Welcome to the MediaGamma blog

The gap is widening between those with a handle on their AI strategy and those only scratching the surface of its potential.

Is AI the right solution for your business: a 2020 assessment guide

Posted by Victor Malachard

AI solution business

Is AI the right solution for your business: a 2020 assessment guide

AI adoption is on the rise at a rate of around 25% year-on-year. That’s an increase of approximately 270% in the last four years alone, and it’s a figure that set to continue rising as big data and tech potential continue to grow hand in hand. Still, despite this popularity, AI is far from a panacea for everything, as is evidenced by the fact that only 23% of businesses report using such solutions regularly. More disturbingly, a reported 85% if AI projects fail to deliver on their intended goals.  

There are a variety of reasons for AI project failure. But one contributing factor is undoubtedly the improper application of AI to projects for which it’s not suited. Whether this is your first project or an expansion, the first question you need to ask yourself is if AI is really the right answer, and if you have the support structure in place to sustain an AI-driven solution.   

Within this checklist, we’re going to discuss some of the reasons that AI integrations stall, and what that means for your business, and your AI focus, moving forward.  


1) Do you have the data you need?

AI and data collation go hand in hand. In fact, 97.2% of the companies currently investing are doing so with data in mind. AI is, undeniably, the fastest and potentially only way to deal with a Big Data landscape that’s set to grow by 1.7 megabytes per second through 2020. This isn’t rocket science, but you might not realise that data also impacts whether AI is the right solution for you in the first place.

Practical AI Guide

Training and implementing any intelligence model, from deep learning through to natural language processing and beyond, depends entirely on your datasets. Quality data is, in fact, the only possible way to develop comprehensive systems that can help rather than hinder your processes. 

On a basic level, a ‘quality dataset’ relies on crucial factors such as: 

  • Representation
  • Diversity
  • Balance
  • Exhaustion
  • Sufficiency

With these critical conditions in mind, consider your datasets and their capabilities by asking yourself key questions, including - 


Do you have enough data for AI solutions?

Businesses are dealing with untold amounts of data, but don’t underestimate just how much information is necessary to integrate AI or machine learning. The more data you’re able to use during the training stage and beyond, the better chance your implementations have of taking you far. As such, you must consider the data you have now, and your capabilities for obtaining more through data aggregators, strategic partnerships, or even the creation of proprietary data. 


Can you avoid data bias?

Biased data is poor quality and thus guaranteed to result in missing variables, inputs, and data paths. Making sure your data is representative is half of this battle, but you also need to verify and review, ensuring reliable AI outlooks and avoidance of data bias at all times.


What about variance?

Vital to understanding the data you’re considering, data variance matters for both avoiding bias and providing the diversity you need. Only by calculating a mean variance can you work out where your data is right now, and where it needs to be to make effective implementation possible. 


2. Is security at the forefront of your efforts?

While it’s hard to pin AI setbacks on any cause, security and processing worries seem to be at the forefront of countless stalling efforts. In fact, with Capital One recently spending as much as $150 million on losses for this exact reason, 49% of companies rate security risks as one of their top three tech worries. 

Data security has become an especially pressing concern since the introduction of GDPR across Europe back in May 2018 — placing data collation in the legal (as well as ethical) category for businesses. Companies are now forever having to consider lawful processing and collation, two things which don’t always sit well with AI as it stands. 

That’s not to say there’s no such thing as a legal data collection in an artificial landscape, but care does need to be taken, with clear plans for data security and processing required at every stage. Even slip-ups during the training process could see you facing hefty fines that simply don’t fit with your money-saving intentions. 

Largely, success here depends on your ability to foresee potential security setbacks where artificial intelligence comes into play, including: 

  • Undetected data loss
  • Malicious data breaches
  • Inadvertent exposure
  • Unexpected changes in cloud infrastructure

Any company intending to implement AI also needs to consider automated malware detection, secure AI domains, and integrations that work in compliance with all relevant regulations. If your current capabilities, budget or infrastructure don’t allow for this, AI could well end up costing more than you’re prepared to spend. 


3. Do you have well-defined AI-outcomes?

Ultimately, AI is only useful if it allows you to move closer to goals with trackable results. Outcomes such as simplified analysis, improved customer service, and automation in all its forms certainly fit this criterion. 

If, however, you attempt to tackle AI without knowing precisely what it stands to do for you, you’re set to lose money and productivity on a broad scale. Even if AI does provide trackable business goals, you may not be entirely aware of the benefits and changes that such integrations are pointing you towards. 

Instead of fitting AI outcomes to your current goals because you feel you should, it’s far better to be honest with yourself. Do you know, right now, what artificial implementations could bring to your work efforts? If not, then it might be best to leave alone until clearly defined goals and targets become more evident. 


4. Have you considered integration?

Even if you're aware of precisely what AI stands to bring to your processes, a failure to consider integration itself could still unravel your efforts. This often comes in at a close second to security for intelligence-based setbacks, with 50% of organisations currently struggling to integrate due to a lack of use cases on the market. 

The simple fact is that there aren’t many companies actually utilising full-scale AI. As such, integration frameworks are almost impossible to come by unless you consider the large-scale and likely expensive efforts of AI frontrunners like Google.

This is a problem considering that a poorly-executed plan could leave you struggling to integrate the data you’ve worked so hard to collate, or even failing to smoothly introduce programs into your workplace without downtime.

Luckily, as AI shifts towards more SMBs, integrations using legacy systems, skills, and control will become far easier to understand and implement. This is also an area where AI specialists and consultants can help. Assessing your in-house skills is the next step on this checklist. However, it’s worth noting that even if you have the in-house capability to procure and deploy an out-of-the-box AI solution, you may need help integrating that with your broader data ecosystem. That kind of hybrid approach is proving a powerful solution for businesses of all sizes. But, fundamentally, make sure you’ve developed a clear integration process from the start to avoid setbacks down the line. 


5. Do you have access to the necessary skills?

If you’ve not integrated AI until now, there’s a high chance you don’t currently have the skills needed to train, model, and implement reliable solutions. That needn’t be the setback you might expect thanks to pre-trained third-party APIs and cloud-based machine learning-as-a-service (MLaaS). However, long-term implementations here still do rely on training and experience for execution. So, you do need to plan for some level of access to AI expertise. 

For the most part, companies face two main choices when it comes to operating AI solutions- 

  • In-house teams

If you have £250,000 to £500,000 to spare in your annual budget, then an in-house team could be your best chance for tailored and on-site expertise. The amount of AI jobs listed on Indeed grew by 29% in the last year alone to account for this rising need. And, joining the fray could see you gain access to top AI talent 24/7. Unfortunately, at this early stage, there is a risk that in-house efforts won’t be financially viable or even sensible, at least until your systems are more streamlined.

  • Consultancy

Competition, speed and cost are all reasons that many companies are turning to AI consultancies. Of senior AI professionals, 56% believe that an AI skills gap is the single biggest barrier to AI implementation across business operations. Consultancies provide a great alternative for businesses struggling to find the right talent in both the short-term and long-term. 

Typically offering services on a contractual or subscription basis, an on-hand consultancy ensures you access to trained AI experts at all times. This can be a great augmentation, even if you have an in-house team — particularly when it comes to very specialised sub-field of AI and machine learning. 

A quality consultancy will not only provide oversight to your AI integrations as they happen, they can also provide you with insight into processes, timeframes, and even profit margins that you’d struggle to reach alone. Note that you will need to ensure a consultancy is data-compliant as per the security concerns touched on above. But, if you choose the ideal consultancy to work with, this could be just the thing to get your AI off the ground at last. 


To AI or not to AI?

Ultimately, the decision about whether AI is right for your business can only land on your shoulders. While consultancies and data awareness can certainly point you in the right direction, you need to consider integration from every angle before you go ahead. Whatever you do, don’t let the growing marketing pressure to automated push you towards solutions you aren’t ready for. 

Once you’ve factored each of these key AI success factors, know that you won’t be missing out if you decide to wait, or integrate later into 2020. The knowledge you’ll glean from this checklist alone will, after all, help you to improve processes and fill gaps to ensure that, when you do turn to the artificial side, you do so with success and your bottom line in mind. Once you have picked a project, it’s time to start planning. 


Don't get left behind - unleash the power of AI. Get started by reading our practical guide to AI for real examples of AI in action. 

Download Guide

Victor Malachard
Victor Malachard
Executive Chairman

Related Posts

Google Privacy Sandbox: What it means for your programmatic ads

Google’s Privacy Sandbox looks like it might be the end of programmatic advertising as we know it, with third-party cookies largely set to become a thing of the past, and reliable automated data collation at risk along with them. The only way to prevent this from significantly unravelling online marketing efforts is to focus on shifting data landscapes, sources, and handling on a business-wide scale.

Is AI the right solution for your business: a 2020 assessment guide

AI adoption is on the rise at a rate of around 25% year-on-year. That’s an increase of approximately 270% in the last four years alone, and it’s a figure that set to continue rising as big data and tech potential continue to grow hand in hand. Still, despite this popularity, AI is far from a panacea for everything, as is evidenced by the fact that only 23% of businesses report using such solutions regularly. More disturbingly, a reported 85% if AI projects fail to deliver on their intended goals.  

How to train your own NLP model transformers

In recent years, search engines have led the way in natural language processing (NLP) and the benefits it stands to bring across business landscapes. Focused around machine learning capabilities, applications like Google’s BERT work to process natural language without human input. Used right, such systems can remove redundant work processes for faster and more efficient handling, analysis, and translation of written or spoken input from various sources.