Platt Perspective on Business and Technology

Building a startup for what you want it to become 26: moving past the initial startup phase 12

Posted in startups by Timothy Platt on September 9, 2017

This is my 26th installment to a series on building a business that can become an effective and even a leading participant in its industry and its business sector, and for its targeted marketplaces (see Startups and Early Stage Businesses and its Page 2 continuation, postings 186 and loosely following for Parts 1-25.)

I focused in Part 25 on the raw data and the processed knowledge that a business accumulates that it uses, in one direction to help shape its strategic and operational plans, and that it uses in the other direction when executing them and evaluating the results achieved. And as part of that, I posed a brief set of questions, the likes of which would go into any quality control effort for managing and more effectively using those data stores, which I repeat here for purposes of continuity of narrative:

1. Is this data reliable, and if so for what? I parse that question, offered here as a general point of principle, into a set of more focused related questions.
2. Where did it come from? And how reliable is its source from prior experience?
3. Is it complete and unedited or has it been pre-filtered or re-represented in some way, by a stakeholder who might be bringing their own biases or agendas with them when offering it? Answers to this question would in most cases be more presumptive than conclusive but evidence of possible filtering or bias should raise red flags and should always be considered as a possibility. As an example of how pre-filtering can be carried out without any intent of adding bias into a data set but still end up adding that in, consider how data can be “cleaned up” before use by deleting from consideration, unexpected and seemingly out of pattern outliers and other “anomalies”, while removing second copies of duplicated records and the like and doing similar data cleansing. That happens and it should raise red flags.
4. Is this data consistent with other data gathered and with expectations in place, or is it divergent from or contrarian to that? Note, new and different and unexpected should not rule out new data findings. But they should prompt closer and fuller examination and particularly if their inclusion would significantly shape conclusions drawn and actions taken.
5. And of course, what would this data suggest, and certainly when considered in the larger context of what is already known?
6. And what are the consequences of that, and both if this data is correct and reliable and if it is not?

I then added at the end of Part 25 that I would “discuss this set of issues in more detail in a next series installment where I will focus on specific types of raw data as business intelligence, and in the more specific context of an at least briefly sketched out working business example.”

My goal for this posting is to set up a conundrum, or at least a realistic-seeming systems example that highlights within it a combination of competing needs and requirements. And I begin that with a set of seemingly simple questions, that I pose in the context of a retail business that maintains a complex inventory of products that it offers for sale. A store of this type is data-driven, with their overall business performance depending on the aggregate sales performance of what they offer, as cumulatively determined on a product by product basis.

• What sells at what rate and at what volume, at any given point in time? This might or might not have a seasonal or other cyclical element to it, as just one reason why this type of question has to be recurringly reconsidered.
• How is this trending, for the various stock keeping unit (SKU) product types offered?
• What is their turnover rate for the business?
• And what is their profit margin when they do sell? That is actually a more complex question than it might at first seem, where a product that simply sits on a self or in back room storage as inventory waiting its turn on a sales floor shelf, accumulates additional cost to the business by taking up room that faster selling items might fill, and more profitably so. But even that more expansive evaluation leaves out the possibility of loss leader product offerings that might intentionally be sold at or even below cost in order to bring in customers, with them offered as marketing tools for driving larger sales.
• What products might be calling for greater shelf space or greater specific model diversity offered, or both? And what would best be reduced for the shelf space that it commands, or even discounted for clearance sale and discontinued, and either for now because of seasonal or other shifts, or permanently?

These are just sample orienting questions, even if they were selected here for their specific relevance to most retail businesses. And all of them require both specific data, and in quantity, and specific analysis to answer them. More generally:

1. What questions would you need to ask and find answers to, in order to more effectively optimize your business, and both to make it more agile and effective in the face of changing market demands, and to make it more profitable and consistently so?
2. Now what data would you need to answer those questions, and both by type and by quantity, and with what data quality control in place?

There are several ways to parse and categorize data but one that is particularly relevant here is:

• Data that is subject to direct statistical analysis, which can mean numerical, binary (e.g. yes or no) or similar
• And data that cannot be so coded and used, such as free form text responses.

For simplicity, let’s assume that all of the data to be gathered and used fits into the first of those categories, making straightforward statistical analysis possible. The more questions you seek to address through such statistical analyses, the more complex they become for types and combinations of data required to address them. And the greater the certainty in any conclusions reached when doing these analyses, the more data you would need in order to carry out these statistical analyses too. And this leads me to my third and fourth questions:

3. How much data do you actually need, in order to answer your statistical questions and do the statistical modeling that you would require?
4. And precisely what data analysis-based questions do you really have to ask in order to meet your business planning and performance review needs?

I will set question 4 aside for the moment and focus on number 3 of this list. Data becomes expensive and certainly in volume:

• To systematically gather it in and store it in usable forms in usable database records
• And with effective data management systems in place to clear out duplicated or defective records, and old and no longer reliable ones (e.g. for “current” customer identification and tracking) – and without adding in bias.
• And data analysis that is based on this, becomes expensive too and particularly when outside expertise is required for carrying out complex statistical tests, on appropriately scaled data sets.
• And the more complex the tests to be performed, the larger the data samples are required to be, and the larger still, the overall pool of raw data that those test samples would be at least semi-randomly drawn from.

Most retail stores would in fact look for outliers (e.g. items that all but fly off the shelf in sales, or that alternatively only gather dust there.) And costs and timing demands would constrain them to at most, doing only simple and even cursory data analyses and business performance modeling for all that lies between those extremes and with that largely based on aggregate analysis of product categories and not of individual item types. That definitely holds for smaller and even for medium sized businesses. Massive retail business systems, tend to be the ones that truly dive into the data, and with all of the effort and expenses involved as mistakes or lost opportunities take on scales of impact, financially, that they cannot leave to chance.

This posting and this series are about startups and businesses that are still small, even if growing, so that second possibility would only be a still-distant one for them. So I finish this posting with some open questions:

• What data is available and in what quantities and with what quality and reliability?
• What questions really have to be answered, and with what level of assurance in that, from this data? (Question 4 from above, expanded)
• And closely related to that, why are answers to those questions important, and specifically so? More specifically, what specific strategic and operational questions would they help address? This bullet point is all about keeping all of this focused and relevant and in a very practical value and returns on investment-oriented sense. And what would be the consequences of simply proceeding on without rigorously addressing them?

I rephrase that last question for a startup context, by asking:

• What really consequential decisions do you face as a business founder, and which of them are data driven and in ways that would be amenable to more rigorous analysis?

If this posting sounds too abstracted from real world businesses, for it to make meaningful sense when planning and executing their strategies and operations, and certainly in young new ones, I will at least attempt to bring its rationale into clearer focus in my next series installment. My goal there is to at least begin a discussion of when and how to add greater rigor and order into a new business, to facilitate its orderly development and growth as it moves forward. Everything, or at least seemingly everything might start out looking ad hoc and new and novel at first. When and how should order be developed and added to this mix? When and how should it take over and become the basic rule for how things are done there? I tend to write about organized and systematically structured systems, and about the ad hoc approach as an often problematical alternative. My goal in the next installment of this series is to at least begin addressing how that circumstance arises.

Meanwhile, you can find this and related material at my Startups and Early Stage Businesses directory and at its Page 2 continuation.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: