DON’T LET TECH DEBT STAND IN THE WAY OF DATA-DRIVEN INSIGHT

Technical debt can put organizations in a headlock, both in the short and long term. Almost nothing casts a bigger, scarier shadow for decision makers — the perception of the time and cost necessary to overcome tech debt looms large, keeping entire companies frozen in fear.

This burden makes it difficult to efficiently extract insights from data. It’s a recipe for stifled growth.

We’re here with a message of hope: you don’t have to dive into a resource-intensive, five-year transformation project in order to manage your technical debt and realize your data potential. You have other options, and it all comes down to prioritization.

What is tech debt?

Technical debt is often defined as the cost incurred when you repeatedly choose short-term solutions rather than doing the (larger, more expensive) work of tackling the big-picture causes of your problems.

But let’s look at the issue through a different lens: what is the nature of technical debt?

Because new solutions are built and deployed every day, all organizations incur tech debt, to some degree, with every system and process implementation decision they make. Even if you implement a new, innovative solution today, there will be a better one available tomorrow. In this scenario, you will still incur tech debt — just less than an organization that makes no updates.

Too many organizations think of tech debt as a problem that can be permanently solved. In reality, it’s a constant that’s renewed continuously by change and growth, and trying to “solve” it completely is a futile pursuit.

While you can’t make tech debt vanish into thin air, you can certainly make it more manageable. If you focus on managing its impacts in an ongoing way, you can deflate its looming, monstrous reputation and get to work on making meaningful improvements in the here and now.

Is tech debt destroying your data-driven dreams?

Analytics bottlenecks are a common issue related to tech debt. Silos slow down the analytics process; if only one person knows where a spreadsheet is and how to extract meaningful data from it, they become the bottleneck.

With each short-term fix and siloed process, data becomes harder to manage, access, and analyze. In turn, drawing insights from that data requires more time and effort, the insights become less timely and less reliable, and informed decision making becomes more challenging.

In other words, tech debt has a way of draining value from data — and the longer you let that debt accrue, the more value you’re losing. Using a prioritized approach to managing tech debt can help you cover more ground right out of the gate, so you don’t lose any more value than you have to.

One way to apply this prioritized approach is with backlog grooming, the periodic process of reviewing and prioritizing backlog tasks (and removing unnecessary or outdated tasks).

How do you prioritize what areas to address?

There is a lot of information available on how to tackle tech debt. Unfortunately, most of it is theoretical. While the abstract stuff can be valuable, if you’re looking for a practical way forward, you need to bring your considerations back down to earth and fold in the business perspective to create a technical debt prioritization plan.

You probably have a lot of tools at your disposal, internally and externally, and resources to leverage. Take a look at what you need to have happen — not theoretically (e.g. eliminating all technical debt by some point in the future) — but actually.

For example, The Gunter Group recently worked with a retail automotive company that was struggling with data debt. It was impacting every area of their business, including employee satisfaction, customer satisfaction, and profit margin. They needed a new approach, but with such a vast problem, it was difficult for them to know where to start.

We worked with multiple teams within the company, including manufacturing, HR, marketing, and technology innovation to create a prioritization plan. High priority initiatives included redefining company-wide KPIs, designing and implementing machine learning algorithms, and improving data literacy across departments.

Though they still have a long way to go on their data maturity journey, this company was able to start making changes where it mattered most, rather than remaining paralyzed by the challenge ahead.

How we work with clients to tackle tech debt

Remediating data-related tech debt requires far more than just technical skills — it requires asking the right questions, gaining a holistic understanding of your organization’s business goals (as well as how they may vary across different departments), and creating a dialogue to explore possible solutions.

Each of these components requires a tremendous amount of time, which internal teams rarely have. In most cases, managing ongoing operational struggles takes priority over transformation, and team members don’t have the capacity to focus all their energy on addressing tech debt. Meanwhile, recruiting new team members is a time-consuming, resource-intensive process, and thanks to the tech talent shortage, it’s more challenging than ever.

Turning to outside help can get the data transformation ball rolling without overwhelming internal teams or opening a can of recruitment worms.

At The Gunter Group, we leverage a multidisciplinary approach (technology, people, strategy, and execution) that enables us to see the long-term big picture while solving the highest-priority problems in the short term.

Combined with our extensive technical capabilities, this approach allows our clients to chip away at their technical debt and reclaim the value of their data as quickly as possible — without the burden of hiring a new team.

Conclusion

Think about a meaningful, specific problem you’re facing right now that’s rooted in technical debt, and what you would be able to accomplish if this problem was being managed proactively.

If you set your sights on eliminating tech debt across your entire organization, you’ll likely get caught up in a complex tangle of issues — and that one major problem that’s holding you back now will still be holding you back in six months.

To accelerate your progress, identify your most pressing issues, and reach out to expert help if you need it. With the right strategy and the right partner, you can mitigate tech debt and use your data to its fullest potential.

Is technical debt slowing you down? Discover how to improve your data infrastructure and decision making with workshops hosted by The Gunter Group.

WHY ISN’T YOUR DATA TRANSFORMATION PROGRAM TRANSFORMING ANYTHING?

It’s a familiar tale — you’ve embarked on a journey to transform the way your company uses data, but all you have to show for it is a lot of very ambitious documentation and hundreds of progress update meetings where time feels like it’s standing still.

Why isn’t your in-progress data transformation program producing tangible results? Here are three common reasons why your efforts might be failing to deliver:

1. Your approach isn’t incremental.

We recently wrote about data maturity models and how they can actually hinder data transformation if you don’t apply them correctly. To summarize, organizations tend to focus too much on the brass ring (reaching some proverbial data maturity nirvana) and not enough on the details (getting more value from data along the way).

Often, the goal is a total overhaul of how data is used from the top to bottom of the business, delivering unimaginable value to everyone, everywhere — an ambitious transformation that will take years to achieve. Even if you’ve had a few updates and helpful process changes along the way, most of the value remains locked behind the “project completion” door.

What does that mean for those who are trying to get some incremental improvement and use data more effectively in the near term? It means they have to just sit tight and wait for the big reveal — otherwise, they’ll be doing work that’s not aligned with the broader strategy.

It’s important to consider the folks in your organization who use data to complete everyday tasks, and how you can make their jobs easier, not harder, during the data transformation process.

2. Technical debt is handcuffing you.

Individual innovation can be a huge asset for an organization. You want people to take initiative, solve problems, and make it happen. But when it comes to managing data, all that individual problem solving can add up to huge technical debt, and make organization-wide data transformation efforts feel insurmountable.

When everyone has their own unique, siloed quick fixes and band-aid solutions, there’s no more unifying structure left to transform — so where do you even start? Think of the chaos you’d unleash by pulling at even one loose thread; mission-critical systems that were crudely patched together would come tumbling down.

As an additional challenge, there will always be responsibilities and processes that can’t be switched off while things are fine tuned. For example, if you’re using your existing data systems to calculate monthly commissions, that presents a huge barrier to making any meaningful changes — you can’t stop paying out commissions until changes are implemented, and it would be risky to just cross your fingers and hope that brand new systems produce accurate results in time for you to cut checks.

Getting out from under technical debt isn’t impossible — it just requires a lot of effort. It’s also infinitely easier to achieve with the right combination of planning, skill, and resources, which is why many companies partner with firms that specialize in tackling technical debt.

It’s also important to remember that tech debt isn’t something you can escape completely, but a challenge that needs to be continually addressed. Good solution design assumes that today’s new solution is tomorrow’s tech debt, everything is eventually deprecated, and all solutions incur maintenance and upkeep. Finding the balance between standardization and innovation is important, as too much of either can be stifling.

3. You don’t have the right people.

Pulling off a successful data transformation project takes a lot of skill sets (strategic planning, project management, change management, ETL development, data science, etc.), and it’s highly unlikely that you’re going to find a single person who can do them all well. It’s also unrealistic to expect your current staff to sustain day-to-day operations and magically find the time to make changes. In 99% of cases, you’ll need outside help.

Trying to build an in-house team is difficult, especially in the midst of a data analytics talent shortage. Planning and recruiting for a team can take anywhere from six months to a year, plus the time it takes to get everyone up and running. If your organizational culture (or even just a particular decision maker) is change-resistant, this process can take even longer — and feel like pulling teeth.

Technical leaders often have to spend a large amount of time jockeying for resources and pitching projects. You may have to argue and advocate for three months to hire one person that was needed to solve a problem three months ago. By the time you get a decision, you’re going through the same cycle with a different problem.

It’s easy to see how years can go by with little to no progress.

This is another situation where it can be helpful to work with a data transformation partner, which gives you access to highly skilled, experienced people, without the pressure or long timeline of trying to build an in-house team.

Conclusion

Our general advice for data transformation can be summed up in two words: don’t wait. Every day you remain stalled out, you acquire more technical debt, and your problems get more complex. You lose more business, you gain less on your competition, and your company loses value.

If you need to see real outcomes from a data transformation project, focus on incremental improvements, and find a partner who can help you overcome the challenges we’ve outlined here.

Above all, remember that data transformation isn’t a “project” that begins and ends. It’s a decision to become a data-oriented organization — and it takes continuous effort and agility.

Not seeing results from your data transformation initiatives? Discover how to improve your data infrastructure and decision making during our upcoming Data Maturity workshop on December 15th at 11:30am Pacific Time.

A QUICK AND DIRTY GUIDE
TO DATA MATURITY

You’re probably familiar with the concept of data maturity — a measurement of how well an organization uses its data — and the maturity models that go along with it.

Understanding your current level of data maturity is the first step toward improving it. But the way you interpret and apply data maturity models might actually be hindering your success.

It’s tempting to look at a maturity model as a straightforward tiered system — your organization exists in one category, and your goal is to move up to the most mature category. But particularly for larger organizations, things aren’t so clear-cut. It’s common for different departments or business units to be at different maturity levels. Additionally, too many organizations get caught up in the long-term goal of data maturity and miss out on opportunities to create value at every maturity level.

In this article, we’ll go over the four categories we use to assess data maturity, give examples of challenges that arise in each category, and offer recommendations for driving more value at that level.

Before we get started, here’s a summary of each category for context:

Note: As you consider where you fall, it’s important to be realistic — think about where you are right now, not where you hope to be after completing a particular project. It’s also helpful to remember that it’s rare for an entire organization to exist within a single category. In most cases, different departments and teams will have different maturity levels.

Siloed

What it looks like:
Point-in-time data is manually exported from various applications and pulled into spreadsheets on an ad hoc basis. Team members compile and analyze data in their own individual workspaces (e.g. a spreadsheet only you have access to, or a platform that’s only used by one particular team).

Finding data requires some exploration — team members are not always sure where to look or who to ask for the data they need. They often end up emailing or asking around until they find someone who has what they’re looking for.

Example of a challenge:
You’re planning to hire people for a new team, and want to use a data-driven recruiting approach. You need access to key data about past recruiting efforts, like cost per hire, time to fill, recruiting yield ratios, and first-year attrition, so you send an email to HR requesting the information. HR then needs to take the time to locate and compile the data into a spreadsheet and send it to you.

Because you don’t have immediate access to the data you need, the recruiting process is already slowed down, and it will take longer to get your new team filled.

How to drive more value:
The importance of business process management can’t be overstated. For a process that will undoubtedly need to be repeated in the future (like looking at recruiting data), it’s important to start out with a deep, thorough understanding of the process itself. The first step to creating a repeatable environment is understanding what you want to repeat and why — then you can work on implementing more efficient processes.

Standardized

What it looks like:
Your organization has standard, scheduled operational reporting. You don’t have to go on an expedition every time you need data, because you know that you can find it in the most recent report (however, you’re still manually pulling data from that report). Even if you’re creating useful insights, they tend to stay with you — they’re not shared with other teams or business units.

Example of a challenge:
Each week, everyone in your department receives an email with an automated report that covers POS data. Most of the time, you don’t find anything in the report that’s actionable for you, so you sometimes don’t even open it.

When you do read it, you have to comb through tons of data to find anything relevant and manually extract it, which is time-consuming and inefficient. You complete analyses that are important to you within your own workspace, and the results aren’t usually shared outside of your immediate department.

How to drive more value:
Once you know where to find your data, determine what makes it actionable and relevant — and who needs to see it. As a starting point, ask yourself the following questions:

• What decisions do I need to make, and how can this data inform those decisions?
• What’s the ideal way for me to receive and view this data? 
• Who else would benefit from seeing this data?
• How can data be efficiently shared across departments so everyone has access to what they need?

Enterprised

What it looks like:
When a particular threshold that’s relevant to your job is met, you automatically receive the relevant information. Rather than getting a standardized report that may or may not contain information you care about, you get notified only when there is data pertinent to a decision you need to make or action you need to take — whether that data is related to a budget milestone, warehouse stock, page views, or something else.

Sources of data are aggregated, so when you need insights, you don’t have to manually combine and analyze data. However, the reporting is still fundamentally historical and often comes too late to help you make informed decisions.

Example of a challenge:
You’re responsible for mapping out an upcoming seasonal campaign and need sales data to inform your plan. Each week, you receive an automatic report containing data that helps you shape the campaign, but you have no way of looking forward in time — you’re stuck with a decision-making process that is reactive, not proactive.

How to drive more value:
With the right timing, data can be leveraged to achieve better outcomes (and predict possible future outcomes). Don’t just think about what data you need and where to find it — think about when it will have the biggest impact on decision-making, and how it can help you course correct before things get off track.

Actualized

What it looks like:
Your data is set up to model and predict future outcomes, perhaps using AI tools like predictive analytics and decision algorithms. Analysts and data scientists are an integral part of the business vision, and insights are created by the company (not requested by specific business leaders).

For many companies in this realm, the data model is inextricably linked with the product or service they provide. For example, platforms like Netflix and Spotify are rooted in predictive data analytics and the ability to make personalized recommendations to customers.

Very few companies reach this level of data maturity — and the reality is, not every company needs to. It takes continuous investment to maintain an Actualized data system and depending on the products and/or services you provide, it might not deliver enough ROI to justify spending the resources and effort.

Example of a challenge:
Your organization is experiencing a slight decrease in customer retention. You already have access to real-time data and analytics that help you understand the problem, but you want to leverage that data in new ways in order to make more informed decisions.

How to drive more value:
Being an Actualized organization doesn’t mean you’ve crossed the data maturity finish line — in fact, there is no finish line.

Only the most advanced companies make it to this level, which means competition is stiff. And considering the incredibly fast pace of data technology, companies that don’t continuously innovate will be left behind.

At this stage, optimization is key. Consider how data can better drive decisions and open up new opportunities for your organization. In the case of the challenge above, building new algorithms for churn models could help guide decision-making and reveal more actionable data.

Conclusion

Identifying your current data maturity level and setting goals for improvement is all well and good, but without taking steps to get more value from your data at your current level, your long-term progress may stall out.

Rather than thinking of data maturity models as rigid paths with set destinations, use them as way-finding tools. Once you understand where you are, you can move forward — no matter what “forward” looks like for your organization.

Regardless of your maturity level, we can help you get more value from your data. Discover how to improve your data infrastructure and decision-making with our Data Maturity Assessment.

SURVEYING THE DATA LANDSCAPE IN 2022

In the past, data wasn’t necessarily important to every person within a company. It was used primarily by analysts, accountants, and other specialists.

But in 2022, companies are learning that becoming a data-driven organization means incorporating data into every aspect of their business — from talent management to customer engagement and beyond — and continuously optimizing how they use data with new innovations and process improvements.

What does being data-driven look like in action? Here’s an example: a west coast retail automotive company employing over 7,000 people across 9 states came to us with the goal of implementing a mixture of data science and machine learning to identify, implement, and improve safety, employee satisfaction, customer satisfaction, and profit margin. The client asked us to work with multiple teams within manufacturing, HR, marketing, and technology innovation to build out the desired capability.

To help this company reach their goals, we provided high-level strategic insight for new initiatives, built out proof of concepts, made recommendations for innovative methodologies, designed machine learning algorithms, helped them redefine company-wide KPIs, and trained their staff on new processes.

As a result, executives are better able to make key strategic decisions and further company goals based on data-driven insights, and the entire organization’s data literacy has improved.

A shifting mindset

A few years ago, the goal for many companies was “fixing” their data processes (a reactive way of looking at data management), with a focus that was often confined to specific departments. In 2022, most organizations are approaching data management differently. They’re aiming to be far more proactive — and to stay competitive, they have to be.

It’s less about simply “cleaning up” messy data, and more about creating meaningful, long-lasting, company-wide change that will continue to drive value and inform decision making in the future. In other words, it’s all about becoming data driven across the board.

Here’s an infographic that breaks down this change in mindset and some common challenges that are forcing companies to rethink the way they approach data:

Approaching data reactively and in silos is a way of the past. To keep up with the intense pace of change, constant innovation, and evolving customer expectations in 2022, a proactive, holistic, organization-wide strategy is required.

This change is positive on multiple levels. It’s not just good for staying competitive — it’s also a way to ensure that each of the common challenges described above (talent optimization, business insights, technical debt, etc.) get addressed so you can reap the benefits of becoming a data-driven organization.

That said, embarking on a large data transformation project can sometimes feel impossible, especially if you can’t promise ROI until months (or years) down the road. At The Gunter Group, we believe in taking a different, more iterative approach that enables organizations to realize immediate value while still keeping their larger goals — and the overall data landscape — in mind.

Ready to reframe the way your organization thinks about data? Talk to the experts at The Gunter Group.

What is tech debt?

Technical debt is often defined as the cost incurred when you repeatedly choose short-term solutions rather than doing the (larger, more expensive) work of tackling the big-picture causes of your problems.

But let’s look at the issue through a different lens: what is the nature of technical debt?

Because new solutions are built and deployed every day, all organizations incur tech debt, to some degree, with every system and process implementation decision they make. Even if you implement a new, innovative solution today, there will be a better one available tomorrow. In this scenario, you will still incur tech debt — just less than an organization that makes no updates.

Too many organizations think of tech debt as a problem that can be permanently solved. In reality, it’s a constant that’s renewed continuously by change and growth, and trying to “solve” it completely is a futile pursuit.

While you can’t make tech debt vanish into thin air, you can certainly make it more manageable. If you focus on managing its impacts in an ongoing way, you can deflate its looming, monstrous reputation and get to work on making meaningful improvements in the here and now.

Is tech debt destroying your data-driven dreams?

Analytics bottlenecks are a common issue related to tech debt. Silos slow down the analytics process; if only one person knows where a spreadsheet is and how to extract meaningful data from it, they become the bottleneck.

With each short-term fix and siloed process, data becomes harder to manage, access, and analyze. In turn, drawing insights from that data requires more time and effort, the insights become less timely and less reliable, and informed decision making becomes more challenging.

In other words, tech debt has a way of draining value from data — and the longer you let that debt accrue, the more value you’re losing. Using a prioritized approach to managing tech debt can help you cover more ground right out of the gate, so you don’t lose any more value than you have to.

One way to apply this prioritized approach is with backlog grooming, the periodic process of reviewing and prioritizing backlog tasks (and removing unnecessary or outdated tasks).

How do you prioritize what areas to address?

There is a lot of information available on how to tackle tech debt. Unfortunately, most of it is theoretical. While the abstract stuff can be valuable, if you’re looking for a practical way forward, you need to bring your considerations back down to earth and fold in the business perspective to create a technical debt prioritization plan.

You probably have a lot of tools at your disposal, internally and externally, and resources to leverage. Take a look at what you need to have happen — not theoretically (e.g. eliminating all technical debt by some point in the future) — but actually.

For example, The Gunter Group recently worked with a retail automotive company that was struggling with data debt. It was impacting every area of their business, including employee satisfaction, customer satisfaction, and profit margin. They needed a new approach, but with such a vast problem, it was difficult for them to know where to start.

We worked with multiple teams within the company, including manufacturing, HR, marketing, and technology innovation to create a prioritization plan. High priority initiatives included redefining company-wide KPIs, designing and implementing machine learning algorithms, and improving data literacy across departments.

Though they still have a long way to go on their data maturity journey, this company was able to start making changes where it mattered most, rather than remaining paralyzed by the challenge ahead.

How we work with clients to tackle tech debt

Remediating data-related tech debt requires far more than just technical skills — it requires asking the right questions, gaining a holistic understanding of your organization’s business goals (as well as how they may vary across different departments), and creating a dialogue to explore possible solutions.

Each of these components requires a tremendous amount of time, which internal teams rarely have. In most cases, managing ongoing operational struggles takes priority over transformation, and team members don’t have the capacity to focus all their energy on addressing tech debt. Meanwhile, recruiting new team members is a time-consuming, resource-intensive process, and thanks to the tech talent shortage, it’s more challenging than ever.

Turning to outside help can get the data transformation ball rolling without overwhelming internal teams or opening a can of recruitment worms.

At The Gunter Group, we leverage a multidisciplinary approach (technology, people, strategy, and execution) that enables us to see the long-term big picture while solving the highest-priority problems in the short term.

Combined with our extensive technical capabilities, this approach allows our clients to chip away at their technical debt and reclaim the value of their data as quickly as possible — without the burden of hiring a new team.

Conclusion

Think about a meaningful, specific problem you’re facing right now that’s rooted in technical debt, and what you would be able to accomplish if this problem was being managed proactively.

If you set your sights on eliminating tech debt across your entire organization, you’ll likely get caught up in a complex tangle of issues — and that one major problem that’s holding you back now will still be holding you back in six months.

To accelerate your progress, identify your most pressing issues, and reach out to expert help if you need it. With the right strategy and the right partner, you can mitigate tech debt and use your data to its fullest potential.

Is technical debt slowing you down? Discover how to improve your data infrastructure and decision making with workshops hosted by The Gunter Group.

ARTIFICIAL INTELLIGENCE & YOUR BUSINESS: 3 THINGS TO KNOW

For Starters: This is Not Skynet

Artificial intelligence is all around you. You have been using it for a while, probably without even knowing it. Gmail finishes your sentences. Your phone corrects your spelling and grammar. Instagram decides what to show you next. Spotify creates perfect playlists of new music. Advertisements know exactly what you’re thinking. You use AI hundreds of times a day.

For some of us, this is an uneasy truth at first glance. We imagine computers ruling our world with cold efficiency, slowly robbing us of our freedoms. But AI is not the villain from our favorite dystopian movies. As fun as it is to get lost in the world of Terminator‘s Skynet, I,Robot‘s VIKI, or Captain Marvel‘s Supreme Intelligence, AI is far less sinister in real life. 

AI is a tool, helping to solve problems that require enormous computing power. It’s lines of code that process millions of haystacks worth of data to pull out a single needle in a matter of seconds. 

The point: AI is everywhere, and it’s not the far-off villain of Isaac Asimov horror fiction. AI is a tool that is seamlessly integrated into hundreds of your daily experiences. It’s not just for nerds anymore. 

Especially in business, there are a few things you should know about this tool if you expect to remain competitive in the coming decade.

3 Things You Need to Know:

(1) AI is Now a Commodity

Until recently, artificial intelligence was mostly the subject of science fiction writers; today it is the subject of your average software engineer. The application of AI has come a long way.

The business community has witnessed an integration of AI on a grand scale. Ubiquitous in all markets, it is written into many of the functions that we use on a daily basis. Furthermore, companies like Amazon and Google have used unimaginably large collections of data to perfect AI tools, and have commoditized them in the form of products like AWS and Google Cloud.

Some have chosen to ignore AI, not seeing value in tools they can barely understand. Meanwhile, fields that lean heavily on AI (like data analysis and business intelligence) have expanded rapidly in recent years. For example, CIO.com lists “BI Analyst” as the most in-demand tech job of 2019. AI is changing business.

A great example comes from an interview with the Harvard Business Review, MIT Sloan School professor Erik Brynjolfsson. He describes an AI program that reviewed recorded conversations of successful sales, and then listened in on active conversations between salespeople and potential customers. While they were on a sales call, the program used the data from successful pitches to make suggestions about words or phrases that the sales person could slip into their conversation to help close the sale. This small application of AI boosted sales by 50%.

Brynjolfsson strongly believes that the only thing holding businesses back is a lack of imagination by business executives on how to use these new tools in their businesses.

(2) Your Competitors Are Using AI

Even if you have a few data analysts on staff, you’re most likely not getting the most out of your software. Since AI is everywhere, it’s hard for CIO’s, tech leads, or business owners to find and use the full range of the tools that are available to them. For instance:

your CRM could be generating leads for your sales department in places they wouldn’t have thought to look

your supply chain solution could be dramatically cutting waste by ordering supplies to be delivered for the lowest shipping cost at the exact moment they are needed  

your security solution could be identifying fraud and malware threats before they strike, saving you the time and money you would have spent recovering from one employee clicking one email

your ERP could be spotting spending trends and suggesting campaigns to your marketing team 

You might ask yourself, Is it worth all the hassle? Do I really need to do all this? I’m getting along fine without AI, why would I change? If you’re asking yourself this question, you’re looking backwards, not forwards. Failing to make the most of AI is not just a missed opportunity; failing to utilize AI is an increasingly significant liability. 

The proof is in the numbers. Netflix claims that a machine learning tool saves it $1 billion a year. Amazon used AI to influence the decision-time of online shoppers and cut it by more than a third. HBR found that companies using AI for sales were able to increase their leads by more than 50%, reduce call time by 60%, and realize cost reductions of 40%. If you don’t take advantage of AI, you will lose out to someone who is.

AI is now a necessity; it’s simply integrated into everything you do. Your CRM, ERP, website, and applications are all using AI. If you aren’t making the most of it, then this low-hanging fruit is spoiling inches from your hand. That is, if it’s not being snatched up by your competition.

(3) The Catch: It’s Not Magic

AI is certainly low-hanging fruit, and it doesn’t take an enormous investment to get more out of it. But it’s not a magic solution that will fix everything. AI is a complicated tool, and getting the most out of it requires knowing how to use it. Utilizing AI takes work. And worse, if you don’t use it correctly, then AI could actually lead you in the wrong direction. Ray Dalio put it best, “Be cautious about trusting AI without deep understanding.”

AI is a tool, and just like any tool it can be used improperly. With AI, bad input means bad output. There’s an art to using this tool.

Here’s a simple illustration. At one time or another, most of us have used the online radio service, Pandora. The process is simple. Tell Pandora a song or an artist that you like, and it searches an enormous music library to play a song that is similar to your input. You rate the suggestion in order to help Pandora hone in on your taste. This is AI at work, learning from your preferences.

But a tool is only as good as its users. If you vote thumbs down on your favorite song, then Pandora won’t play it again. Or if Pandora hadn’t invested in a large and diverse enough music library, it wouldn’t be able to return songs similar to the ones you like. The tool needs to be used properly in order for you to get the most of it. 

AI solutions in business are no different: you need to use the tool properly in order for it to work properly.

So How Do I Do it Right?

There are three main components of a good AI implementation in business: know yourself, know what you need, and use the right data. If you don’t have all three of these components then at best you’re not getting the most out of AI, and at worst the tool will lead you in the wrong direction.

First: Know Yourself

An AI solution isn’t worth the investment if it doesn’t solve the specific problems facing your business. This makes sense in theory, but is hard for executive leaders to get right in practice. 

The reason for this is not hard to grasp. CIO’s or VP’s of Sales have deep knowledge of their own departments and the business verticals relevant to them, but good tech integrations require organization-wide implementation, and this always pushes beyond the knowledge of a particular individual or department. It’s hard to see beyond the boundaries of your silo.

We begin every project with a current-state assessment. This seems like a logical first step, but it’s often overlooked. It involves gathering requirements that clarify the current-state needs and processes that are affected by a solution. This gives you a clearer understanding of what you need in the future. Many executives assume they already know this, but even the best leaders have blind spots. 

A current-state assessment is the best starting point for any kind of project work, but it is especially important with AI. If you don’t have a crystal clear understanding of what you need from an AI solution, then all that will change is the speed in which you receive unusable or incorrect answers to your business problems. 

A worthwhile software integration must always begin with a careful look inward, with an up-to-date assessment of requirements gathering and process mapping. Failing to do this has its consequences. If AI is integrated into an organization’s workflow without this look at your current-state, the result is solutions that don’t fit your business or market. 

Second: Use the Correct Inputs

What sets real-life AI apart from fictional AI is one key aspect: general intelligence.  AI can solve some problems faster and better than humans, but it can’t think for itself. 

For example, AI programs have bested world champions in Chess, Go, Texas Hold’Em, and Jeopardy!. But there’s an important detail: the same AI that beat champions in chess can’t even play the game of Texas Hold’Em. Another example: an AI program has to sample tens of thousands of photos before it can identify animal pictures with any reliability, whereas a 2-year-old can correctly identify cats after only seeing one example. 

But it’s not just games and image recognition: there are darker examples of AI falling short in big ways:

Developers at MIT were excited about the accuracy of their AI facial recognition software, until they realized that they forgot to build inputs into the software that could identify darker skin tones.

Biases built into AI solutions in law enforcement yielded inaccurate results with huge consequences, such as falsely singling out minorities for recidivism or counseling police to target ethnic neighborhoods. 

Amazon used an AI recruitment tool that spent 4 years sorting out the resumes of female applicants, even specifically flagging the word “women” as cause for downgrading a resume. 

AI tools are narrow, specialized solutions: you can’t expect to solve problems without teaching it how. It takes work to shape the tool to work correctly. Well-defined and clearly-articulated problems are inseparable from successful AI integrations. The payoff comes once a computer knows how to do a task properly, and can do it at a speed and volume that humans could never achieve. The good news: this work is absolutely within your reach, and most off-the-shelf software has easy-to-use feedback loops built in to help you!

Third: Use the Right Data

Imagine searching through a deck of playing cards to find the midday market report. Or searching through a 4-pack of crayons looking for an exact match to Robin’s Egg Blue. If your data set isn’t large enough or doesn’t fit your questions, then you aren’t going to find meaningful answers. This is especially true for artificial intelligence. 

This can be daunting for someone new to AI. How do I know if data is high-quality? How do I know if I have a sufficient quantity? Without the help of experienced input, executives might be making data purchases that are unhelpful, or even harmful. The consequence of using AI with insufficient or bad data is inaccurate solutions and misdirection. 

One Last Consideration: Don’t Reinvent the Wheel

Your business is unique, but your problem is not. Why spend time and money custom-fitting an AI solution to your business when a tool has already been developed for just that problem? Finding the right solution might just be a matter of having someone who knows the market helping you find the solution that fits your business.

Using Artificial Intelligence Well: A Case Study

A client of ours was experiencing stagnation in their financial and customer growth for the first time in their history, and couldn’t identify the reason for the slowing growth. They turned to The Gunter Group to help them revamp their digital strategy in order to expand to new customers.

This client had years of data on their customers that they didn’t know how to leverage. They offered great service, but they didn’t understand their customers’ behavior. So we started there.

We began with collecting their data, which consisted of several different types that needed to be aggregated into one system. We helped them build a unified repository, so that any insights they sought maximized the value of their data. In addition to helping them improve the quality of their data, we also helped them refine the insights they hoped to gather. At the beginning of the process, we engaged our experienced Business Analysts to help them integrate their knowledge of their organization’s structure and business goals into the process. 

With the 3 important ingredients in place, we were ready to make the most of an AI integration to explore the data. Our team helped craft complex algorithms to create customer segmentation, cohort development, churn prediction, and market share analysis. They were able to use these insights to launch highly effective marketing campaigns, and began a path to predictive analytics to enable real-time interventions in the future. 

This kind of example abounds in the business community today. Artificial intelligence is quickly becoming a commodity, available to all. You can’t afford to stay behind the curve. 

The Gunter Group partners with organizations in Portland, Vancouver, Bend, Salem, Reno, and Sacramento, helping them to know themselves and seize the low-hanging fruit of AI. If you are interested in learning how we can help you to do the same, reach out today! 

USING AN ARCHEOLOGICAL APPROACH TO DELIVER TRANSFORMATIONAL DATA ANALYSIS

Archeologists seek to understand human behavior through stuff – physical stuff. These artifacts of human existence are often ravaged by time, leaving the subtlest of clues for us to examine. Without context, archaeologists piece together the puzzle and hypothesize why humans of the past made certain decisions and how they interacted with each other and their environment.

As a former archaeologist, I’ve transitioned from the academic world into management consulting. At first glance, the two may seem unrelated but I’ve discovered that my background gives me a unique perspective and a surprisingly transferable skill set. I certainly know my way around a transected excavation pit and a trowel, but the skills I’m talking about are more akin to business intelligence, behavior science, and organizational development. Additionally, the academic discipline itself, like most in the liberal arts, provided an incredibly strong interdisciplinary education and honed my ability to learn new things and synthesize complex data quickly.

Archaeological education, training, fieldwork, and research commonly incorporate a wide range of disciplines, from botany and geology to sociology and economics. One day I might read a dense medical journal and the next study climate data, all while making novel connections to illuminate meaningful patterns. By now, I’m sure you can see how this would translate to a more corporate setting where business analysts, data scientists, and market strategists regularly flex the same muscles. What I love most about consulting is taking these insights and the scientific method and turning them into action so clients can better predict, understand, and adapt to their customers’ behavior.

For example, I recently helped a client understand how wait times impacted the customer experience and subsequent loyalty. Their assumption was that lower volume would reduce wait time thereby increasing volume again. I rolled up my sleeves to excavate their data and find out if this proved true. It turned out to be more complicated than that so I dug deeper to uncover the sweet spot at which wait times, volume, and customer loyalty could work in harmony to optimize results.

Such a dynamic situation can be difficult to articulate so I leveraged a commonly used framework developed by in-depth ecological research: the predator/prey model. When applied to the above business scenario, our analysis revealed a pattern that led us to actionable insights. We determined the retail equivalent of “ecosystem equilibrium” which indicated an ideal range for how long customers could be kept waiting without negatively impacting loyalty. We are currently testing our hypothesis of the maximum theoretical wait time in the real world. By engineering specific wait times, they will be able to collect additional data which will then be fed back into our model and drive continuous improvement.

Another recent example of archaeological expertise adapted to the business world comes from our work with a large retailer. This client was puzzled as to why one location showed double the sales compared to another only a mile away. Customer profiles, inventory, staffing, etc. were all so similar so why the disparity? 

For this scenario, I invoked another scientific model: least cost pathways. The least cost pathways model essentially helps us determine the path of least resistance using weighted costs associated with various routes. In archaeology, this model might help explain seemingly odd trade patterns or dietary choices. In a modern business scenario, the same approach can be used to map human movement, both physical and virtual, and optimize location of brick-and-mortar stores, button placement on a website, or product placement on a shelf. 

Through this lens, our analysis revealed that the under performing store required a U-turn and a left hand turn across four lanes of traffic. Meanwhile, the other store required only an easy right hand turn off a freeway entrance. Given this information, which location would you choose to do your shopping? Suddenly customer behavior made complete sense and store managers were able to direct their attention to the right issues and avoid unnecessary time and expense trying to solve the wrong problem. Going forward, executives have revised their strategy for key real estate and business decisions to incorporate these insights and avoid costly mistakes.

The potential of data analysis, especially in this age of “big data,” is immense but it’s important to use appropriate models to help explain that data and to ask the right questions in the first place. By thinking like an archeologist and working like a data scientist, I’m able to solve puzzles that save clients time and money.