Why should SMBs consider investing in BI and which kinds of SMBs don't need it?

While I agree with most of Pauls' sentiments, I would add to this list:

SMB's who are looking to compete with larger shops, or who want to break in to the market share race need to seriously consider BI. When done properly, the analytics can provide insights in to trend analysis that otherwise can't be seen. BI can also provide insights in to the cost-of-acquiring new customers over time, and how those costs are related to "customer gain or loss".

SMB's who have no competition (for now), probably don't need BI (for the reasons stated above).

I would say this: BI may be a desired implementation if the data quality is so bad, that the business is just barely functioning. One of the results of BI is to "point-out" the gaps between the operational applications and the business processes. Metrics or impacts of "changes to the business process or operational systems" cannot be measured if the Gap's are not monitored over time. Without BI, the business will only "see" what is happening today.

It may be, that bad-data might be a driver or a cause to implement BI, and to force the business to take a serious look at what the bad-data is truly costing them. Then, and only then, they will understand the benefits of cleaning it up.

So I guess, I disagree with point #2 above, in that there can be "gold" in the metrics produced on Bad Data as well, as metrics produced on Good Data... Too often the business management believe that only "good data" can actually be beneficial in a BI world. Unfortunately I have found this to be a false-belief.

Cheers,
Dan Linstedt

Any small business that wants to be larger!

BI is one of those terms that should be banned from modern jargon. What's the alternative? Business Stupidity? Wait... The acronym mightn't work...

If I as the owner of a tanning salon analyze my customer list and determine time between visits and determine that customers in beachside towns visit me less often than inland customers. If I then use that information to market to those segments differently have I performed and act of Business Intelligience? Well yes. Off course I have. Did I need an expensive software package to do that? No.

BI is what intelligent business owners should be practicing every day. True amount of money they invest to automate the collection and analysis of their data is an important question and is actually what many people get caught up in. Implementing BI takes the desire to answer a question , not fancy software. Intact implenting a BI system without knowing the questions you want answeres to is a prime example of BS.

Mid-sized and smaller companies have the same reasons as large enterprises for using BI and analytics: to better understand the performance of the business, to be more competitive, to support decisions for future direction, to identify problem areas. Mid-sized / smaller businesses will have fewer BI projects but the impact of the intelligence outputs can be just as significant as in large enterprises.

Excel is still likely the most-used “BI tool” overall for any size business, and is likely the one that mid-sized and smaller businesses look to first for an approach to BI. Obviously Excel is a seemingly cheap solution for smaller companies, but frequently it is misused and underused by these constituencies for BI purposes. With Excel acting as the default BI tool, it’s harder to determine adoption rates for mid-sized and smaller companies simply because they already own it, if they are using Microsoft Office.

A major hurdle for connecting SMB companies with the right BI and analytics solutions is this: most software vendors tend to treat “SMB” as one big blob of a market, failing to derive in-depth segmentation to reflect the extreme differences from the lowest tier of the SMB landscape to the highest. When most software vendors say “SMB” they really mean “M” – the Mid-Market – and even then, they usually mean the upper tier of the mid-market: companies with revenues of $500M - $999M. To be successful with BI and analytics offerings that work well for mid-sized companies (or smaller), software vendors must more correctly understand the diverse small and mid-market segments to uncover real needs, goals and constraints.

Constraints for mid-sized and smaller companies adopting BI tools include small to non-existent IT staff. So any BI solution that requires a lot of IT intervention is not feasible. There are self-serve analytics solutions coming along which might be good fits for “SMB” – these solutions are frequently based on SaaS / Cloud services (which can make such analytics services more affordable). For smaller companies to benefit from such services, it is essential that the following are provided by the analytics offerings: self-serve training, extensive guidelines for using services, project templates, overviews of basic analytical thinking and asking the right questions, problem-solving with BI /analytics.

There are other gaps for mid-sized and smaller companies: how to use the right data sources, how to understand where the data is and how to access it, how to understand if the right answers are returned by BI and analytics projects. Also important: knowing how to corroborate and validate results with other data points and research. Data quality and the right data integrations are big issues for mid-sized and smaller companies, just as they are for larger enterprises.

Frankly there are many large enterprises that still have not figured out how to get the most from BI and analytics – and there is a significant failure rate for BI implementations in large enterprises where considerable time and money have been spent. It may be through vendor efforts to provide simpler approaches for BI and analytics as good fits for mid-sized and smaller businesses, that overall BI / analytics solutions will improve for all enterprises.

At a simplistic level, BI turns data into actionable information. While that is simplistic, it's a point with many implications that all too often are missed when companies become enamored of implementing "BI".
The implications are several, of which some are the following:
1) There has to be enough data available so it can't be readily understood without assistance. There's no point in implementing a BI if the data volumes are so small a person can look at the data and reach accurate conclusions.
2) The BI portion is only as good as the quality of the data. If there's not a way to ensure the data is accurate and sufficiently complete, the BI is useless. Worse yet, the business may draw incorrect conclusions and thus be worse off than with no BI.
3) Investments in BI technology tend to be multiple and large. The data has to be gathered, cleansed and/or normalized, and loaded into a database in an appropriate form. Tools must be deployed to extract the data from the database and display it in a form that is useful to the company. All this tends to be rather expensive and complex, although some of the packages available today greatly reduce the cost and complexity.
4) The whole thing is useless unless the company is prepared to make decisions and change direction based on what is found from the data.

Overall, BI can be a key differentiator for a company if the business model is appropriate and the BI selection, deployment and usage is handled the right way. This is all very difficult for SMBs. Many large enterprises struggle with these challenges.

So the conclusion has to be that SMBs should only consider investing in BI if they can address at least the above minimal list.

In general, all organizations require BI at some level. Wayne's comment above about all companies using BI at some level is very accurate. The issue though is whether the use of spreadsheets and other general tools provide the most value to organizations.

One of the issues facing many SMBs is the fact that many internal solutions are either homegrown or hosted externally to the organization. What this creates is an organization with many data sources in multiple formats. So when business decision makers try to get an accurate view of what is happening across the organization, they can't. Yes, they may be able to understand sales performance, or website traffic and general analytics, but tying in clickstream data to the lifetime value of a customer and how they interact with the organization may be out of reach.

What BI offers SMBs is a way to consolidate disparate data sources to create a consolidated and broader view of how the organization is performing, what gaps exist, and hopefully help provide answers to how to make things more efficient.

In terms of SMB adoption rates, I do think that most organizations use BI in some format, but that SMBs are starting to adopt more traditional BI methods (in essence, moving beyond Excel and spreadsheet use). Unfortunately, a true understanding of the market and what options exist for SMBs is still lacking on a broad level because much of the content available is either marketing hype, or targets organizations familiar with BI use.

The most important factors in judging what BI tooling, if any, an SMB requires are the complexity and scale of its base data and the size of its decision-making unit. Measures such as revenue, profit or number of employees have far less relevance.

The base data of a busienss today can be far more complex and extensive than in the past, because the web has made far larger volumes of a wider variety of data types available to even the smallest businesses. In a traditional business, decision making requires consolidation, cleansing and historicization only of internally-generated, operational data. The BI needs of such an SMB depends on the size of its operational data and, principally, on the number of sources. Where there is only one or two data sources, cleansing and consolidation needs are limited and BI tooling choice focuses solely on the analysis and reporting needs of the users.

As other respondents have noted, Excel is often the default and de facto BI tool in many SMBs - simply because it's there, it's familiar and has a lot of basic function to allow users to play with data. A newer class of BI tools with more analytical and visualization function and the ability to handle larger data sets, such as QlikView, Tableau and Lyza, has been gaining traction recently.

SMBs dealing with extensive web-sourced data often have very specific cleansing and consolidation needs, and use open source approaches, often based on Hadoop.

As mentioned above, the size of the decision-making unit is the other key factor in BI tool selection. If decisions are typically taken by a very small management team, BI tooling seldom goes beyond using Microsoft Office.

I came in late to this discussion, so most of what I can say has been said by others, particularly my friends Lyndsay Wise and Barry Devlin. I would add however, that SMB's are a category for marketing and not a true category. If "small" can include a distribution company with $3,000,000 in sales and 7 employees to one with $500,000,000 in sales and 1200 employees, it's a little difficult to generalize (Wayne alluded to this too).

I've worked with extremely large companies that purchased 1000's of seats of BI software, but in reality, only a dozen people actually use the tools while the rest just viewed output, at best. Many will say that an SMB is better served with one of the NextGen BI products, either a SaaS provider, a pure-play or a commercial open source product because the price point is lower. But if you can wrangle a good deal from one of the BigBI companies for only that handful of users who will actually use the tool, it might be as good a buy, just forget about dreams of "pervasive" or "self service" BI.

Echoing Barry again, a "small" company may have very big data requirements, especially if it's business is all or even partially deriving revenue from data. One of my clients has 40 employees and about $20 million in revenue, but they process over a terabyte of data per day. That makes them a large company in this context. On the other hand, I had a client that manufactured gas-fired turbines for electric power generation and had revenue of about $9 billion/year, but their entire data warehouse was only about 50 GB (not a typo). But when it comes to counting the beans, there is usually a more direct relationship between size of the company and complexity of the data and BI requirements.

One more point: whether you're an SMB or an elephant, classic BI failed to achieve its mission (in most cases) because it declared victory at the level of informing people. That wasn't enough. Newer generations of BI are showing real promise in terms of usability (which means relevance in integration with other work, not pretty GUI's), collaborative decision-making, visualization, gamification and iterative scenario analysis. The "users" of these tools are increasingly people who are one or even two generations younger than those of us who started this party. They will not tolerate the quality of the legacy tools. Their experience is not in enterprise software, it's the consumer WEB, where things are engaging and actually work without a three-day training class.

Lacking from most (all?) discussions from BI is the word "model." It always starts and ends with data, but data is nothing more than the footprint or the shadow of some activity or event that already took place. To be useful for analysis, planning and decision-making, data has to be placed in the context of some model. Until now, those "models" have been the data models of the data warehouse, which are deficient for business decision-making. Instead, the locus for modeling, for organizations large and small, is the spreadsheet. Users have voted with their feet and walked away from BI because it doesn't provide the framework for relating and "what-if"-ing, and deriving other conclusions that are not already extracted, transformed and loaded for consumption. With all of the aforementioned desirable features I've seen in the NextGen BI products, this is still missing. And the irony is, we had it in the 70's and 80's with DSS tools before BI and data warehousing came along.

Even in the case of extremely small businesses, a strategic deployment of a Business Intelligence solution can have major impact on the growth and profitability of the company.

Having a clear view of the profitable customers, products, regions and market segments is fundamental to understand the causes and expand upon the successes.

Equally important is to find those customers, brands, markets, segments and competitors responsible for draining cash and quickly stop the bleeding.

Should we treat all customers equally? (If you say yes, I’d suggest reading “Angel Customers & Demon Customers”)

In a company that has more than thirty customers, the Pareto Principle will be evident. Looking inside the famous 80/20 rule one can find that five percent of the customers generate close to 50% of the profits while the bottom 50% of the customers generate only 5% of the total profit.

Many companies have a hard time identifying which customers belong to each group. A strategic deployment of Business Analytics provides the answers instantly.

Even when sales revenue is growing, management should be able to ask key questions and find the answers right away, so the train of thought isn’t interrupted:

Is revenue growing profitably? Where? How and Why?

Are we paying sales reps commissions for bringing in unprofitable tonnage?

Can we quickly tell whether the growth trend is just over last month, last quarter, same quarter last year or year to date?

How about the profit growth of the last 52 weeks compared to the previous 52 weeks? Is it really growing?

Is our growth accelerating or decelerating? How much?

How is our profit growth versus budget or business plan?

Where is it failing to meet objectives? Why?

Which competitors are threatening our business?

In what regions and market segments can we maximize our growth?

A BI solution is fundamental to find answers to the seven layers of WHY's in order to get to the root cause of issues. Being able to understand and correct these issues faster than the competition provides the company a competitive advantage regardless of how small the business is.

In the past deploying a BI solution was not affordable by small companies, not only due to licensing cost but also because the internal and external resources needed for set-up and maintenance.

This situation has changed though. Today, there are many new analytics applications, either on premise or in the cloud, that are powerful, user friendly and very cost effective; making them ideal to bring small companies to the forefront of the 21st century technology to become true analytic competitors. http://blog.strat-wise.com/2011/08/04/what-is-bi-30.aspx

Regards, Bill

I think it is more important for SMB's to adopt BI. Small businesses should be more adaptable to the environment. Any SMB that maintains an inventory, ships products to customers, and maintains lists of customers will need to have insight into their customer base, demographic and geographic data for their customers, which products are the most popular, which products are the most profitable.

Many applications that are used for SMB's may not have a full picture of this information because some key portions of the data will not be maintained within an application.

Providing a clear picture of this information is BI's goal.

I'm not sure the rate of adoption for question 1.

For question 2, the pentaho stack, (ETL, BI, Analytical tools, and dashboard development) is a great starting point for SMB's. Many SMB's choose open source tools to get started because they are trying to get their business off the ground inexpensively and effectively. Keeping with the Open Source philosophy and choosing tools like Pentaho will meet many needs very well.

The understanding is growing in the SMB space, but in my experience long term "big" Data warehouse initiatives and drawn out BI deployments will not be adopted quickly by SMB's.

Doug

Hmm, sorry about that, of course you are. I've known you longer than Lindsay or Barry, but it might be handy if you spelled my name correctly!

Sorry, though, I disagree. The answer is not to create "a more robust architecture for power users." They will always find a way. The question is, how do we assist the process and decision-making of ordinary human beings? I'm also scratching my head over "sandboxes" of Hadoop clusters. Those things are a bear to set up and maintain, are only inexpensive with respect to their scale, and (at the moment) take real programmers to operate.

Other than thart, I agree. LOL

Wayne, my friend, you've overlooked the need for people, even the little people, to be more than informed. The need to have seamless integration between historical data (defined as anything that happened up to a nanosecond ago) and direction for what to do next is the what 80% of people need. Your comments apply to the segment of the population engaged in analyzing the past, not those operating in the present, There is a huge gap here.

Where have you seen "once Hadoop is set up?" There may be a cluster in place, but the feeds and analyses are mostly custom, and require programmers. How expensive will that be? Where are you going to find them? When will we have high-level languages instead of PHP or Java? I'm not opposed to Hadoop, quite the contrary, but I see it as barely an edge technology for some niche analysis, and not for everyone. In fact, most companies won't develop Hadoop capabilities, they will outsource them from data aggregators much like pharmaceutical companies do with syndicated pharmacy data.