Main content

ERP - and the buying of it - has fundamentally changed. Here's how and why that matters to enterprises

Brian Sommer Profile picture for user brianssommer May 3, 2024
Summary:
As software evolves, so too must the methods used by buyers to pick an acceptable solution. The latest innovations/evolutions in application software should make every company re-examine how and what it is purchasing. Here’s a critical look at the future of software buying and how it’s getting harder to do well.

evolving

The old ways of selecting software are dying. Yep, time to break out the dark suits/dresses and get those condolence cards written. 

For some time, people compared one application software product to another based on the functions and features that each product possessed. That method was already passé while I was at Accenture and by the early 1990s, my team there and I re-wrote the firm’s methodology for selecting software. The new approach kept some of that function/feature focus but that was supplemented with descriptive problems or difficult business scenarios that the client wanted a vendor to solve.

Some software buyers added process demonstrations to the mix. In doing this, the goal shifted from understanding what function points were in a package to which vendor solved a prospect’s toughest, ugliest, costliest, most inefficient and/or most troublesome processes in the most elegant, efficient and cost-effective manner. Bonus points might be warranted if the proposed solution also delivered some kind of competitive advantage!

A few years ago, I even wrote a book on this subject

Those methods and foci are no longer all that valuable as we enter a new age of software. This new age includes solutions where:

  • Generative AI helps users complete tasks, suggest solutions/answers and encode transactions
  • New, advanced technologies may possess a raft of new risks that must be surfaced and mitigated
  • New solutions may run on public cloud, hyperscaler computing facilities that may possess additional risks and costs
  • The newness of today’s solutions may mean that users may not fully understand the new capabilities, the advanced capabilities behind them and the risks that may be present
  • Machine learning and no-code capabilities help tune or reengineer processes 
  • The ‘platform’ underneath a solution is possibly more important that the pre-supplied functions and features found in the applications
  • A vendor’s partners take on a new importance beyond just implementing a solution quickly. Does the implementer know how to ‘extend’ applications, develop vertical capabilities, etc.? 

And much more.

Right now, every vendor seems to sound the same. They all claim to have things like platforms and AI. The reality is that there can be big differences between vendors in just these two new areas. Buyers have to be more savvy to discern what the real differentiation is in today’s application software offerings. This is because the software that companies can buy today does NOT resemble what was being sold at the start of 2023. As a result, the methods to acquire it have to change. 

The new selection criteria

Let’s look at several major technologies that are (or should be) influencing buyers and what issues buyers should be considering for each. While not exhaustive, this list includes platforms, AI, workflows, employee experience (EX) and more. 

Platforms

Starting in the summer of 2023, I was getting frustrated with vendor/analyst briefings as all anyone wanted to discuss was their platform or how excited they were about generative AI. At first, I thought this was some sort of new-tech fueled mass hysteria that would hopefully subside soon. But, it turns out, platforms and AI are here to stay. 

The apex of platform mania, for me, occurred at the October 2023 HR Tech Conference. Everybody wanted to tell me about their platform. These vendor briefings assumed that I’d never heard of a platform before and they wanted to go through all of the layers and layers of intricacies within them. You can’t do a detailed platform discussion and still cover cool function bits about the product all in 30 minutes. It can’t be done.

In this piece I did regarding that HR show, I noted:

Of those 500 or so exhibitors, I’m pretty sure that 400 of them truly love and adore the HR technology platform their solution rests upon and they want to tell the world about. And each of them believes their platform is different and/or competitively advantageous.

Unfortunately, that rosy self-assessment may not be true. One vendor’s CEO that I met with argued with me for almost 45 minutes that her product’s platform was competitively differentiated. She spoke like a technologist about her technology. I had to repeatedly stop and ask her to explain just what exactly is the problem she is solving for customers and what experience or experiential aspects of her solution are resonating with these buyers to solve those problems. A platform that doesn’t connect up to a business problem is an interesting technology going nowhere.

The corollaries to that are:

  • No-one comes to this show specifically looking to buy a ‘platform’. Nope! They attend because their firm has a business problem (e.g., they can’t find enough qualified workers) that they need solved. 
  • No-one can get budgetary approval to attend a show like this to look at ‘platforms’.
  • If a prospect has to struggle to figure out what your platform’s raison d'être (i.e., reason for existence) is, your firm needs a new marketing person. Life’s too short for this.

The problem at this show was that pride of authorship and a lack of competitive awareness made all of these discussions generic, overly long and unfocused. If only these vendors had taken the time to see how many attributes of their platform look and sound like their competitors then they could have sharpened their messaging and made their platform pitch more targeted and succinct. Seriously folks, I don’t ever need to see another vendor’s wheel or layered bands of platform stack components for the rest of my life.

The other cardinal sin was the inability of vendors to show the value of their platform. Yes, it enables more rapid deployment of advanced capabilities into applications. While I can state this in a sentence, vendors are incapable of doing it in an hour of presentations. If it doesn’t speak to solving one of a prospect’s top business problems, it’s probably not worth mentioning. 

What I also didn’t see happening last Fall was that the way software buyers must examine new offerings has to change. Yes, a platform is now quite important but no one has a killer list of the most important characteristics within a platform that customers should expect. 

This discussion on platforms should NOT get confused by the platform narratives that a handful of ERP vendors pushed over the last decade or so. Those stories were intended to convince software buyers/customers that the vendor had changed the underlying plumbing of older on-premises solutions to now work in the cloud. This was predominately a technical change but was marketed as being something greater. It wasn’t. 

We need new software selection questions for buyers to consider. Buyers need to know:

  • Whether the platform possesses entire, distinct layers for specialized processing? Are there services for integration, generative AI, machine learning, security, controls, etc.? 
  • How extensible are the products built with this platform? Do they share the same data model? Can customers and integrators build or extend applications? How easily is this accomplished? 
  • Can users (not IT) use no/low-code capabilities to extend applications?
  • Can workflows be designed, automated and reengineered by end users?
  • Can non-vendor data be easily connected to and addressed by the software? Is this data stored in a data lake and is there any latency with this information?
  • How does the platform protect a user’s data without impacting the usefulness of new AI tools? 
  • How do third-parties (e.g., implementers, software partner firms, etc.) connect into the vendor’s platform and broader ecosystem? Do all of these players abide by the same security, governance and privacy terms?
  • Etc.

The key issue isn’t whether a vendor has a platform or not. In fact, most firms will gladly point out that they possess one. What is key is:

  • How robust the platform is
  • How well the platform can flex to support future technical environment, process, controls and other changes
  • Whether this platform is markedly more complete or powerful than that of a competitor’s 
  • How much of the of stack belongs to third party infrastructure/hyperscaler firm, is open source, or rests with some other third-party
  • Whether the platform infringes on some data sovereignty matters
  • How well the platform can protect your internal company data/IP and whether your data is being used to train LLMs or aggregated for resale to third-parties

AI – Artificial Intelligence

AI, especially generative AI, took off like a rocket ship last year. It’s now a core component of most every software vendor’s marketing spiel and vision. The importance and risks associated with AI are so great that these capabilities must be fully understood, pre-purchase, by software buyers.

Like platforms, not all AI capabilities are consistent (or even desirable) across vendors. The challenge for software buyers is that they need to do some soul-searching of their own first before starting a software selection. The reason for this is that each customer will want to identify how it will and won’t use AI and what kinds of risks it can/cannot live with.  I stepped through these issues in this Jan. 2023 article. Upon review, it’s a solid piece with tons of counsel, suggested questions for vendors and more. I won’t repeat that here but I’d highly recommend anyone looking at new application software to review it. 

AI includes a number of disciplines and use cases. Your firm may want to prioritize what it wants/needs before engaging with potential vendors. For example, a very risk adverse firm may only want to pursue machine language (ML) solutions. ML is frequently used in workflow automation applications. Other firms might want to use chatbots, content generators, code generators, etc. There are many different uses for AI and different AI tools might better aid in specific use cases. 

Likewise, not every AI tool/use case combination has the same risk profile. For example, some AI tools trained on large third-party databases could be more prone to hallucinations or return copy with objectionable content (e.g., curse words, sexually explicit pictures, etc.). 

AI tools that utilize external data exclusively (e.g., weather forecast data) may have a low risk profile while other tools that have access to sensitive internal data or intellectual property may be deemed too risky to use. A major risk concern is that private data can become part of a public AI database. Assurances from vendors that the data will be aggregated or anonymized may not be adequate protection. 

So, before a firm can begin selecting its next application software, it has to decide what its AI risk profile is and how it will judge potential solutions. In this article, I took a deep look at the readiness of HR executives to acquire new AI-powered HR software. The picture it painted wasn’t a good one.

The risk profile of your firm requires you to decide what kinds of guardrails, co-pilots, safeties, etc. should be part of your AI enabled new software. 

The better application software vendors are already working with standards bodies and governments (globally) to ensure their AI efforts will not run afoul of rapidly emerging new laws and regulations. These better vendors are focused on a few AI priorities: 

  • Designing ethical, safe, protected, legal, defensible solutions
  • Providing audit trails, traceability, etc. in how AI recommendations or content were generated
  • Developing a feedback mechanism to help users or human co-pilots to report aberrant, illogical, offensive, incorrect, illegal, etc. results and get these resolved
  • Using systems thinking to imagine how some users will abuse or misuse these tools. These malevolent actions could alter future users’ results, denigrate the effectiveness of the software, cause reputational harm to the software buyer firm or other damages.
  • Carefully introduce AI-assisted capabilities into their products
  • Clear guidance as to who exactly owns the training data as well as the outputs from AI and algorithmic tools
  • Identifying how long data will be used and retained by AI tools (e.g., When will AI tools trained on your employee data purge records of former or deceased employees?)
  • Documenting when some datasets will be replaced and how this will be accomplished

Smart software buyers might look at how vendors are:

  • Drawing a bright line as to what AI capabilities will use private data vs. public data
  • Prioritizing AI development efforts in areas like:
    • Chat
    • Assisted decision making
    • Assisted form or field completion
    • Workflow automation
    • Forecasting (or forecasting more accurately)
    • Production optimization 
    • Repetitive task automation
    • Etc. 
  • Deploying different AI capabilities carefully, thoughtfully and productively throughout their product line
  • Using AI to reshape not just how users enter transactions but also how AI can radically transform how work is completed and what work should actually be. How is the user experience being impacted by AI?
  • Anticipating and bulletproofing their AI processes, data and workflows from miscreants
  • Rolling out AI capabilities. Are they moving too slow/cautiously or too fast/reckless?
  • Carefully rolling out AI capabilities to smaller and/or less technically sophisticated users

In the last few months, some differentiators appear to be forming. The best vendors:

  • Include AI capabilities (all of them) within their platform
  • Have a data lake to complement the customer’s transaction data. This data lake can be a form factor greater in size and incorporate all kinds of other data (e.g., third party databases, sensor data, images, etc.). The data lake becomes the feed stock for AI insights, content, etc.
  • Start off sensibly
  • Ensure each AI utility delivers productivity improvements, insights, AI assisted decisions, content or other value

AND

  • Provide AI capabilities as part of the standard SaaS pricing. These capabilities are NOT priced a la carte or are separate SKUs in the vendor’s sales system.

The (incomplete) vision

Watch a vendor demo (or drop into an analyst briefing) and it’s likely that 90% of the presentation/conversation concerns the first two topics (AI and Platforms). What is missing from these exchanges is considerable. 

When vendors describe how AI will change work or the employee experience, you’ll likely see a wee bit on workflow technology and a mobile/cellphone demonstration. What you won’t see is telling though including how non-employees will use the technology. More specifically, the processes never consider all of the other technology and data that suppliers, customers, jobseekers, regulators, hackers, trolls, miscreants and others might use.

The lack of systems thinking is appalling today. No one is discussing how outsiders will impact processes. For example, there are jobseekers using AI to write resumes, cover letters and thank you notes for them; develop great answers to common interview questions; apply to 5,000 job openings in one week; and more. These persons are not only gaming the system to their favor but making life difficult for other job seekers and for recruiters. This is a big problem as today’s vendors are thinking too incrementally. Their thinking is constrained and the resulting technology solutions they are suggesting you buy will likely be short-lived and problematic.  In evaluating new solutions, seek to understand how others might use and abuse the same technology.  

There are other vision issues to consider. For example, has the vendor produced a new set of metrics, performance measures, controls, etc. that should become part of your implementation?  Today’s new technology can allow your employees to work beyond the limitations of transaction entry and processing. What does management look like in a world where automation deals with most everything transaction oriented? Does the vendor have a point of view on this? 

Likewise, you might the vendor’s ‘vision’ lacking in several areas like:

  • The software may come with some pre-supplied process workflows marketed as best practices. These will likely only provide competitive parity when your firm would want competitive advantage. Does the vendor offer alternative process designs? Are the suggested designs merely artefacts of yesteryear’s business methods or do they represent something radically new, inspired and differentiating?
  • Does the vendor’s vision go beyond AI and platforms?  If the vendor only recently ‘discovered’ AI, how great can their vision be anyway? Can the vendor articulate a compelling vision of what management, leadership, processes, workforces, and your industry will be like in 2, 5 or 10 years’ time? 

This is an important issue as most companies are loathe to replace their ERP and other application software in less than a decade (or two). But, if they pick a vendor who can’t help them navigate technical and business changes in a timely fashion, they will need to make an untimely, disruptive and expensive software change in only a few years. Is this vendor’s team mostly a sales organization, a financially-driven group, or a group of visionary leaders you can expect to be ahead of the curve? 

Make sure your potential vendor can paint a compelling vision for the ‘afterlife’. That is, how your firm will look/operate after it implements their solution. If the vision is unattractive, so is the vendor.

What happened to multi-tenancy?

For the last couple of decades, software buyers had to ascertain when a vendor would move from single-tenant, hosted or on-premises software to a multi-tenant cloud solution. Eventually, most major vendors made the shift (although they took a very long time rearchitecting their solutions to support this). Multi-tenancy permitted many different software customers to share the same technical infrastructure and code while their data was walled off from other customers. 

Now, the market has shifted again. The new deployment and architecture style that is in vogue involves a public cloud deployment. In this situation, a hyperscaler (e.g., Amazon AWS, Google Cloud, etc.) provides almost limitless disk storage, memory and computing power to a software vendor’s customers. The software vendor places its software in the hyperscaler’s servers and maintains the product. Customer data is not comingled but does reside in the hyperscaler’s data centers. Some customers are demanding a dedicated instance of the software so that they have control over the timing and configuration of the software. 

This new approach is really a blend of past methods where customers get more choice in how software is installed, upgraded and configured while also ensuring that their data is walled off from others. 

Software buyers need to understand exactly the kind of architecture they want and get these needs documented in their RFP/RFI. Some industries, like State/Federal Governments, Financial Services, etc. have very explicit demands here. Major ERP vendors may be the most accommodating with these needs while smaller firms may lack critical flexibility. 

All software buyers should carefully examine what data sovereignty requirements they have and how these align with the server deployments and data storage locations of the vendor and/or hyperscaler. Buyers can’t just assume this will not be an issue.

Combination possibilities

Much of the current thinking espoused by software vendors has been linear and incremental.  You will see it everywhere: let’s add a chatbot here and a smart recommendation there. The really big ideas and imaginings are likely a couple of years off. 

Nonetheless, buyers should look for vendors that are combining two or more advanced technologies into a single solution. For example, it is cool to see how AI can automatically monitor workflows and recommend different/new controls and workflows based on emerging patterns. Likewise, anomaly detection tools can be reinforced with smart risk assessments to automatically recommend which of these odd occurrences be looked at first. Workflow, algorithms and AI can be a powerful combination especially as it helps reduce risk, prioritizes activities, and suggests responses. 

Software buyers should see combinations of advanced technologies in areas like: accounting consolidation and closing activities; processing of Payroll to General Ledger journal entries, accruals and reversals; and in the development of management/board briefing books.

Verticalization

If your firm is in a vertical that a vendor doesn’t fully support now, you might want to consider looking elsewhere for new solutions. Software companies are going to be slammed trying to modernize their existing applications and verticals for the foreseeable future. Their development resources are trying make every process, every application, etc. a modern, AI-powered solution. While we may see a flurry of quick incremental enhancements over the next couple of years, don’t expect a software vendor to move into a net-new vertical anytime soon. 

On this point, expect vendors to work on their cross-industry, horizontal applications (e.g., Finance and HR) first and verticals will be a distant second. 

Partners

Are partners changing, too? They could be. 

Customer expectations of partners will like grow as partners (e.g., implementation, software, development, services, etc.) find new avenues of revenue expansion with changing software vendors. Buyers should understand how partners are:

  • Developing local and vertical extensions (e.g., are they using ethical, secure practices)
  • Integrating third-party tools into new workflows
  • Monetizing their intellectual property products
  • Altering their implementation processes to account for new technologies and business practices
  • Creating and implementing solutions to deliver competitive advantage (vs. parity)

Smart software buyers need to review any product extensions built by third-parties. No software user wants to find out, post facto, that a partner firm is:

  • Using a copy of their proprietary data to train AI LLMs for an unrelated product
  • Trying to monetize an aggregated or anonymized version of their data
  • Susceptible to basic hacking efforts
  • Inappropriately monitoring your transactions
  • Less than reputable
  • Developing product extensions in a restricted country or with forced labor
  • Not paying a livable wage to its employees or contractors
  • Shipping, storing or processing your data outside of your country of operation
  • Etc.

Vendor shortfalls

In their rush to show us their newfound AI and platform capabilities, vendors have gotten some of the technology in place but their marketing and sales aids may be lacking.  As a result, software selections may be more complicated and obtuse. Smart vendors need to do a great/better job of:

  • Articulating their long-term vision not just for the technology but also for their customers’ businesses
  • Showing the competitive differentiation 
  • Helping potential customers decide between solutions 
  • Explaining how their data is secured each and every place advanced tech (e.g., AI) is in use
  • Showcasing more than one potential workflow
  • Stating, in plain language, who is liable for direct and indirect damages due to the use of new AI tools
  • Indemnifying customers for any real and substantial damage triggered by a rogue AI tool or the bad powering it
  • Identifying exactly what role the customers’ personnel must play in monitoring AI outputs and how failure to do so will absolve most or all of the vendor’s liability for losses triggered by AI tools
  • Providing very accurate cost estimates for customers. Yes, I know that vendors can’t predict everything (e.g., how many images you’ll place in your data lake) but they should at least provide customers with the pricing formulas and the projected vs. actual costs prior customers have incurred. Great vendors don’t pass the buck (to a hyperscaler).
  • Communicating which AI or advanced capabilities come with the software and which ones require separate pricing and agreements.

One really big item that has not been surfaced in any briefing I’ve attended is the vendor/customer prenup and its provisions. This is the part of a subscription agreement where the two parties agree to smooth, phased cessation of services and clarify exactly how limited access to data and apps will continue for a limited timeframe. This is important as the new solutions have massive data lakes, well-tuned predictive models, custom LLMs, proprietary algorithms and more that make it really hard for a customer to just pick up their data and go elsewhere without some real pain. Don’t buy any software until both sides have this prenup nailed down.

My take

A great software selection team will want to identify, vet and test each and every instance where an advanced technology embedded in a software solution. Buyers need to make sure that each advanced tool ensures their data remains protected, secure, and does not become part of a vendor’s AI training data. This will not be a trivial effort. 

New solutions should be creating not only better, ‘deskless’ and more positive user experiences, they also should be using AI to predict what users like/want/need, the data they’ll input, etc. But through it all, new applications should be creating experiences for the individual that are tempered by the content and context of the data they are working with at that moment. 

For me personally, I would love to get into another great big software selection project if only to document all of the new requirements, negotiating issues, discovery matters and more that a savvy buyer will want to explore. Those are highest value producing efforts out there – and – they’re illuminating. Is it time to shine a bright light on your software selection project?

Loading
A grey colored placeholder image