovered that 80% of the consumers who are driven by (non-monetary) values are willing to spend more on products that align with their personal values.
In this article, I have:
- Made the case for the need of speed in marketing, as well as the need to understand the behavior of the right group of people at the right time.
- Demonstrated how the tool can be useful, and affirmed what it does and its limitations.
Despite all these, I merely scratched the surface.
There are a lot more case studies to come.
Stay tuned!
