Category Archives: Spend Analysis

The State of Spend Intelligence is The Same Today as it Was 20 Years Ago. There is NONE!

In this age of AI hype, every vendor with an “Agentic AI” or an “AI Employee” is claiming their LLM-based garbage app is going to provide you with the “spend intelligence” you need for your Procurement success. The reality is that it is as much hogwash today as it was 20 years ago when I wrote that There’s No Such Thing as Spend Intelligence.

The words I wrote exactly 20 years ago today still stand.

There is no such thing as a spend intelligence solution.

Let’s start with the definition of intelligence.

Intelligence is a most complex practical property of mind, integrating numerous mental abilities, such as the capacities to reason, plan, solve problems, think abstractly, comprehend ideas and language, and learn.

And since spend management solutions are software, let’s review a definition for software.

Software is the (collection of) program(s) that enable a computer to perform a specific task, as opposed to the physical components of the system (hardware), where a program is the collection of source code and libraries which have been compiled into an executable or otherwise interpreted to “run” in (active) computer memory, where it can perform both automatic and interactive tasks with data.

Simply put, intelligence is a property of mind and software is a property of machine. And despite the efforts of the artificial intelligence community, I do not expect the property to cross the chasm anytime soon. Artificial intelligence is simply a collection of very sophisticated algorithms processing large data stores, instruction sets, and probabilities very quickly to come up with reasonable responses to queries – it is not thought, although it might appear to be thought since today’s computers can perform billions of calculations in a second.

The only difference is that today’s processors easily do hundreds of billions of instructions per second and with multi-threading and massive parallelization, average racks do hundreds of trillions of calculations per second and custom AI Data Center racks do quadrillions of calculations per second. But calculations do not equal intelligence. And since the models are still probabilistic, they haven’t gotten any better and, in fact, with LLMs, they’ve gotten worse!

In fact, with hallucinations a core function, not only can you not depend on the math (which these models get wrong all the time) but you can’t even be sure they’re working on the right data. We’ve went from insight to inaccuracy. That’s not intelligence. That’s idiocy … at it’s finest!

Next Generation Analytics NEEDS to Surface Root Cause Analysis …

… but relationship modelling alone is NOT going to get us there!

In another great article by Xavier Olivera of Hackett Spend Matters, he dives into the topic of how procurement analytics needs to work – from visibility to orientation because current procurement analytics offerings, while reasonably good and actionable at the process level compared to where they were a few years ago, are poor at helping users orient themselves when a specific goal or problem comes into focus.

He notes that when a procurement leader decides they want to improve X, the challenge is no longer visibility. It is knowing which analytics matter for that objective and which do not. But all the analytics platforms give them today is metrics, they don’t give them direction. Even if the user knows what metric to drill into first (because it is the highest, lowest, or outlier), all they can see is the data that contributed to that metric. For spend, the transactions. For a supplier rating, the Net Promoter Scores. For a process, the time in each step.

The users see the immediate “what”, but not the “why”. Why were the transactions high? Is this market price, has the quantity gone up, or is the supplier charging above the agreed upon rate. For a rating, is it because the performance wasn’t up to spec, the delivery is consistently late, or the service/interactions are very poor. For a process, which time was too long (compared to average), unless you can dig into another level (and even then, why it was too long).

According to Xavier, in situations like these, analytics has to work different. When a procurement leader wants to improve contract compliance, the starting point should not be a full review of all compliance metrics, benchmarks and dashboards. It should be a guided path that surfaces the specific reports, KPIs and comparisons most likely to explain the gap, given the organization’s operating context.

Which is a great start, but just surfacing those reports, KPIs, and comparisons that are statistically relevant or deviations from a norm doesn’t explain the gap, it just captures the gap. Not only is it the case that a KPI only becomes meaningful once it is examined in the right context, but it only becomes useful if there is enough data to allow the system to determine, with high statistical likelihood, the root cause and actions to take that could address the root cause (and not just the symptom these systems surface today).

Xavier than tells us that the ability to orient analytics effectively depends on the data’s structure, which is partially right, but doesn’t quite capture the entire requirement. He goes onto state that Procurement outcomes do not arise from isolated transactions … they emerge over time from relationships and analytics is most effective when the underlying data model can express these relationships explicitly. Which is closer. But the reality is that this still isn’t enough for proper root cause analysis.

It’s critical, because without relationships you can’t trace the end metric back to the source data, but just being able to identify the source data only tells you what is fundamentally wrong, not why, or what you need to do about it.

That’s where analytics needs to get to.

If your steel category transactions are high, you can trace back to the contracts and whether or not the rates are per contract, the shipping is per carrier quote, the tonnage as expected, and the breakdown across steel categories appropriate for your current product lines or construction products. If any rates or tonnage don’t add up, you know the issue is the invoices — but you don’t know why they are being paid. Were the new rates not properly encoded? Were the tolerances within acceptable limits and the automatic OK-to-Pay issued despite the mismatch? Are category managers blindly overriding the system because the supplier was threatening late shipments if payments didn’t appear on time?

In Xavier’s example, if contract compliance is low, why? Is it just a few suppliers, or even a single supplier, across a category. If just a few suppliers, are they unaware of the contract because of personnel changeover? Did a new industry regulation adversely affect them? Was it actually the fault of a carrier or sub-tier supplier they had no control over? This is what you need to determine to ensure that compliance actually improves and stays improved.

In other words, you need more than the data, you need models that capture what the data element used in a KPI is, who or what creates the data in the first place (and how they create that data), what the data range and typical mean/median/mode values are, what positively or negatively impacts the data, and what can be done if a shift is desired in the data.

Without this baked in intelligence into the model, even if the root data in the system can be uncovered, a user won’t understand what it means or where to start doing something about it. That’s where analytics needs to get to for analysts to be proactive instead of reactive.

And this is another area where the Busch-Lamoureux approach to Exact Purchasing will help. When you define your categories at a granular level appropriate to to the quadrant of the pocket cube they occupy, you not only know what influences their cost, but what also influences their supply, what defines their quality, and what role third parties (that you may have to monitor) play. You have the foundations for doing real proactive analysis and identifying not only what “good” is but what is most likely contributing to a “not good” metric or data point and what standard options exist to address, and try to improve, the data point (as you need to mitigate high risk and manage high complex categories at a detailed level).

In other words, the future is knowledge-based models that capture more than data points and calculations, but what the data points actually mean and what factors (represented by other data points) directly influence the data points you are analyzing.

Analytics Must Drive Source-to-Pay, but not necessarily Gen-AI

Xavier recently penned another great piece on Analytics in P2P: From visibility to actionability where he highlighted the failures in analytics in traditional P2P:

  • static, backward looking, spend by category, invoice cycle time, approval rates, compliance rates
  • insights only after transactions are processed, payments are made, and cycles completed
  • late payments multiplying, exceptions accelerating, and supplier risk accumulating
  • lack of operational insight

According to Xavier, P2P can only be modernized if the embedded analytics shift from descriptive to diagnostic.

  • don’t report KPIs, explain the root causes (which approval paths contributed the most to approval time)
  • don’t report exception rates, identify suppliers that consistently cause them
  • don’t report spend anomalies, break it down and identify root causes

It’s a great start, but where it needs to get to is actionability. Xavier begins to address this point by stating the next step is “predictive awareness” where the system anticipates likely outcomes within active processes, such as predicting which invoices are likely to miss payment terms, which requisitions are likely to stall in approval or which suppliers are likely to generate disputes based on current patterns as that allows a Procurement professional to intervene before issues arise.

Finally, Xavier gets to the main point — the real inflection point comes when analytics begin to recommend actions and influence execution paths. Prescriptive analytics in P2P requires tight coupling between insight and control. If analytics identify a high-risk transaction, the system must be able to route it differently, apply additional validation or prompt a specific decision. If analytics detect a low-risk, repetitive transaction, the system must be able to reduce friction without manual intervention.

But it needs to go one step further. It must not only route differently, and apply more controls, but it must still do so automatically based on the diagnostic and predictive analytics. It can’t just apply a “one-size-fits-all” approach for automation and kick every exception out for human processing. You can’t always make the default path smarter because there should be different paths depending on the cost of the purchase, the risk associated with the purchase, the discrepancy between the invoice, goods receipt, PO, and/or contract terms and conditions. You need multiple streams that are auto-selected by predictive analytics that support the right actions given the assessment of the conditions.

The reality is this — except for truly exceptional situations, once you’ve made the decision on what to purchase, procurement should be 100% automated. It’s all e-document exchange, analysis, authorizations, and (payment) transactions. Unless something is really off, a buyer should never be involved once all the workflows, rules, and authorizations are setup.

But this automation should extend back into, and through, source-to-contract. Building on the Busch-Lamoureux Exact Purchasing pocket-cube framework, there are categories that are low risk, low value, and low complexity — you should NOT be buying these manually. “Agentic” automation should be taking care of these for you, considering that even a worst-case screw up will be of little impact. Then there are categories of moderate risk, value, and/or complexity which can be fully automated if all of the necessary data is available and there is a cost and supply history to build on, there are no special situations that need to be taken into account, and a worst-case analysis indicates that even a statistically unlikely “bad buy” will be of minimal impact. These should be 90%+ automated from the decision to buy to the recommended award, with extensive analytics and augmented intelligence for human review. And if the buyer likes the default recommendation, it should be just one click for the process to go from award to e-signed contract.

All of this requires very extensive descriptive, diagnostic, predictive, and actionable analytics and intelligence with extensive, adaptive, robotic process automation ([A]RPA) that can automate everything that should be. The reality is that while everything should be sourced (or exactly purchased), when you have all of the (market) intelligence, the standard processes, and the organizational goals encoded, then there’s no reason that the systems shouldn’t do the majority (or the entirety) of the work for you.

While buyers won’t be replaced by agentic systems (despite the over-hyped BS claims of AI Employees), they will be heavily augmented by them when most categories aren’t complex, risky, or strategic enough to require human review or intervention.

The Lack of Adoption of Analytics is NOT Complicated!

According to THE PROPHET, the reason that we’ve never seen a breakout $100M+ pure-play (spend) analytics vendor is it’s complicated. (Source: LinkedIn)

But the reality is that it’s really not.

First of all, approximately one third of all multi-nationals are headquartered in the US. In other words, one third of global enterprise is based out of the US, where the strategic decisions are made. Let’s say that again, one third!

Secondly, and this is the real explanation, in our age of participation trophies and only focusing on the positive (when there really isn’t any), no one is willing to state the truth, and that is most of the employees responsible for strategic [spend] analysis are just too math stupid.

Analytics, at its core, requires good mathematics skills and, with traditional analytics applications, good computer skills.

However, the US, where many multi-nationals are based, consistently ranks in the lower part of the OECD international rankings and is currently 34th in the PISA [out of 79 scored countries] (with an average numeracy score of 249, below the TOTAL OECD average of 263, with over 1/3 of its adult population at level 1! This means they can’t even do basic arithmetic and problem solving [or calculate a tip FFS, but that does explain why they believed their administration when they lied and said other countries pay the tariffs] — and that’s the average business employee in the US, since anyone with a level 2 on the OECD can likely fake it in a STEM career in the US.

As for THE PROPHET‘s reasons as to why Spend Analysis has consistently underperformed the hype:

  • While 3/4 of solutions have always been reporting in drag, I’ve been highlighting at least a dozen Best of Breed solutions consistently for the past decade. They have existed for the past 20 years, you just had to look (and understand what to look for. But this site did a great job of helping you with that!)
  • Yes, scale came at the cost of dumbing down the UX (for the US market in particular)!
  • Unfortunately there is no faster way to die as a Spend Analysis vendor then to get scooped up by a (mega) suite or a Big X Comsultancy.
  • Actually, the analytics and optimization is not powerful or complex enough in most solutions. Again, the problem is that the vendor didn’t add incremental levels of simplification (i.e. dumbing down) so each user could take advantage of it at their mathematical (in)competency level.

But the real reason, as hinted above, is that employees resisted these advanced spend analytics solutions because they knew they didn’t have the mathematical skills to use them. (Which the US Education System should be blamed for [and why it should be fixed, not dismantled], not the employees, unless those employees went to University and chose not to take math courses to try and make up for the failings of the public education system they were subjected to.)

As for THE PROPHET‘s signals that the times they are a changin’:

  1. Good + Cheap = Dangerous
    Faster? Check! Cheaper? Check! Smarter? Well … Ask Woody!
  2. Analytics is Merging with Execution
    This is key for adoption of analytics — do it when you need it and apply the findings right away.
  3. Intake, Orchestration and Agentic Tech
    I guess I have to say it again!
    ????? ????????????? ?? ???????? ??? ??? ??????? ????!
    When what we really need is a Revenge of the Nerds! (If the USA even has any left!)

However, the real reason that we may finally be entering a new era in analytics is the following:

4. Most companies are trying to stave off bankruptcies as a result of US trade, market, etc. decisions that have already bankrupted many SMEs and they now realize that analytics is a key part of that solution. You can’t optimize spend you don’t understand, or understand the impact of a sudden 145% increase in tariffs if you don’t understand how much you are sourcing from the country in question.

2025 Is Just Another Year … But Is It All Doom and Gloom? Part 2 (Real Tech!)

As per our first instalment, it all depends on your point of view and whether you are willing to look beyond the hype, buckle down, and get the real job done.

For instance, just the following five technologies will eliminate 95% or more of your tactical sourcing, procurement, and supplier monitoring work — and all you have to do is find them, properly implement them, and use them. Let’s talk about them briefly.

Real DIY Analytics

The ability to analyze the data you want, when you want, how you want, enriched and augmented using the auxiliary data you want … and not in predefined dash-boards or hidden “AI Agents” which may, or may not, do the analysis you want (and need) … cannot be underestimated! Real value comes from ad-hoc analysis and investigating hunches, abnormalities, and trend changes when you discover them; not days, weeks, or months later when the “cube” has been refreshed, and it might be too late to correct a problem or capture an opportunity!

Remember, this is not 2005, this is 2025, and there are at least half a dozen great DIY (spend) analysis solutions that will do most of what you want, for a price tag that is a fraction of what you might expect, and if you are okay with full DIY, some of these start at a price you can put on a P-Card. For example, Spendata Classic (which can handle data sets up to 5 Million Rows) can be obtained for $699 a year, and Enterprise, which can handle data sets up to 15 Million records, which comes with unlimited use for 5 users (and view licenses for more), and some consulting and setup, starts at an amount that will surprise you. (You can still put it on a P-Card if you pay monthly.) And there is literally nothing you can’t do in it if you’re willing to apply a little elbow grease. It truly is The Power Tool for the Power Analyst.

(Strategic Sourcing Decision / Supply Chain Network) Optimization

Yes, it’s math. But you know what? Math works! And when you use deterministic math, it’s 100% accurate, every time! And it’s one of only two technologies in S2P+ that was been proven (by multiple analyst firms) to repeatedly identify 10%+ savings year-over-year (but since this was pre-COVID and pre- the 47th, we need to amend this finding to adjust for inflation and tariffs). And as an FYI, the other technology was NOT AI. (It was proper DIY spend analysis. Only Human Intelligence can intuit where to look for previously unidentified opportunities, the best AI can do is just follow a script and run standard analysis. Furthermore, the thing about spend analysis is that an analysis that identifies an opportunity only helps you ONCE — once you capture the opportunity, the analysis is useless. You need to do a new, and different, one.)

Rule Based Automation

When you think about most tasks across Source to Pay, most of them are just execution of simple, easily defined processes — most of which don’t require much (if any) intelligence and, thus, don’t need AI (and shouldn’t use an unpredictable AI agent when you can encode a process that gets it right, guaranteed, every single time. (Plus, the way you want to source, buy, pay, track, manage, etc. is probably a little bit different than your peers, and who knows how the AI Agent would do it for you. You certainly don’t!)

With rule based automation, you can easily execute an entire sourcing event in the background all the way to award if you like. It can run auctions, it can run multi-round RFPs with detailed feedback (it’s all calculations, response comparisons, and decisions on what data you want to share and how blinded you want it), it can run analyses and optimizations, it can calculate recommended award decisions subject to the constraints and goals that matter to you, present that to you for acceptance, or, if it’s a simple winner take all or top 2 situation, create the award automatically, send it out, get supplier acceptance, assemble the contract, and send it for e-Signature. You don’t need Agentric/Gen-AI, just tech we’ve had for over a decade!

Machine Learning

Now, when it comes to Enterprise Master Data Management and Administration (E-MDMA) and Invoice Processing, it can be quite a lot of work to keep up with the mapping, cleansing, and enrichment rules, and you don’t want to have to manually define all the new rules every time a new data element appears or a new invoice format arrives, especially if the system can auto-detect/”guess” 90% of the time through rule re-use and generalization. With machine learning, the system can keep track of your corrections, mathematically extract models, and adjust it’s rules to handle the new mapping again automatically as well as improve its suggestion logic when it doesn’t know what to do — increasing the chance that you just have to “accept” a new rule vs. defining it from scratch. (Unlike Gen-AI which just tries to find similar patterns somewhere to present you with something that may or may not have any correlation to your business and even reality!) And we’ve had great non-(pure-)Neural Network machine learning that works great with enough data for decades! Predictive analytics was making huge progress late last decade before this Gen-AI BS took over and could have helped Procurement departments automate 90%+ of what they wanted to automate with just a bit more development and effort by the leading vendors — it just required a bit more time, money, and focus. (Gen-AI has set us back a decade!)

Analytics Backed Augmented Intelligence

We don’t need machines to make decisions for us (especially when they can’t think, or even reason), we need machines to do calculations for us that help us make the right decision quickly and effectively. We need the machine to automatically identify and retrieve all of the relevant data, do all the relevant situational and market analysis, do all the predictive trend analysis, identify all of the typical responses with respect to the situation, predict the likely success of each, and present us with a set of ordered recommendations, complete with the calculations and supporting analysis, so we can pick one or realize that the machine didn’t/couldn’t know about a recent event or a human factor and that none of the responses are right (and that only we could craft one, with full information on the situation). The machine may not think, but the thunking it can do far exceeds our computational ability (billions of computations a second, all flawless), and that’s EXACTLY what we should be using the machine for.

If we give up on this Artificial Intelligence BS (even if the current models are right, machines need to be 100 Million times more powerful for it to even “mimic” human intelligence. That’s not happening any time soon) and instead just give all the machines all the (boring) grunt work, leaving us free to do what they can’t (strategy and relationships). If we do so, we can be at least 10 times as productive as we are now and deliver on the promises Gen-AI / Agentric AI / AGI never will, and do so at a small fraction of the cost. And oh, we have that tech today … we just need to deploy and integrate it properly!

And this is just the beginning of what you can do when you look beyond the hype and use your Human Intelligence [HI!] to cut through all the BS.