The discovery call

We sit with the people who will use the screen and the people who own the data, and we listen. The recording becomes the specification.

In the room

• Edward, Managing Partner (the LP question)
• Priya, Finance Director (the numbers must tie out)
• Daniel, Head of IT and data (access and security)
• Sophie, portfolio analyst (builds the board pack today)

What they actually said

What we captured

• One morning cockpit with a red, amber, green heatmap across every company
• Budget versus actual, net debt, covenants, headcount
• Click any headline number to jump to the company behind it
• A data health layer so the numbers can be trusted

And where the data lives

From the same conversation we map every source the numbers come from: the systems, the grain, and how it all reconciles. That map is the data architecture.

[00:00] Muneer Right, I think we're all on. Thanks everyone. Quick intros, on my side, this is Bharathwaj from Indian Insights Company, IIC. They're the AI delivery team I mentioned; they'll be building the thing. I'm Muneer, pHRism, we've been looking after Meridian's payroll and finance-systems work for, what, three years now. Meridian side, do you want to go round?

[00:24] Edward Sure. I'm the managing partner here. I run the funds, I sit on most of the boards. I'm the one who gets asked by our LPs "how's the portfolio doing" and currently I can't answer that in less than about three days. So that's my interest.

[00:41] Priya Finance director, and I head up portfolio operations. I care about the numbers actually tying out, budget versus actual, net debt, covenants. I'll be the one telling you what a metric means.

[00:55] Daniel Head of IT and data. I'll be the one telling you where everything lives and, more importantly, where you can't go. Security, access, all that.

[01:04] Sophie And I'm the portfolio ops analyst. I'm the person who actually builds the board pack. So... I have feelings about this.

[01:11] Edward (laughs) She does. Go on, tell them.

[01:14] Sophie Honestly? Five days a quarter. That's what it costs me. Five working days, copy-pasting out of twelve different systems into one master Excel, re-keying it, then redrawing the charts in Power BI. And every quarter at least one company's numbers don't add up and I'm emailing their FD going "your revenue by segment doesn't match your total revenue, which one's right?"

[01:38] Bharathwaj That's exactly the kind of detail I need, so thank you. Let me set up how I'd like to use the time, and please cut across me whenever. I want to come out of this knowing three things: one, what each of you actually wants to see on a screen. Two, where the underlying data physically lives. And three, just as important, the open questions, the things we don't yet have an answer to. I'd rather surface those now than discover them on day five. If I get that, I can have a wireframe back to you inside two days.

[02:08] Edward Two days. Good.

[02:10] Bharathwaj Two days for the wireframe. A working build, live on your data, in about a week, but, and I'll keep saying this, the week only starts the day we get read-only access to the data. No access, no clock.

[02:24] Daniel Noted. We'll come to access.

[02:27] Bharathwaj Perfect. Edward, can I start with you, since you're the one with the LP problem. If you could glance at one screen in the morning, what's on it?

[02:36] Edward One screen. A cockpit. I want to look at it for ten seconds and know whether anything's on fire. So, first thing, a heatmap. All twelve companies, red, amber, green. Green means it's fine, leave it; red means call the chairman today.

[02:53] Bharathwaj What drives the colour? Red on what basis?

[02:56] Edward Off-plan on EBITDA, or a covenant issue, or they haven't reported. Priya can define the thresholds properly, but conceptually, performance, leverage, lateness.

[03:09] Priya I can give you the rules for that.

[03:11] Edward Then the top line. Consolidated revenue and consolidated EBITDA across the whole portfolio, against budget. In sterling. Everything in sterling, please, half these companies report in euros or dollars and I'm sick of doing the conversion in my head.

[03:28] Bharathwaj So FX consolidation to GBP, baked in. Hold that thought, it's going to come up again. What else?

[03:34] Edward Net debt across the portfolio. And a weighted-average leverage number, the whole book's leverage in one figure, weighted by size. Number of covenant breaches, as a count, right at the top, if that number's not zero I want to know before I've finished my coffee. Total headcount across the portfolio, because the LPs ask about jobs. And the one I really want, a value bridge. Where did we buy each business, what's it worth now, walk me from entry enterprise value to today.

[04:05] Bharathwaj A value bridge from entry EV to current. That's a portfolio-value-creation view, multiple expansion, EBITDA growth, debt paydown. Do you think in MOIC as well, or just absolute value?

[04:17] Edward Both, ideally. But the absolute bridge first. The MOIC's a nice-to-have on top.

[04:23] Muneer Just to translate for the build team, when Edward says "entry EV" he means what they paid for the business including its debt, at the point of acquisition. That's captured in their deal records, not in any of the portfolio companies' accounting systems. Different source. Flag that, Bharathwaj.

[04:41] Bharathwaj Flagged, entry data is a GP-side master record, separate from the monthly accounts. Good. Priya, over to you. The finance and ops view.

[04:51] Priya Right, mine's less glossy and more... forensic. Budget variance, every company, actual versus budget, on revenue and on EBITDA, in pounds and in percent. I want to sort by who's furthest off plan.

[05:06] Bharathwaj Variance, sortable. Period, month or quarter?

[05:10] Priya Both. Management accounts come in monthly. The board pack is quarterly. So I look at it monthly and I report it quarterly. I'd want to flip between the two.

[05:21] Bharathwaj So a monthly-versus-quarterly cadence toggle. That's actually a design decision I'll note, I'll come back to whether the default view is monthly or quarterly.

[05:30] Priya Default monthly for me, quarterly for him (gestures at Edward). See, that's your first problem.

[05:36] Bharathwaj (laughs) That's a real one, and it's fine, role-based default. Keep going.

[05:41] Priya Covenant headroom. By facility. I don't just want "breach yes/no", I want how close each one is. Company's got a senior term loan with a leverage covenant at, say, 4.5 times, they're sitting at 4.2, that's a sweat, I want to see the 0.3 of headroom shrinking before it goes red. Same for the interest cover covenants. We track all that in a separate spreadsheet today, which is its own nightmare.

[06:08] Bharathwaj So headroom per facility, per covenant type, leverage and ICR, not just the breach flag. And it's currently in a standalone sheet, divorced from the financials.

[06:18] Priya Completely divorced. Which is mad, because the leverage covenant is net debt over EBITDA, and net debt and EBITDA are sitting right there in the management accounts. They should be the same number. They're maintained separately and they drift.

[06:32] Bharathwaj That's a reconciliation we can compute and check, recompute the leverage from the accounts and compare it to what the covenant sheet says. Where they disagree, that's flagged. I'll come back to that under data-health.

[06:43] Priya Yes. Then, a list. The late and missing submissions. Who hasn't reported, who reported late, who reported but it's still marked draft. Right now I find that out by Sophie chasing people.

[06:56] Sophie It's always the same two companies.

[06:58] Priya It is always the same two. And one of them, the carve-out, genuinely went missing for a month last year, no submission at all, and we didn't catch it until the quarter close. That cannot happen again.

[07:10] Bharathwaj So a submission-status tracker, Final, Late, Draft, Missing, surfaced, not buried. Got it. Anything else on the finance view?

[07:18] Priya Adjusted EBITDA. The add-backs. When we report EBITDA to a lender or an LP it's adjusted, we add back one-offs, normalisations, management adjustments. I need to see the bridge from reported EBITDA to adjusted EBITDA, the add-backs itemised, per company. Because the covenant tests are usually against adjusted EBITDA, and people fiddle with add-backs.

[07:42] Bharathwaj An adjusted-EBITDA bridge with the adjustments itemised. And presumably you want it consistent, the same definition of "adjusted" everywhere.

[07:50] Priya That is the whole battle. Right now company A's FD and company B's FD adjust differently and nobody's signing off on one definition.

[07:58] Bharathwaj I'm going to write that down as an open item, who signs off the metric definitions. Because if EBITDA isn't computed one consistent way, the consolidated number is fiction, and the dashboard would just be presenting fiction faster. Who owns those definitions today?

[08:13] Priya Me. In principle. In practice it's a Word document from 2023 that nobody reads.

[08:19] Muneer We can help formalise that, by the way, that's squarely a pHRism finance-systems piece. We've done the chart-of-accounts harmonisation work before.

[08:28] Bharathwaj Good, so there's an owner. Sophie, you're the one who lives in this. What do you need?

[08:34] Sophie I need to stop reconciling by hand. That's the headline. But specifically, I need to be able to drill into a single company and see everything about it on one page. Its P&L. Its revenue broken down by segment, product, channel, geography. Its headcount by department. So when Edward says "what's going on at company seven" I click company seven and it's all there, instead of me opening four files.

[08:59] Bharathwaj Single-company deep dive, P&L, revenue by segment, headcount by department, all on one drill-through page. And you want to get there how, click a company on the heatmap?

[09:09] Sophie Click it on the heatmap, yeah, or pick it from a list. Either.

[09:13] Bharathwaj And the reconciliation pain, describe the actual checks you do by hand.

[09:18] Sophie Okay so. Every month: does the sum of revenue-by-segment equal the total revenue in the P&L? Half the time a segment's just... missing. Someone forgot to fill in "concessions" or whatever and the segments add up to less than the top line. Then: does the payroll cost in the HR system match the staff costs line in the accounts? Those two are never exactly the same and I never know if it's a real difference or a data error. Then duplicates, sometimes a company submits twice and I get two rows for the same month with different numbers. And FX, occasionally a rate looks insane, like someone's fat-fingered a euro rate, and the whole company's sterling numbers go haywire.

[09:58] Bharathwaj This is gold, honestly. Every one of those is a rule I can automate. Segment sum equals total revenue. Payroll sum reconciles to staff costs within a tolerance. No duplicate periods. FX rate sanity, flag if a rate jumps more than, say, a quarter month-on-month. That whole list becomes an automated exceptions worklist. We call it the data-health tile.

[10:22] Sophie If that exists I will personally buy you dinner.

[10:25] Bharathwaj (laughs) Let me explain the data-health tile properly, because it's the bit that's genuinely different and it ties to what Priya said about covenants. The tile makes a distinction that most BI tools don't. There are two kinds of "red" on a dashboard. One is a data problem, the segments don't sum, there's a duplicate row, the FX rate is stale. That's a bug. You fix the data. The other is a genuine business risk, a company has actually breached a covenant, EBITDA has actually collapsed. That's not a bug. That's real, and a partner needs to see it and act. The tile separates those two. "These eight things are data errors, here's the fix. These three things are real, escalate them." Most dashboards throw both at you in the same red and you can't tell signal from noise.

[11:10] Edward That's the bit I've never had. Last quarter we nearly called a chairman over a "breach" that turned out to be a spreadsheet error in our own model. Embarrassing.

[11:19] Bharathwaj Exactly the failure the tile prevents. Right, Daniel. The unglamorous half. Where does this data actually live?

[11:27] Daniel Brace yourself. Twelve companies, and near enough twelve different systems. Off the top of my head, a couple are on SAP, two or three on NetSuite, several of the smaller ones on Sage, one on Microsoft Dynamics, the most recent acquisition is still on QuickBooks, and the people data, headcount, payroll, that's split, some's in Workday, the payroll runs through ADP.

[11:51] Muneer And pHRism manages the Workday-and-ADP layer for several of them, so that side we can get you into relatively painlessly. The ERPs are the harder ask.

[11:60] Bharathwaj So there is no single source of truth. There are twelve sources, in different shapes, different currencies, on different reporting clocks. That's the real shape of the problem, and to be clear, that's normal, every PE house I've seen looks like this. The dashboard's value isn't in any one system, it's in marrying the twelve and catching where they disagree. Question: do the management accounts at least come to you in a common format today, or do you get twelve different exports?

[12:24] Daniel Priya's team gets a monthly pack from each company. It's meant to be a common template. It... mostly is. There's drift.

[12:33] Priya It's a template they all received. Whether they fill it in the same way is another matter.

[12:39] Bharathwaj Understood. So we could start from those monthly packs as a landing point rather than wiring directly into twelve ERPs on day one, that's a phasing decision. But Daniel, on access, what's realistic? What I need is read-only. I never write back to your systems.

[12:56] Daniel Read-only is the only thing I'd entertain, so that's good. But it's not just my decision per company. Each portfolio company's data is technically theirs, we have information rights as the owner, but granting a third party access, even read-only, may need each company's FD to sign off, and probably an NDA per company. And I need to know where the data goes. Where does your platform sit? Which region?

[13:22] Bharathwaj Fair, and important. Data residency, we can host in-region, UK or EU, your choice; I'll confirm the specifics in writing. Nothing leaves the region you nominate. On access, what I'd suggest, and this de-risks your sign-off problem, is we don't start on live company data at all. We build the wireframe, and even the first working version, on synthetic or anonymised data. Then there's nothing confidential in flight while the lawyers and the FDs do their thing.

[13:50] Daniel That genuinely helps. If the first build is on dummy data I can run the access approvals in parallel instead of as a blocker.

[13:57] Bharathwaj That's the idea. And it means the two-day wireframe has zero dependency on your security process. The week-to-live clock starts when real read-only access lands, but the wireframe doesn't wait.

[14:09] Daniel One more thing, and it's why we're even on this call. We're on Power BI now and the licensing is killing us. It's per seat. Every partner, every analyst, every company FD I want to give a view to, that's another licence. I'm not paying for forty seats so people can look at a chart.

[14:28] Bharathwaj That's central to what we do, so let me be direct about the differentiator. Our front end is AI-generated and it is licence-free on the viewing side, there's no per-seat charge for people to look at it or share it. You can roll it across all twelve companies and out to the FDs without the seat-sprawl. And it's editable in plain English, and I mean by you, not by us with a two-week change request. Priya, you'd literally type "show me EBITDA versus budget for just the euro-reporting companies over the last eight quarters" and it reshapes the view. No ticket to me.

[14:58] Priya Hang on, I can type that? Not raise a request and wait?

[15:02] Bharathwaj You type it. That's the point. The thing that takes Sophie five days and me a change-request in the old world, you ask for it in a sentence.

[15:11] Sophie ...okay, now I'm paying for dinner.

[15:14] Edward (laughs) Let's see it first.

[15:17] Bharathwaj Quite right, see it first. Let me make sure I've got the open questions, because these matter as much as the wishlist. Can I read them back?

[15:25] Edward Go on.

[15:27] Bharathwaj One, data access ownership. Who grants read-only access, and is it per company with an NDA each? Daniel owns scoping that. Two, single source of truth versus twelve: confirmed it's twelve, and we need to decide whether we land off the monthly packs first or wire ERPs directly. Three, cadence: monthly versus quarterly, with role-based defaults. Four, the chronically late reporter: how do we want the dashboard to handle a company that's missing, show a stale figure, or a big "MISSING" flag?

[15:55] Priya Big MISSING flag. Never show a stale number dressed up as current. That's how you mislead a board.

[16:02] Bharathwaj Good, that's now a rule, not an open question. Five, the authoritative GBP FX rate. This is a real one. When you consolidate to sterling, which rate, sourced from where, and dated when? Month-end spot? An average? Whose feed?

[16:18] Priya ...that's a good question and I'm not sure we're consistent. Probably month-end spot, but I'd have to check what Sophie actually uses.

[16:26] Sophie I pull it off the same financial data site each quarter. Month-end. But there's no policy, it's just what I do.

[16:33] Bharathwaj So that's an open switch, we need an authoritative FX source and rule, signed off, otherwise every consolidated number is slightly arguable. I'll list it. Six, metric definitions sign-off, especially adjusted EBITDA and the add-backs, so it's computed one way everywhere. Priya owns that, pHRism assisting. Seven, security, NDA, data residency, Daniel owns, we host in-region, and we de-risk it by building on synthetic data first. Did I miss anything?

[16:58] Edward Which fund. We've got three funds. Don't try to do all twelve companies across all three on day one, pick one.

[17:06] Bharathwaj Excellent point, I'll add it, scope decision, which fund or funds we start with. Do you have a preference?

[17:13] Edward The flagship buyout fund. It's the bulk of the value, it's got the companies with the covenant pressure, and it's the one the LPs ask about. Get that right and the rest is the same pattern.

[17:24] Priya Agreed. And it's got the late reporters and the breaches, so it's a proper test.

[17:30] Bharathwaj Then we'll wireframe against the flagship fund as the pilot, designed to extend to the other two. One small tangent if I may, Muneer, you said pHRism already holds the Workday and ADP side for several companies. If we can prove the headcount and payroll-cost reconciliation early, payroll cost from the HR system tying to staff costs in the accounts, that's a quick, visible win and it's data we can actually reach without waiting on twelve ERP approvals.

[17:57] Muneer Agreed, and we can stand that up fast. The HR-to-finance tie-out is bread and butter for us. That'd be a good first proof point, and it's exactly the kind of thing that goes wrong quietly: a payroll cost that's ten percent off the staff-costs line and nobody notices.

[18:13] Daniel That'd genuinely be useful. We've never reconciled those two systematically.

[18:18] Bharathwaj Then we'll feature it. Let me play back the full picture so the wireframe matches your heads. There's a top-level cockpit, heatmap of all companies red/amber/green, consolidated revenue and EBITDA versus budget in sterling, portfolio net debt, weighted-average leverage, covenant-breach count, total headcount, and the value bridge from entry EV. That's Edward's ten-second glance. Then a finance-and-ops layer, budget variance sortable by worst, covenant headroom by facility for both leverage and ICR, the late/missing/draft submissions list, and the adjusted-EBITDA add-back bridge. That's Priya. Then drill-through to a single company, P&L, revenue by segment, headcount by department, that's Sophie's deep dive. And threaded through all of it, the data-health tile: the automated exceptions worklist that separates data errors to fix from real business risks to escalate. Currency throughout is GBP-consolidated. Cadence toggles monthly/quarterly. Have I got it?

[19:08] Edward That's the most anyone's understood it in one go, yes.

[19:12] Priya The only thing I'd add, when it flags a variance or a breach, I want to click straight from the cockpit number into the company behind it. Don't make me navigate.

[19:21] Bharathwaj Click-through from any headline number to the company driving it. Adding it. Sophie?

[19:26] Sophie Just, please make the exceptions list exportable. When I chase a company FD I need to send them the specific thing. "Your November segments are short by this much."

[19:36] Bharathwaj Exceptions list exportable, with the specific figures. Done. Daniel, anything on the non-functional side I've missed, performance, audit?

[19:44] Daniel Audit trail. If a number changes I want to know it changed, when, and from what source. Auditors will ask. And no surprises on where it's hosted, confirm the region in writing.

[19:55] Bharathwaj Audit trail on data lineage and changes, region confirmed in writing. Noted both. Muneer, anything from the relationship side before we close?

[20:03] Muneer Just to manage expectations on both sides, the two-day wireframe is on synthetic data, it's a picture of the dashboard, not your live numbers. The live build is a week from the day Daniel gets you read-only access. So the single most useful thing Meridian can do this week is start the access conversation with the flagship fund's companies. That's the long pole.

[20:25] Daniel Understood. I'll start the access scoping for the flagship fund companies, work out per-company sign-off and NDAs, this week.

[20:33] Bharathwaj And to be totally clear on sequencing: wireframe in two days regardless; live build clock starts on access; we build on synthetic or anonymised data until the real access and NDAs are in place, so security is never the thing that holds up seeing progress.

[20:48] Edward That works for me. When do we see the wireframe?

[20:51] Bharathwaj Two working days. Today's Friday the twelfth, so I'll have it to you by Tuesday the sixteenth, and I'd suggest we get back together that day or the day after to walk through it.

[21:02] Edward Tuesday review. Book it.

[21:05] Priya I'll get you the metric definitions before then, the colour-coding thresholds for the heatmap, and a first cut at the adjusted-EBITDA add-back rules. The FX policy I'll need a bit longer; let me confirm what Sophie actually pulls and we'll formalise month-end spot from a named source.

[21:21] Bharathwaj Perfect. Even a first cut of definitions makes the wireframe sharper. Sophie, could you send me one redacted, anonymised example of a current monthly pack? Company name stripped, just so I can see the shape of what you receive. It'll make the synthetic data realistic.

[21:37] Sophie Yes, I'll anonymise one and send it over today.

[21:41] Bharathwaj Brilliant. Then I think we've got a genuine spec. Muneer, you'll coordinate?

[21:46] Muneer I'll pull together the action list and send it round this afternoon, and I'll lock the Tuesday review in everyone's calendars.

[21:53] Edward Good. This is the first time I've felt like the quarterly scramble might actually end. Don't let me down.

[21:59] Bharathwaj Tuesday. You'll see it. Thanks all.

[22:03] Muneer Thanks everyone. Speak soon.