July 2, 2025

Dan Thompson: Using AI to Solve Forecasting’s Hidden Problems

Dan Thompson: Using AI to Solve Forecasting’s Hidden Problems
Dan Thompson: Using AI to Solve Forecasting’s Hidden Problems
AI Pathfinder for Private Equity Podcast
Dan Thompson: Using AI to Solve Forecasting’s Hidden Problems

Summary In this episode of the AI Pathfinder for Private Equity podcast, host Steve Budd speaks with Dan Thompson, CEO of Kluster, about the critical role of accurate forecasting in private equity. They discuss the challenges faced by private equity firms in forecasting, the concept of private equity grade forecasting, and how Kluster's unique approach leverages AI and data integration to enhance forecasting accuracy. The conversation also touches on the importance of data quality, the dynami...

Summary
In this episode of the AI Pathfinder for Private Equity podcast, host Steve Budd speaks with Dan Thompson, CEO of Kluster, about the critical role of accurate forecasting in private equity. They discuss the challenges faced by private equity firms in forecasting, the concept of private equity grade forecasting, and how Kluster's unique approach leverages AI and data integration to enhance forecasting accuracy. The conversation also touches on the importance of data quality, the dynamics between private equity firms and their portfolio companies, and the future of forecasting in the industry.

Takeaways

  • Accurate forecasting is crucial for private equity firms
  • Different stakeholders have varying definitions of forecasts
  • AI can significantly enhance forecasting accuracy and insights
  • Data quality is often a major barrier to effective forecasting
  • Sales cycles are lengthening, impacting revenue predictions
  • Standardising forecasting definitions across teams is essential
  • The future of forecasting may involve more autonomous systems

Chapters
00:00 Introduction to AI in Private Equity
00:54 The Importance of Accurate Forecasting
03:07 Challenges in Forecasting
06:41 Understanding Private Equity Grade Forecasting
08:28 Kluster's Unique Approach to Forecasting
12:08 Data Management and Its Impact on Forecasting
14:14 Changing Behaviors in Data Capture
17:06 Surprising Patterns in Sales Data
21:13 The Dynamic Between PE Firms and PortCos
23:55 Turning Forecasts into Action
25:34 Raising the Forecasting Game
27:54 The Future of Forecasting
31:23 Incorporating External Factors into Forecasting

Links

Dan's LinkedIn: https://www.linkedin.com/in/dan-thompson-kluster/

Kluster website: https://www.kluster.com/

AI Pathfinder Private Equity AI Strategy Meet-ups:

https://www.aipathfinder.co/private-equity-ai-events

Have a question or would like to suggest improvements, please contact steve@aipathfinder.co

SPEAKER_01

Hello and welcome to the AI Pathfinder for Private Equity podcast. My name is Steve Budd, and I'm on a mission to help private equity firms go from AI curiosity to confident action by building a network that shares what's working, what's not, and how to unlock meaningful value with AI. On this mission, I'm meeting some brilliant people who approach AI from diverse perspectives with domain expertise that private equity can greatly benefit from. Today we're focusing on a topic that every PE firm and Portco cares about, forecasting. It's a critical input for board reporting, growth plans, and ultimately value curation. And getting it wrong can kill confidence or momentum fast. Joining me is Dan Thompson, CEO and co-founder of Cluster. His company specializes in what they call private equity grade forecasting, helping portcos gain earlier and more accurate visibility of their top line. Welcome, Dan. Thanks so much for sharing and sparing the time to talk to me. Thank you, Steve. Great to be here. Good stuff. Well, look, first off, can we um just give the listeners a bit more background? I did give a little intro, a sneak preview, I suppose, there. But a little bit more on yourself, your background, and the story behind cluster, and really why did you decide to go deep into forecasting?

SPEAKER_00

Sure, sure thing. So, yes, I'm I'm co-founder and CEO of Cluster. My background, I'm a theoretical physics graduate. My first role was at Actress, an insurance technology SaaS platform in the very uh early days of SaaS. Um I was a data analyst slash data engineer there, I guess, for a few years. Um and then I moved into the uh insurance and reinsurance market, and uh I was working as an actor consultant for insurance and reinsurance companies. So my job would be to go into these companies across the uh Lloyds of London insurance market and essentially build them forecasting models to get the coverage they needed for normally uh large unpredictable events like natural catastrophes, marine aviation disasters, civil war, pandemics, etc. And so I'd build the uh large forecasting models and make them as flexible as possible because they've got many different use cases. Some of it is for pricing the risk and pricing the financial products you're selling, essentially. And then there's the other side of it, which is uh regulatory risk for reporting to the Bank of England, et cetera. So I was doing that for the best part of a decade, I guess. And while I was doing that role, I was talking about forecasting with someone who was a very successful sales leader, and he mentioned that um while he was running a high performance team, uh, his forecasting was was often inaccurate and done on a whiteboard. And so that was when we decided uh to bring the kind of forecasting acumen, essentially, that that exists in many sectors, uh particularly the financial sector, um, into the world of B2B to be organizations.

SPEAKER_01

Great. Okay, that that's yeah. Look, uh you probably see a whiteboard behind me, so very familiar. Although, you know, when it's just one person, it's it's slightly easier to forecast. That's for sure. Absolutely. Well, uh, um, I think we should get straight straight into this. You know, what why is forecasting so hard, even in well-run businesses? Uh, you know, I I I've I've been in in those as well. And again, you know, the whiteboards uh do still exist. Why is it misunderstood? And I and I I you know, having a look at your your LinkedIn as well, you you've sort of a little anecdote there around forecastings like playing the stock market, you know, it is really difficult.

SPEAKER_00

Absolutely, absolutely. So I could uh I could talk about this for many hours, but I think I'll just I'll just cover two points. The first point would be uh when it comes to particularly private equity-backed uh organizations, um, particularly if they've been growing uh inorganically through through uh acquisitions, etc., the data gets more and more uh complex. So even if you've only got one system, you've often got a plenty of standard objects, custom objects, custom fields, rules, formulas, validations, the the company may have been restructured over time and the meaning of certain sales stages, certain types of reporting, etc., all gets gets changed and reformatted. But furthermore, as a company gets older and as it grows inorganically, they've often got multiple systems, multiple CRMs reporting uh for what used to be different companies and is now different products within within a roll-up company, uh uh ERPs, data backed up into data warehouses, et cetera. And so having multiple systems and connecting them is hard enough, but actually having one central place that's always live at runtime 24-7 is extremely difficult. And so that is the main reason why forecasting and even just reporting is very hard, because you've got multiple moving parts all changing all the time, and you have to bring them all together, which is why A, it's hard, and B, it's often done with with a pretty material time lag. Um and for reporting, the time lag is is annoying, but for forecasting, it actually, by definition, makes the very forecast you've been creating uh even less accurate, the the the longer the lag has been. So from the more technical side, that's that's one of the big challenges. You of you obviously need a really comprehensive, essentially a data cube that's that's connected, um, live all the time and also augmented with all the trends and insights that you need as well. But on the other side of things, um what's what's really interesting about this this specific problem is that the definition of a forecast and the definition of different types of forecast um and the incentives uh behind it are very different depending on the stakeholder, even within a company. So um a CFO might be wanting a forecast that is the maximum number that you're 95% confident you won't go below, and you essentially run your cash flow off of that. Salesperson or someone in the commercial team uh might have some sort of bias to be anchoring their forecast near their near their target or their quota, even if there isn't the empirical evidence uh to underpin that, whether that be you know uh historic pipeline generation performance or current pipeline or current deals there that they're running. And so that's the first step, but then secondly, when you really drill into a forecast, what does it mean to have a best case forecast or a commit forecast or you know, optimistic or cautious forecast essentially? The very specific definitions of those is often not clearly defined. So even when a large amount of work is gone in uh going into it, you can find that all the people saying different numbers with all the different meanings between them. And the real problem there is when that's presented to the board and when there's the strategic strategic discussion around it, it becomes very, very difficult to actually make those decisions.

SPEAKER_01

Yeah, and and um, you know, as as as you're um certainly, you know, the front page of your website talks about P grade forecasting. Can you just just give us a bit of an uh understanding there of what you mean by that? But also in terms of also the chain are there are there changing expectations as well around forecasting now? Is there more expectation on the on the port code to have that up and running?

SPEAKER_00

Absolutely, absolutely. So um regarding the whole private equity grade forecasting, if something within within business or business operations is private equity grade, it's essentially the the top tier that you can get as expected by a top-tier private equity fund. And so when you look back to the private equity uh model, let's say they they invest or uh buy out a company, they will have a value creation plan uh over a period of time, let's say five years, and they need to get the transaction done, get in there, and they've got a timeline, whether it be through technology or across the people or across the strategy of the business. And they need to roll that out in phases and furthermore, just as importantly, be reporting on it so that you can ensure that every part of your value creation plan is on track, and if it's not on track, you can jump on it as quickly as possible. Because the private equity game is all about IRR. So with any phase that is delayed, you're therefore creating a downstream effect for all the other phases of the value creation plan in terms of slowing that down. Um the result of that would be uh a lower IRR essentially. So when we talk about private equity grade forecasting, what we mean is the most resilient, robust, and comprehensive forecasting solution you could get where every number, furthermore, is backed by empirical evidence and reportable on. So you can see the number but also understand why. So when it comes to forecasting, your number is 10 million, and that is based on these pieces of empirical evidence. You've closed X, your pipeline generation is Y, there might be a new marketing campaign coming out, etc. etc., underpinned by facts, but with any other part of reporting or forecasting, it's always backed by uh the KPIs they need to make sure everything's on track for their for their broader plan.

SPEAKER_01

And so um obviously we're getting close to the the solution here with cluster. What what's what are you doing differently that hasn't sort of existed previously? And and and obviously we're gonna talk about you know how how and you've been running for nearly 10 years now. It's not like um it's the generative AI piece that's that's an enabled you necessarily, but it'd be good to hear, you know, how how is that supporting the the uh the proposition, the product?

SPEAKER_00

Absolutely. So what makes us different is that there are there are plenty of players in the market on the sales intelligence side, and on the other side though, that there's plenty in the kind of office of the CFO space uh across FPNA and and and much more. Now, a lot of these vendors in the market are kind of software or SaaS solutions that either sit on data or they process data to go in and then report back on it or do something else on it coming back out, like a workflow. Now, Cluster is an enterprise data company, essentially. So uh we we deploy it as a solution because as everyone knows, forecasting and reporting isn't just a technology fix. If you've got sporadic data systems uh all over your company, all over the world, you can't just get a SaaS platform and plug it into them and um and it works. So what makes us different is that we connect all the CRMs, the ERPs, marketing system, data warehouse, et cetera, and work with the team. Members of the cluster team uh will work with our customer to ensure that all the business logic, all the taxonomy is centralized, aligned, in line with what the board wants, what the CFO wants, and what what the private equity sponsor wants as well. And we pull the data together, we clean it, we make sure that it's all live 24-7, and then we amplify it over a thousand times. So once it's pulled together into one central, essentially, data cube, we can explode it with every insight, every risk, every opportunity, and every trend across your organization. And only once we've done that, you can then layer on top your historic reporting, we can do your bookings, your revenue, your revenue recognition, forecasting across every product line. But you can't do any of that if your data is in different siloed systems. And so we're the only uh we're the only vendor on the market, we're the only solution on the market doing what we do in our space. And that's that's what makes us different. And going to your your AI question, we've been we've been at first since since the beginning, as you mentioned, five plus years ago it was more machine learning and then neural networks, etc. came out. We very heavily leveraged that side of things. But similarly, you can only get real value out of um out of machine learning and AI if your data is clean and in a great place and as up to date as possible. And it depends on the use case. So if you're looking at trends, uh we use the most the most advanced libraries to spot risks in deals, in entire cohorts of pipeline across any part of your business and alert those to you. And when it comes to the more LM side of things, we are investing heavily in essentially using that both in the in the data, but also in the user experience. So uh so obviously throughout the cluster application, the the the end user will be able to query the data, talk to the data, and and find out everything they need to know to make the decisions as quick as possible, uh as soon as possible.

SPEAKER_01

The the port codes are diverse, aren't they?

unknown

Yeah.

SPEAKER_01

And you know, the discussion that we had just last week at the breakfast briefing I host was around data foundations and you know the importance of that, and something that's been been asked for. What are I'm sure you get to see a lot of good and not so good levels of of data sort of management. Is that a big issue? Is that one of the blockers to really sort of getting good forecasting? Or or indeed your solution, do they need to be a certain level to even sort of get to the point where they're able to plug in what you do?

SPEAKER_00

No, there's not there's no certain level. So when it comes to what cluster does, common objection is is our data's not ready or our sales reps, our account execs and account managers are only are only filling out three fields and they should be filling out 20 per deal. The reality is that if you're the end user or or sales leader or someone within the company, you might go into your CRM and look at the deal and see three fields. And you might say, right, you've got the stage, you've got the close date, which is too close, which is too optimistic, it should be further back, you've got the amount, which is probably too optimistic as well, it's probably too high because you know there'll be some discounting going on, etc. So you'd look at that and think, well, our data isn't ready yet. But going back to the point on the on the the reality is that if a sales rep has been running a large deal, a six or maybe seven figure deal and it's been going on for six plus months, they will have created the opportunity of the deal six plus months ago. And every call, every email they've made is a touch point on that timeline. Every time they update the close date or the amount or whatever it might be and click save, that is another data point on the timeline. So instead of having three data points that you might think may not be very accurate, we might have 300, 600 plus data points in a time series for a single deal. And at private equity scale, there are obviously a pretty substantial number of deals, substantial amount of pipeline. We can aggregate that and then build out time series across every every part of the pipeline too. So when it comes to what we do, any any private equity scale company will will always have enough data. We've we've never come across a company that didn't. Um but in terms of a recommendation more broadly, um, it would always be to get started and simplify things. Quite quite often, a an instinct would be to define a complex data set and make it even more complex by adding more fields or tables or views or whatever you might be doing, and that actually has has the opposite effect.

SPEAKER_01

Just before we get into some of what you've been seeing in the in the trillions of data points, sales data points, is are you seeing a change in behaviour um uh around the data that's being captured and the importance of that data once you've you've come on board and they're starting to see these insights?

SPEAKER_00

Absolutely. I think to to preempt that when it comes to use cases for for AI broadly, I'd say there are that there are kind of two buckets, one of which is to make make an existing workflow or process much more efficient, and another one would be to gain an edge. Now, now cluster's position is to gain an edge. We give your your numbers and your forecasts and your insights much earlier. A very large proportion of the market is in automating away or accelerating workflows, and particularly with people in a commercial team, a huge amount of the AI is all around the call recording, the conversation intelligence, cadences of emails, etc., drafting the emails for you and making that much faster. And so the prevalence of these products and features across the sales intelligence market means that as every well, as every quarter goes by, there's more and more data in their systems. And that's not because not necessarily because the team have become more adept at filling out their CRM. It's simply because these new um these news new features and products are automating it for them. So that there is more and more data, and it is increasingly clean and in a better format, and that's that's thank how the market's moving forward, essentially.

SPEAKER_01

So do you have like a sort of an ideal tech stack that would just mean then that the quality of the data you're inputting is like a gold standard?

SPEAKER_00

It's an interesting point. I think what's going on at the market right now, particularly with what I just mentioned, is we will be moving towards large enterprise deals for these tools. But right now, there are a lot of up-and-coming new tools, and the the sales team will be putting it on their company credit card for $9.99 per month, whatever it might be. So you might have multiple people using different things, and whatever they feel accelerates their their work and helps them perform better is is what works. I think as as AI has, or the the the applied part of AI, in particular in the commercial, commercial space of the companies we work with, as it moves forward, it is um a bit more commoditized than technology might have been five to ten years ago. So there are companies that have come out of the blocks extremely quickly solving for one particular use case, whether it be in making a call and sending an email or answering a question or customer support, and they've come out of the blocks very quickly and scaled to a very material point. But equally, I think the flip side of that is the potential for churn and swapping out could be higher as well because they're leveraging the new new AI models, all of which would be you know off the back of OpenAI or any of the others like that, too. So to answer your question, I wouldn't I wouldn't say an ideal tech stack for them because the data is all recorded and put in in the same format. So I would always recommend keeping things as as lightweight as possible, have one clean data system, and as long as whatever's pushing into the data system is is in a consistent format, then um then it will it'll always work.

SPEAKER_01

Yeah, uh, I was just gonna say about standardization, um, and that must be really difficult right now. There's the amount of tools on market. Um just well, going back to uh I mentioned that that you've analyzed um a lot of sales data, what's what sort of patterns are surprising you that you think well, you know, you'd be useful for the for the audience here?

SPEAKER_00

Yeah, sure. So um one I I one main point that jumps out across all almost every company that we work with and across everyone I'm talking to is deals taking longer than expected and the market not catching up to that when it comes to their forecasts. And this also overlaps with what I mentioned at the beginning around people uh having different different semantics and different taxonomy for for what they're describing. So you could have um a chief sales officer or a VP sales saying, right, I've got these 10 deals and they're all in commit, and I'm saying I'm I'm 90%, I'm gonna give them 90 or 95% confidence they're gonna happen and they're gonna close. Now, what the data says is the percentage of those deals that slip to subsequent quarters has quadrupled in the past uh two years. So when the market started, really started to um kind of plateau and then and then contract, uh, I guess starting in very late 2022 and then into and across 2023 and beyond. What really happened when we were starting the sales cycles is is mostly near the end of the sales cycles, the procurement team is spent uh taking longer, uh, the CFO is getting more and more involved, there's more and more budget constraints, people are pushing things further back, they're asking for higher discounts, there's more and more hurdles, um, particularly with selling, um selling internationally, uh around um barriers uh such as tariffs more recently, etc. And so what we see is a deal appears to be moving at the speed it always has done for the past 10 years, and then once you're 80% in, the remaining 20% will will take four times as long uh as it has before. And so that is a trend we've seen absolutely everywhere. And the big problem there is obviously you could get to the end of the year and the VP of sales says, Well, I told you I was right, because 90% of those deals do actually close. But the CFO would come back and say, Well, only half of them close in the course that you said they would. And then our, for example, our new business revenue uh was was half or less than half of what we needed. Having that open conversation and making sure that everyone's sure that when they say 90%, do they mean in general, or do they mean within the exact financial reporting period you're talking about? And that is across the board. It's kind of plateaued now, and people are just about getting to grips with it. But that's the kind of thing that sounds like it's not that big a problem, but the actual impact on the revenue performance of a business is absolutely huge. It's it's one of the biggest drivers of it.

SPEAKER_01

Yeah, I can imagine. And and I assume this gives well obviously more confidence around the data points, but having more open conversations about this too.

SPEAKER_00

Absolutely, absolutely. And so when we're when we're working with a company, as mentioned, we pull all the data together and then amplify it with all the insights and and risks and opportunities. And so the second we spot a risk that this might be happening, the deal might be decelerating, there's fewer touch points on the deal across a week. We can then compare it with other deals selling the same product, potentially with the same sales rep, or maybe a better sales rep who can get deals done faster for the same product, and you can say, oh well, actually, the ideal the ideal path for this deal would be to have four stakeholders involved at the prospect, and you've only got two, you don't have the CFO involved yet, you should practically pull them in rather than hope they don't get pulled in by the other prospect, etc. So we we look out, we listen out for the signals that this is happening and then compare against uh against um against the deals that did work out and then and then find the path to get there as quickly as possible.

SPEAKER_01

Okay. Okay, look, I I um I I've worked in a a number of P back companies, but still P is relatively new to me. So when you've we've got this forecasting tool and and obviously this is not a new uh concept um around forecasting, the dynamic between the private equity firm themselves and the portco. How you we obviously normally have a um somebody sitting on the board from the PE firm. I'm just wondering how the the dynamic is there and whether there is almost like a dashboard purely for the PE firm themselves and and how they are looking to access this information more regularly, or hopefully that they have something like this across the whole portco. Yeah, I'm just wondering a little bit more on trying to understand that piece.

SPEAKER_00

Absolutely. Yeah, so what we would recommend and works well, um, obviously depending on on the culture of the business and and what people choose to do, is that the the dashboard is essentially the same from the board all the way down to the most junior sales rep, but obviously the data going into it is is is different or a subset of the top level. And and what you get when that is up and running, it's really fantastic. You get these kind of water cooler moments or coffee machine moments. Essentially, where the CEO or someone from the board can be talking to one of the junior sales reps and say, right, these are the top three KPIs we care about. What are you on? And at the company, you might want you know 10,000 leads, whatever it is, and the salesperson wants 200. It's a much smaller number, but the KPIs are identical themselves, the way they're calculated are identical, and the cadence at which they're reported over, whether it be weekly, monthly, or quarterly, is identical too. So that's that's a very powerful thing to do. But in practice, some people prefer to not do that. And at the private equity level, we've seen it be really powerful when you standardize the reporting KPIs. So we we can add on a layer on top of how the the portco operates to essentially translate the numbers and the taxonomy into a single into such the portfolio can say, right, these these X companies are all running on cluster. What is their high, medium, low forecast? What are these KPIs versus plan? Are we on track to be deploying uh the next the next phase of our value creation plan as well? And then that that helps massively with their portfolio operations. And you know, as mentioned at the end of the day, everything's an IRR game. So the faster you can you can get these things done, the faster you can move on to the next step and and and you can get to where you need to be.

SPEAKER_01

Uh you find it that I mean it'd be good to talk through if possible, obviously not naming no names, but how it's how that insight is sort of given private is sort of a case to um, I suppose potentially put um more capability, resource into a firm because they're seeing something with this with this data.

SPEAKER_00

Yeah, absolutely. So if you're not actually creating a decision and an action as a result of the forecast, you're you're obviously not getting the vast majority of the value off it. So with with our customers, we add an average of two months uh visibility. Um and what that means is 93% of the market only get confident once or two weeks for the end of the quarter, whereas with cluster, it's within the first few weeks of the quarter they get confident. And so what that means is most businesses will get to the end of the quarter, they will realise, as is often the case, that they're going to miss quota by five, ten, maybe twenty plus percent, and then only at that point they have to be re-planning when it's far too late, because their recurring revenue is lower, their revenue is lower, their balance sheet is weaker, um, etc. And then they might have to downgrade their future projections too. Whereas if you're running on cluster, then you're two weeks into the quarter, and we can say these two teams are going to miss their quota by 50%. And the reason for that is because they've got plenty of pipeline, but it's spread too thinly, or it's spread at the top of the funnel rather than the middle and the lower part of the funnel, or these reps are going to miss quota, etc. And once you get that information, what happens is you can pull in Linda top performing sales reps to help close the deals, make sure they close on time without discount, or you can reallocate the deals where you need to be, but you can level up at the reps that are forecasted to not perform so well, or you can you can move on from them. But either way, when you get to the end of the quarter, two plus months later, you're you're more than twice as likely to hit target. So our customers outperform the market by 103%, so over double in terms of attaining quota. And so that's turning the forecast into action. You've got two months of warning, you know exactly what to do, and you you're not retrospectively having to cut costs, downgrade your projections, etc. etc. You take those actions before the the bad event actually happens.

SPEAKER_01

Well, that sounds like something that that I'm sure those listening to to now want to achieve. But what could they do, do you think, right now to raise their forecasting game?

SPEAKER_00

So right now, if you literally had to start right now and raise the game, I would say firstly define what each word means. If you've got a commit forecast and the best case forecast, etc., what does that actually mean? Every deal that's in commit should have a fixed criteria. Now that can be different per business, but what's what's really important is it's identical across every deal that's in commit. And the same with best case. Every single deal across the entire organization should have an exact same meaning. The meaning of a forecast should be identical between the CFO and the CRO or VP of sales, so that everyone's talking about the same number, you're not talking about a conservative estimate or a higher number, etc. So once you've got that out of the way, which surprisingly is prevalent across companies in the market, and that's fine. We, you know, they just need to they just need to uh sort it out, which is quick to do. The next step would be to start with multiple simple forecasts rather than building one big complex one. And what's really powerful about this is if you if you base them on empirical evidence, like actual facts that no one can deny, and then calculate the number a lot more frequently and a lot earlier than you would be comfortable doing, you actually converge towards an extremely accurate number pretty quickly. So what I mean by that, you might have a stage-weighted forecast, say how much pipeline have I got at every stage that's closing this quarter and next quarter, get the conversion rate per stage, account for slippage, and then just have that number and snapshot that number. You might update that once a week or even once a day, and then do the same separately, maybe using your capacity, so the headcount of your team, and have multiple models. We've got some templates on our website to help with this, but essentially the a big problem people have is they only submit a forecast when they're confident in it, and they're only confident in it when it's far too late. So the the person submitting it is by definition almost doing themselves out of a job because by the time it's far too late, the CFO could have ballparked that number in their head anyway. Whereas if you sit there and say four months out, right, if our headcount is going to be X in four months, if our pipeline is going to be Y, and assuming the demand generation is anywhere between 10 and 20, spread is this. And even if you give a big spread with a big deviation, you say given the evidence at hand, this is what the outcome is going to be. And if you're really relaying that information every week up the chain of command, that becomes really powerful uh very quickly.

SPEAKER_01

What's going to happen next then in terms of forecasting? What what what's your uh view on the on the future there? And and are we getting to the stage where we're thinking, you know, full fully autonomous state?

SPEAKER_00

It's it's moving in that direction, certainly. And I think that'll make a lot of people happy. All the people have to pull together a forecast. It's a thankless task to spend lots of time doing something that's guaranteed to be wrong. Um, but essentially, the the perfect solution looking into the future is a solution where every number is at your fingertips live, and every single number is based on empirical evidence. So you can essentially ask the system what is going to happen to product line B in the United States over the next six months, and it would just come back and say, right, the spread is between this and this, the big needle moves are these three enterprise deals, your sales rep just got off a call with the prospect three minutes ago, and the and the call recording implied that it's moving forward, but they haven't updated the CRM yet, et cetera, et cetera. But you can ask them any question about any part of your commercial, any part of your company, and and the answer would be there live and in real time. I think in terms of autonomous, there's there's both the the information and gaining the edge, which is very powerful, but there's also the workflow of sales reps managing their deals, managing their pipeline, the sales leaders managing their deals and their pipeline, and having that operational cadence is is very important, not just for the forecast accuracy that it delivers, but also it's very tightly aligned with the operational cadence that private equity like to roll out uh post-transaction. So at the very beginning, you said, you know, what does private equity grade actually mean? The reality is when you've got the value creation plan, you've got the people plan, the technology plan, and all these other plans, it will be almost a spreadsheet or a table or a grid or a Gantt chart over time. Are we moving in the right in the right direction? Is the is the one-to-one meeting happening every day or week or month? And what was the output? Is that red, amber, green? You've just got a dashboard of red, red, amber, green of what's happening. It becomes very powerful. And so if you think back to the the kind of more workflow side that AI enables, a a forecast meeting could be that AI says, right, the empirical evidence says boom, boom, boom, boom, boom. This is your leads, this is your pipeline, this is your demand generation, there's a marketing campaign, or there's an event happening in two days. We expect that the pipeline created to be X. We believe that your forecast should be between eight and ten. What do you believe? And I've drafted some notes for you, but also want your your input input too. Um having that cadence from the people is very powerful. And just as any organization, uh whether you're you're in business, you're in the military, or otherwise, it's still a very important workflow that can be augmented by and the expectation on that commercial leader is gonna be going up, isn't it, essentially? Absolutely. But I think I think this is really powerful for commercial leaders because at the moment, the reason why they have the tendency sometimes to submit a higher number that's close to their target or to their plan is because you don't want to go to your CEO or your CFO and say, right, I'm forecasting I'm gonna miss. But actually, if you've got the evidence, then you can very aggressively go back and it actually helps everyone and say, My pipeline is X, my qualified pipeline is Y, marketing demand generation is diminishing, and and we've got two fewer people in my team than than we'd initially planned. All else being equal, I can't possibly hit the number that you also asked me to hit. These are your three options hire X people, spend X on marketing, or downgrade the plan, etc. And then everyone very quickly gets to a strategic conversation on what to do as quickly as possible, rather than different people floating around different numbers, which are then proven to be incorrect many months later, and then the IRR of the of the investment collapses.

SPEAKER_01

Just circling back in terms of your um your your way you started in your career around risk, and like every every day there's a there's something happening in the world that has some impact. Do you look at external factors as well within the forecasting and to support you know a company that could be very close to getting a deal and then disruption happens?

SPEAKER_00

Absolutely, absolutely. We we bake in data, it's it's very much driven by the internal data of the company, but we we can augment with with benchmark data across uh across the sectors, and then you know, looking forward, we will be incorporating macro data too. But I think what's really important is no matter how um no matter how far you move forward with this, if you present a number and then the evidence, the CEO or the CFO or the or the private equity sponsor or people on the board will be looking at the company and looking at the the numbers and saying, I know that I'm actually thinking about hiring X more people or opening an office here or cutting costs there, or whatever it might be. There's always going to be information that isn't in the in the data set, which is what makes the empirical evidence or you know, scientifically speaking, the assumptions. You say, These are my assumptions, this is the number. The person reading it can say, right, four of those five assumptions are correct, but one's wrong because I know I'm going to do something else in the future that no one else knows because I haven't written it down or put it in a database. Um yeah, going back to your point, we do augment it uh with a lot of other data, both for forecasting, but also benchmarking against the market, how you perform against your peers. Um but um at the end of the day, um it's it's the human that needs to have the accountability and responsibility. So uh so we present all the findings in a clear way back to them.

SPEAKER_01

That really fascinated conversation, an area I don't think gets enough airtime, and um well, I think that this will be uh you know case in point that we should probably do this again at some point. It it certainly has a major impact on value creation. So thank you for your time. In the show notes, I will make sure that you're uh that we have a link to cluster and and yourself on LinkedIn so people can uh get in touch. But for now, thank you, Dan, for your time and I'll see everyone again soon.

SPEAKER_00

Fantastic. Thank you, Steve.