Big decisions with big data

Featured Content

This article is courtesy of TechRepublic Premium. For more content like this, as well as a full library of ebooks and whitepapers, sign up for Premium today. Read more about it here.

Join Today

Big data means big decisions for companies as they consider whether to incorporate big data into their organization. Big data can also mean big dollars, or big failure. There are clear steps to take while building a case for, or against, big data.

There are big decisions to make with big data. Consider these two statistics: 70 percent of all change efforts fail; and projects with excellent organizational change management (OCM) average 143 percent return on expected value. The first statistic surfaced in the 1980s when the Standish Group started tracking success rates of IT projects.

Since then, you would hope things would have improved; however, that's not the case. A 2008 IBM study confirmed that the majority of change projects still fail. The second statistic comes from a McKinsey study that was published in 2002. Other top firms like Boston Consulting Group and Bain have similar findings. For a large company, a failed project is unfortunate, but life goes on. For a small business, a failed venture is devastating. That's why most small businesses don't make it to their third birthday.

I'm not trying to freak you out, or sell you on the idea of organizational change management. My point for juxtaposing the two radically different outcomes is to demonstrate that there are methods and strategies successful companies embrace that unsuccessful companies don't. Over my 20 years as a consultant, I've had the privilege of working with large companies and small, and some patterns have emerged. Once you have an idea of how big data will be used in your corporate strategy (core, supporting, or operational), you must clearly understand whether it makes business sense to proceed with your approach. Successful small businesses not only have great visions but they also make great decisions about if and how big data will grow their business.

Enjoying this article?

Download this article and thousands of whitepapers and ebooks from our Premium library. Enjoy expert IT analyst briefings and access to the top IT professionals, all in an ad-free experience.

Join Premium Today

Building a case for big data

You absolutely, unequivocally, undeniably, must build a business case for big data from the top down: Benefits to cost. I've seen far too many companies spend money on big data without a clear idea of how they're going to use it. There's so much hype around big data and misguided advice, it's no wonder people are confused. This confusion translates to irrational behavior that doesn't make any business sense. I implore you to do the opposite: build a rational business case first, and then spend money on big data if it makes sense.

A business case is not a panacea, but it will mitigate egregious strategic errors. When I consulted for Visa, part of my responsibility was to help build solid business cases for the 150 projects that were moving through its enterprise data strategy. Although all 150 projects were on the radar, the executives would not authorize any project that didn't have a strong financial case. This is a good practice that you should follow as well. A strong business case for big data will quantify your investment decision.

Your business case stems from your strategy. You'll need to determine the upside potential of your corporate strategy and big data's contribution to this upside. To determine your upside potential, compare what your profits would be if you launched your strategy against what they would be if you did nothing. The difference represents your strategy's potential value. For instance, let's say your profits this year are $10 million. If you do nothing, in the next three years, your profits are projected to be $11 million, $12 million and then, $13 million. However, if you launch your strategy, you profits are projected to be $11 million, $15 million, and $20 million. The potential value of your strategy then, is $10 million ($3 million difference in year two, plus $7 million difference in year three).

Next, decide the chances of your strategy's success. This tempers the potential value to something more realistic. Not all strategies work out, especially if they involve big data. I suggest a very conservative estimate of perhaps 50 percent or less. It may come as a shock to think that your strategy only has a 50 percent chance of success; however, this is the reality. Expect the best but plan for the worst. In our example, if we adjust our potential value of $10 million with a 50 percent chance of success, we arrive at an expected value of $5 million.

The last step is to determine big data's contribution to your strategy. Up until this point, big data has been implied; however not explicitly involved in any of the calculations. This is the point where big data becomes explicit. If you're using big data as your core strategy, big data's contribution is 100 percent. However, if you're using big data to support your core strategy, or to fortify the operational capabilities that power your strategy, big data's contribution will be something less.

At this point, don't think about cost; think about contribution to success. For instance, Progressive Insurance is in the business of selling insurance; however, they're using big data quite effectively to adjust premiums based on real-time driving data collected from sensors placed in their customers' cars. So, big data isn't their whole strategy, but it may contribute 65 percent to its success. Using this rationale, if the company in our previous example came up with the same percentage, the expected value of big data's contribution to its corporate strategy would be $3.25 million (65 percent of $5 million expected value). This gives us a rational place to start when considering an investment in big data. For instance, if you spent $1 million on big data, based on these numbers, you're looking at better than 3 to 1 return. That's not a bad decision if you ask me.

Strengthening your case

Now that you have a general idea of how your business case for big data should be built, let's explore some ways to make it stronger. You shouldn’t spend too much time in thinking and analysis without taking action; however, you shouldn’t spend too little either. On the rare occasion that I actually see a business case built for big data, it's not tenable. Instead, it's laced with wild assumptions that have weak or non-existent support. Large companies typically have strong governance to protect against weak business cases; however, as a small business leader, it's your responsibility to reinforce your business case with a little armor.

The biggest risk you run is in step one: projecting the profits from your strategy. Since profits are revenues less cost, everything hinges on your revenue projections. This is where small business owners tend to get careless. It's good to be optimistic about your ideas; however, it's dangerous to base financial decisions on hopes and dreams. I've even seen people reverse-engineer revenues from how much they want to spend on big data! Sorry, it doesn't work that way.

Regardless of your strategy, understand your market. This is the best advice I can offer to a small business owner. I've seen a lot of brilliant independent consultants forced into a salaried job because they didn't know how to market their talents. A small business is no different. It's great to have a big data insight; however, it's a completely different thing to have a big data insight that's vetted against your target market. Make sure your revenue projections are supported by some form of marketing analysis. This is easiest for those using big data to support a market-driven strategy—by definition they should already understand their market well. However, all companies must know who their target market is, and be able to make an educated guess as to what their customers will buy.

Another area of the business case to attack is step two: your chance of success. I recommend an extremely conservative chance of success because big data initiatives are inherently risky. That said, there are ways to improve your chances. If you take an agile approach and deliver tangible results from your strategy as soon as possible, not only will you increase your chances of success by clarifying unknowns in the marketplace, but you'll also accelerate your payback period, which is the amount of time it takes to recoup your investment. Practically, this is easier said than done; however, it's worth the extra effort.

Some other key things that increase your chance of success are scorecards and risk management. It's best to follow a data-driven approach, instrumenting your success with a scorecard. A scorecard first defines your success criteria, and then monitors your success during strategy execution. It also monitors key risks and assumptions. Big risks shouldn't come as a surprise. Your scorecard should give you adequate warning to make adjustments.

Big data shopping

Once you have a good idea of how much you can spend on big data, it's time to make a shopping list. Don't be disappointed or upset if everything on your shopping list adds up to more than you should spend. This is actually a blessing in disguise; your diligence in building your business case has just saved you from making a bad decision.

Obviously, you shouldn't spend more than the expected value of big data's contribution to your strategy; however, this is not your spending threshold. In our earlier example, we had a $3.25 million expected value of big data's contribution. If we then spent $3.25 million on big data, we would have no return on our investment. That's not too smart. In fact, your spending should only be a fraction of your expected contribution value—one tenth is ideal, but one third isn't bad. You really shouldn't exceed half of your expected contribution value. So in our example, if your big data shopping list came up to $2 million, I'd recommend you pass, even though you're staring at a vetted $10 million upside.

The biggest ticket on your shopping list will be talent, so I would concentrate my efforts there. Data scientists are not cheap; however, there are some cost-effective ways to find similar talent if necessary. The challenge with data scientists is that they have both analytic and programming skills. This is a little hard to find now; however, we should start seeing more talent enter the marketplace in the near future. In the meantime, I suggest looking for people that either have advanced degrees in some form of mathematics or physics. It's more likely that they'll have some programming experience, than finding a programmer that has advanced math skills. Another avenue you might try is out-of-work Six Sigma Black Belts. If they have a proven track record, they should have good analytical skills. And if they've also done some programming, you're in luck.

Unfortunately, you can't stop with just one data scientist. It may take a team of data scientists, and at a minimum, you'll also need analytic leadership, analytic management, and business experts (especially marketing). Ideally, you'll also add change leaders (remember the opening statistic?), team builders, governors, and quality specialists. Even if you have resources currently in place (including yourself) to fill these roles, do they really have the time to devote to building out a new strategy, while simultaneously keeping the business running? Probably not. A consultant(s) could be used; however, they'll cost more than a new employee. That said, a consultant might be more efficient, as their expertise is deeper and wider than a typical employee. You'll have to run the numbers and do what's best for your situation.

One final caution: don't cut costs just to cut costs. Balance costs to build a reasonable business case, but don't overdo it. You get what you pay for, and you don't want to skimp on talent if you can afford an expert. Like anything else, you take huge risks when you center your approach on reducing cost. Low quality talent is unpredictable and the last thing you need on a big data initiative is more uncertainty.

Conclusion

Launching a strategy that involves big data can be exciting, but it can also be fatal. The success rate of companies taking this journey swings wildly from dismal failure to wild success. It's not all random. It certainly takes some luck, but it also takes the ability to make smart decisions, especially if you're dealing with the limited resources of a small business. Before launching a strategy that includes big data, be sure your investment makes economic sense.

A good business case provides a rational explanation for your big data investment. By quantifying big data's contribution to your strategy, you extricate its real benefit in financial terms. This is the information required to make a wise investment decision: the bulk of which will be on big data talent. Of course, you get what you pay for, so make your decisions based on value, not cost.

I've walked you through the mechanics of building a strong business case for big data, so take some time today to put some rough numbers in place. It won't take very long and it will give you a good sense of whether to consider a business relationship with big data. If the numbers work out, then great, go for it. If the numbers don't work out, go back to the drawing board. Either way, pat yourself on the back for making a good decision. Continue making decisions like this and you're well on your way to success.

Read more about big data and the path to success with TechProResearch's article The big data rush: How small businesses can strike gold.


Join Premium Today