A full review across both halves of the engagement: the data we build and score for you, and what then happens to it on the phones and in the CRM. The data is dense with genuine, high-net-worth, asset-relevant investors — including people who already own racehorses. The results don't yet match that calibre, and this review shows precisely where, between a strong lead and a closed deal, the value is being left on the table.
Three findings, evidenced throughout this document. We profiled the calls end-to-end and read the live pipeline. The leads hold up under scrutiny — the losses happen after the lead is delivered.
We located 640 of the 641 brochure leads in the call records and profiled every one with a substantive transcript — 349 dossiers in all. Of these, 153 contained a real four-minute-plus conversation (long enough to assess), and of those 68% are Tier 1 (the strongest band), 87% carry a clear wealth signal — business owners, property and precious-metals investors, finance professionals — and 14 already own bloodstock.
Whatever data tier is dialled, results vary widely by who is on the call. The team average is respectable and one agent (Kate) is exceptional. The variable is how the call is handled and tagged — not the data.
~48+ live calls were logged as "no answer"/"answering machine"; ~670 brochure leads sit in "Lost"/"Inactive" against 17 wins; and the CRM records €0 on every win, so revenue can't be measured.
Self-reported wealth figures throughout are treated as claims, not verified facts; the robust signal is the consistency of the pattern across hundreds of records. A genuine, correctly-tagged "not interested" is never counted as a mis-tag.
Thirteen years, one specialism: investor data. At the centre is Axiom — our engine that holds real investor records, scores every one, and works out which should go where. It is built from thirteen years of who actually invests, so it can't be bought off a shelf.
Hints at the decade someone was born.
Old ISP vs fresh Gmail — age + tech-habit signal.
A real clue to the wealth likely held.
From a file full of investors → the rest probably are too.
The single strongest signal they'll do it again.
A fresh record behaves differently to a worked one.
Inferred from name, email vintage and history — sets life-stage.
Occupation and industry — a direct read on earning power.
Modelled earning bracket — how much is realistically investable.
How warmly the record has responded so far.
Age, job title, income band, engagement level and recency are examples of the 17 wealth tags Axiom assigns every record. The engine weighs them into one score out of 100 (source counts most, then email, then name); the score sets the tier. Axiom scoring model
Each tier down is ~10× bigger but lower-converting. The agents work the people the engine surfaces as lookalike audiences — modelled on real, converted investors and drawn fresh from the same 13-year base. This is the system we've always run: pre-AI we hand-built fresh lookalike / wealth-tagged lists every day; Axiom now does it continuously. Because the lists are fresh and unworked, lookalikes convert as well as — often better than — the seed audience they mirror, and the data is worked, not sprayed.
We matched 640 of the 641 brochure leads to their call records and profiled all 349 with a substantive transcript. Of the 153 that contained a real four-minute-plus conversation, the calibre is striking — and the strongest proof is the perfect-fit segment: prospects who already own racehorses.
| Prospect | What they told the team (verbatim / paraphrase) |
|---|---|
| Roger White | "I own six of my own, so I don't need any more" — owns six racehorses |
| John Cotton | "my wife and I own a few horses" — buys yearlings at £100–120k; owns a 100-year family company |
| Giles Paul | Horses placed across Paul Nicholls, Nicky Henderson, Jamie Snowden & William Haggas |
| James Cairns | "I've owned racehorses… various syndicates or ownership" |
| Barry Leslie | Household owns multiple shares (incl. Speedstep); "very much an equine family" |
| Mark Westbrook | "I've owned a few racehorses"; also owns 50% of a 160-staff manufacturer |
| Andy Newton · Lee Cartwright · David Craster | Prior syndicate share (accountant) · prior racehorse shares + property portfolio · racehorse shares + cash-bought property |
| …and 5 more | Chris Evans, Stuart Hinton, James Badham, Alexander Haywood, Dean Rogers — current or former owners (14 in total) |
| Prospect | Wealth profile (as described on the call) |
|---|---|
| John Richardson | "near on 30 different businesses that we've invested in" — properties, gold, crypto; a trade-show / Dragon's-Den-style exit |
| Francis Ball | "I was UK MD of Costco until 2006" — software-company director and non-exec of an IFA |
| Barry Maloy | "not long bought an additional HMO… most of my money is probably in Tesla" — property portfolio, businesses, crypto |
| Robin Flannery | Farming-company owner; "invested £150k" in physical gold, "increased in value by 1.2 mil"; whisky |
| Anthony Malone | "shares in ExxonMobil… about 4,000" plus two rental properties and premium bonds |
| Ricky Assi · Stuart Albert | Multiple properties, 30 yrs of stocks/shares, crypto, whisky casks · retired engineer with a Hargreaves Lansdown FTSE/S&P portfolio |
| Robin Humble · Iain Purves | States he sold a pharma company for ~£750M; commercial property, EIS, gold, classic cars · founder of a high-end property-development firm |
| Michael Harris · Shane Zhang | 35 years a property investor with a corporate-finance firm and Dubai interests · a 30-year Mayfair shipbroker who also collects fine art |
| Joshua Hall · Alexander Haywood | Holding Bitcoin since 2013 alongside blue-chip equities · owns a company that builds data centres, plus whisky-cask investments |
| Prospect | Profile (as described on the call) |
|---|---|
| Garry Schofield | "I am a financial adviser… I prefer stocks and shares. I've done it for forty years" |
| Jim Lavery | "I was an area director for Lloyd's Bank. I had wealth managers working under me" |
| Martin Smith | "I was an IFA… I'm retired" — gold investor, affluent lifestyle |
| Guy Mason | Private-equity consultant; prior start-up investments; "liquid… good risk appetite" |
No reasonable reading of these calls supports the idea that the data is poor. It delivers the audience the product needs — repeatedly, across every agent's list — including, uniquely, people who already own racehorses.
These weren't quick brush-offs. The Tier 1 conversations averaged 17 minutes on the phone, and 89% were followed up with at least one email (3 on average). The engagement is real on both sides — the prospect gives a long, candid account of their wealth, and the team logs follow-up. The question the rest of this review answers is what happens after that strong start.
The fairest measure of effort is billable time — the hours the agents are actually paid against. Across the whole window the timesheet reports it (late May onward, once it begins populating), the four agents logged about 133 billable hours between them — ~126 of those in June, the first clean full month. Set against the 59,328 dials placed across the campaign, that is the number that tells you how much real calling work the data received.
| Agent | Billable | of which live talk | Calls | Contacts (≥conn.) | Talk share |
|---|---|---|---|---|---|
| Fin | 53.4h | 30.3h | 5,854 | 4,995 | 57% |
| Josh | 33.3h | 22.1h | 2,939 | 2,327 | 66% |
| Kate | 30.0h | 24.0h | 1,872 | 1,158 | 80% ★ |
| Georgia | 16.8h | 11.1h | 1,663 | 1,234 | 66% |
| Team | ~133h | ~88h | ~12,328 | ~9,714 | ~66% |
Calls and contacts above are the billable-window count (late May → June), not the full-campaign 59,328 dials in Section 4 — they cover only the period the timesheet records billable time.
Against a roughly 8-hour paid day, June billable time runs ~1.5–2.8 hours per agent per day — under a third of a shift. A dial averages ~34 seconds and only ~25% last 10 seconds or more, so most are short no-answers, exactly as a power-dialer produces. The realistic ceiling on output is being set by billable calling hours as much as by data or skill.
Of the ~133 measured billable hours, about 88 were live talk and ~46 were dialing and wrap — so roughly two-thirds of paid calling time was actual conversation, the rest the unavoidable overhead of a power-dialer. But the split varies sharply by agent, and it tracks the conversion story exactly: Kate turns 80% of her paid time into live talk; Fin only 57% (Josh and Georgia ~66%). The fastest dialler spends nearly half his billable time dialling rather than talking. The workload was also concentrated — Fin alone accounted for ~40% of the team's billable hours and ~48% of its calls (5,854 of 12,328). Georgia was active only across weeks 22–23 (~16.8h) and then drops out, so from mid-June this was effectively a three-person effort, settling at roughly 25–30 billable hours a week between them (Fin ~13h, Josh ~8–9h, Kate ~6–8h). The realistic way to lift output isn't more dials — it's more agents sustaining steady billable hours, since two-thirds of that time already converts straight into conversation.
Plotted week by week, the shape is clear. Team billable effort peaked in the first full week all four were active (~42h between them) and then fell to ~25–30h — roughly a 30% drop — and stayed there. Two things drove it: Georgia billed two weeks and then dropped out entirely (she'd also worked ~7h the week before this window), and Fin tapered from a 16-hour peak to ~11–13h. Josh and Kate held broadly steady at ~6–9h. The decline isn't a data effect — it's hours coming off the phones. Weekly billable, Aircall timesheet; wk 29 Jun omitted as incomplete.
From 59,328 dials came 768 brochures and 17 wins. Measured per dial the rate looks low — but a dial is mostly a ring-out. Measured per conversation, the data converts well, and how the call is handled is the single biggest lever.
| Agent | Dials | Contacts (≥10s) | Brochures | Per dial | Per contact |
|---|---|---|---|---|---|
| Kate | 12,940 | ~3,235 | ~346 | 2.67% | ~10.7% ★ |
| Georgia | 3,501 | ~875 | ~44 | 1.27% | ~5.0% |
| Fin | 30,499 | ~7,625 | ~284 | 0.93% | ~3.7% |
| Josh | 12,387 | ~3,097 | ~92 | 0.74% | ~3.0% |
| Team | 59,328 | ~14,832 | ~768 | 1.29% | ~5.2% |
Per conversation the team converts ~5.2%, not 1.29% — and Kate converts ~10.7%, roughly 3× the team. Put as conversations-per-lead: Kate lands a brochure roughly every 1 in 10 conversations; Fin needs ~1 in 33. On 22 Jun her calls averaged 4m 50s against Fin's 46s — she dials far fewer but holds real conversations. The lever is the call, not the data.
The mechanism that buries good leads and makes the contact rate look worse than it is: answered, engaged conversations recorded as "no answer" or "answering machine".
An asset-relevant contact is dialled and picks up.
Sometimes ten minutes; sometimes describing significant wealth or existing ownership.
The lead is buried and the contact rate looks artificially low.
A ten-minute, fully two-way conversation with Steve Jackson — who already owns live racehorse-syndicate shares (Old Gold Racing). He engaged warmly, named his horses, was recommended a £2,000 share and agreed to a Monday follow-up — and was logged as a machine that picked up.
| Agent | Clean mis-tags | Clearest examples |
|---|---|---|
| Georgia | ~25 of ~95 read | Lambert, Patel, Stone, Nick (FCA adviser), David L Martin — answered, tagged "no answer" |
| Fin | 7 | Steve Jackson (above); Coram, Darren, Carl, Carroll "answering machine"; Robert Mayson (family office) "no answer" |
| Josh | 6 (+2) | Alistair Morris, Simon Powdrell, Robert Sidaway — engaged HNW calls tagged "no answer" |
| Kate | 1 + soft | Anthony Holness — live two-way call tagged "answering machine" |
| 26 Jun sweep | +9 | John "no answer"; Martin Kinsella "answering machine"; Nigel Troup & Mike Plaskett "callbacks" |
| Total surfaced | ~48+ | and rising as more transcripts are reviewed |
This is the most common defect across the whole review. It simultaneously loses the lead and manufactures the impression that the data has a low contact rate. The 100-call audit found a 61% connect rate — above the norm for HNW outreach, and higher still once these records are tidied. Most of these calls are recoverable.
And it costs you twice. A mis-tagged lead isn't just buried — it starves the data engine. When a real, engaged lead is logged "no answer", the system never learns that record converted, so it can't go and find more leads like it. Disciplined tagging is what makes the data sharpen over time; without it the engine is flying blind. It also means the weaker-looking agents' true performance is genuinely unknown — where the tagging is wrong, we can't tell how well they actually did.
The CRM holds ~2,452 open deals. In the brochure-led funnel, most leads are parked, and across the book most receive very few touches.
| Stage | Leads | Read |
|---|---|---|
| Brochure → Attempting → Recommendation → Paperwork | ~164 | In flight |
| Client / Win | 17 | ~2% of the funnel |
| Inactive | 186 | Parked |
| "Lost" stage | 484 | Parked |
| ~670 parked (≈79%) vs 17 wins | 851 | The core question |
| Owner | Open deals | Wins | Avg emails | Avg activities | Pattern |
|---|---|---|---|---|---|
| Finley | 21 | 1 | ~4.6 | ~7.1 | Highest touch counts of any working agent (~4.6 emails, ~7.1 activities per deal) — but on a small owned book of 21. As a high-volume cold-caller these are cold brochure leads he's nurturing, not an existing-investor book; the depth per deal is real, it's just applied to very few of the leads he generates |
| Callum | 183 | 1 | ~1.1 | ~13.3 | Phone-led — high call activity, light email |
| Charley | 541 | 9 | ~2.6 | ~2.0 | Largest book + most wins; works a subset |
| Josh | 85 | 0 | ~1.6 | ~2.4 | Bimodal — several with zero touches |
| Kate | 39 | 1 | ~1.1 | ~1.1 | Mostly single email then parked |
| Admin pool | 1,578 | — | ~0.1 | ~1.3 | One auto-dial each, no human follow-up |
Most leads never entered the process we built. Of the 851 brochure leads, only ~164 were ever moved into the structured stages (Brochure → Attempting → Recommendation → Paperwork); the rest sit at entry or parked. The deal-stages aren't "unused" — most leads were simply never put through them.
Two-thirds of the entire pipeline (~1,578 deals) sits under an admin account with ~0 emails and a single logged touch — leads that received one automated dial and no human follow-up. The five working agents nurture a comparatively small slice (~870 deals). And because the structured deal-value field is €0 on every win, the pipeline can't total revenue per win, agent or month — so the operation can't yet measure its own results.
Touchpoint averages are indicative 8-deal samples (10 for the admin pool) read live from Pipedrive on 23 Jun 2026 — they characterise the pattern, not a full census. "Activities" includes automated call-logging, so high counts are partly call-driven.
One consistent picture across the engine, the calls and the CRM. The data is genuinely high-net-worth and asset-relevant — including racehorse owners. What's costing results sits downstream of the data: in the conversation, the connected hours, the dispositioning, the follow-up and the pipeline hygiene.
None of these is a data problem — they are all execution and hygiene, and all fixable: how the call is opened, how it's tagged, how leads are routed and followed up, and how the pipeline measures itself.
Worked the way the data is designed to be, the same pipeline delivers materially more from the same leads. The upside is sitting in the records you already have.
Rigour: wealth figures are prospect self-reports (claims), not verified facts; the robust signal is the pattern across hundreds of records. Call/hours figures from Aircall (to 18 Jun); pipeline counts read live from Pipedrive 23 Jun; 640 of the 641 brochure leads matched to their call records; all 349 with a substantive transcript profiled, of which 153 contained a four-minute-plus conversation scored against the tier rubric. The lifespan model is illustrative; per-agent contact figures use the team ≥10s ratio applied per agent.
Public record, not opinion. Any investor doing basic due diligence on "Opulence Thoroughbreds" finds a company in liquidation — and the current approach keeps the old name, email and domain, linking the relaunch straight to it.
| Status | Liquidation |
| Resolution to wind up | Passed 25 September 2025 (extraordinary resolution, form LRESEX; filed 2 Oct 2025) |
| Liquidation | Voluntary liquidator appointed + Statement of Affairs filed 1 Oct 2025 — a creditors' voluntary liquidation |
| Before that | Compulsory strike-off began (First Gazette 22 Jul 2025), then discontinued and replaced by the wind-up |
| Accounts | Micro-entity only; last made up to 31 May 2023; now overdue |
| Sister entity | Opulence Thoroughbreds Breeding Limited (14520967) — Active, but shares the brand/name |
The due-diligence point: the standard practice when relaunching after a wind-up is a clean break — new entity, brand, domain and contact details — precisely to sever the association, because the association doesn't disappear when the old entity does. Dissolved and liquidated companies remain on the public register as historical entries, and search engines, reviews and press make no distinction between the old business and the new one, so the prior history follows the name. Retaining the Opulence Thoroughbreds email and domain does the opposite of a clean break: it preserves the very link the relaunch should cut, and it's discoverable in minutes through a Companies House lookup, a search of the domain, or a WHOIS history check — the routine first moves of anyone vetting a counterparty before committing capital.
There is also a direct legal dimension. This was a creditors' voluntary liquidation — an insolvent wind-up — and the relaunch runs with the same directors, same team, same office and the same email identity. That puts both triggers for Section 216 of the Insolvency Act 1986 squarely in place: it restricts anyone who was a director in the twelve months before an insolvent liquidation from being involved, for five years, in a business trading under the same or a similar name — and the prohibition expressly extends to trading styles and branding, not merely the registered name (the category an email address and domain fall into). Breach is a criminal offence, carrying fines, potential disqualification and personal liability for the new business's debts. On the facts as we understand them this isn't just reputationally risky — continuing to trade under the Opulence Thoroughbreds identity looks like a live Section 216 exposure, and should be put to a solicitor before the relaunch goes further. Evidence: Companies House CRN 12607224 (overview + filing history); Insolvency Act 1986 ss. 216–217 and Gov.uk guidance on prohibited names; Companies House guidance on dissolved-company records remaining publicly searchable. Legal exposure stated on the facts available — confirm with counsel.
The question was whether the content sent to leads adds value or is filler. The brochure itself does go out — to most leads. The gap is everything that should come after it.
The reach is fine — the nurture is missing. Read directly from the Pipedrive book, roughly four in five brochure leads have email correspondence logged (the brochure send plus replies), so this isn't a reach problem. What's absent is a structured nurture programme: beyond the initial brochure send, the leads we traced received no designed multi-step sequence — at most ad-hoc replies — nothing that deepens understanding of the model, builds the investment case, or keeps a prospect warm between calls.
The right content is in use — but it's thin. A templated "Introduction to Opulence Bloodstock" email carries the Opulence Bloodstock brochure (the broodmare-syndication overview Cloudworkz produced, with illustrative returns), converted clients get a "Welcome to Opulence Bloodstock" onboarding email, and single-horse packages (Sea The Boss, Fixette) go out for specific shares. So the on-brand brochure is being sent. The gap is depth and reach: it's essentially an intro + a per-horse sell-sheet with little mid-funnel nurture that deepens understanding or builds desire over time between calls. The legacy racehorse-ownership brochure also still exists and is off-message if used. Read directly from the Opulence Pipedrive (record-linked emails + attachments).
This review keeps arriving at the same root: the data is strong, but there's no joined-up system around it to turn conversations into conversions or to measure what's happening. These are the foundations that close that gap — most of which you can build with your own team; we only take on the parts you want us to.
| Recommendation | What this review found it fixes |
|---|---|
| Investor personas | The data is full of HNW, asset-relevant investors — even racehorse owners — but nothing documents who they are, so calls and content can't be tailored. Define the four-plus personas and the profile behind each. |
| CRM structure with real deal stages | ~670 of 851 brochure leads sit parked and only ~164 ever entered the structured stages. Clear stages and thresholds let a lead move through a journey instead of stalling. |
| Call dispositioning & phone discipline | ~48+ live calls were logged as "no answer"; conversion reads 0% for every agent because outcomes aren't tagged. Proper call setup and dispositioning makes performance measurable from day one. |
| Lead nurturing / content bank | ~80% of leads receive the brochure, but there's no drip that builds the case between calls. A nurture programme keyed to persona and journey stage. |
| Sales playbook & training | Conversion swings ~3× by who's on the call (Kate ~10.7% vs ~3% per contact). A documented playbook spreads the best approach across the team. |
| Management reporting & dashboard | €0 is recorded on every win, so revenue per win, agent or month can't be totalled. Reporting turns this report's manual reconstruction into a live number. |
| ROI calculator & marketing roadmap | Gives every conversation a consistent value story, and sequences the work above into a plan. |
| Brand guidelines | One consistent, on-message identity across brochure, emails and follow-up (and the clean-break rebrand the corporate-status finding above calls for). |
None of this is exotic. It is the standard foundation under any sales-and-marketing operation that converts well — persona → CRM → process → content → measurement — built once and then maintained.
Recommendations don't stall for lack of merit; they stall in the decision. Usually a poor decision is the product of poor data — but here the data has been on the table, and the call has repeatedly gone the other way. That points to two things worth naming plainly: the decision process itself, and decisions taken outside the area they properly belong to.
| The decision | What was known at the time | Where it has left things |
|---|---|---|
| Relaunch under the same name, domain, email and office, with the same directors | The prior entity was in a creditors' voluntary liquidation (public record); a clean break is standard practice, and Section 216 creates direct legal exposure | The brand carries a discoverable liquidation history into every prospect's due diligence |
| The March agent evaluation and training plan was not taken up | A like-for-like scoring of the team against a defined rubric, with a plan to close a ~3× conversion gap between agents working the same data | Re-scored in June against the same rubric, the same patterns persist — the gap is wider, not narrower |
| The strongest converter moved into management; a volume bonus introduced, then withdrawn | Kate converts roughly 1 in 10 conversations against a team average nearer 1 in 19 | Time and stability taken away from the one agent the data says to protect and learn from |
| The calling team repeatedly reshaped — openers and closers switched, agents rotated in and out | A small phone team converts best with consistency and coaching | Momentum lost to churn rather than compounded |
| Finite top-tier data worked by the highest-volume callers | The apex is ~4,000 non-renewable names; the volume callers convert it least efficiently | The best data spent for 1–2 deals where it could yield ~20 |
None of these needed more information to get right — the information was already there. That's why the fix isn't another report; it's a way of deciding and acting that stops good evidence from stalling.
| A clean-break rebrand | New entity, brand, domain and contact identity — the single move that removes the liquidation association and the Section 216 exposure at once. |
| An agent training programme | Turn the March evaluation into structured, repeatable coaching for whoever is on the phones — work that was already scoped once. |
| Match the data tier to the agent | Hold the finite Tier-1/Tier-2 data back from the highest-volume callers, who convert it least efficiently — it protects the apex and frees budget. Optionally trial a few hours of our own agents on the same data for a like-for-like read on return. |
| Resource on the evidence | Decisions on who stays on the phones should follow the performance data. Where there's another reason to keep a caller, that's legitimate — but make it a deliberate, budgeted choice, not a default the results are then expected to excuse. |
| Claim R&D tax relief | Set up correctly, a new entity can reclaim a meaningful share of qualifying expenditure — including salaries, not just marketing — by structuring roles around the development of the product rather than pure sales. At ~20% of qualifying spend, a business spending ~£1m a year can recover on the order of ~£250k a year. This is a specialism of ours: we can scope and run the claim, not just flag it — and we first raised it with you in 2024. |
As with the foundations above, most of this was first put to you in 2024. The consistent thread across this whole review is that the data, and the advice, have been sound and stable — the missing piece has been a decision-making setup that lets good evidence turn into action. That is the thing most worth fixing, because it's the thing that unlocks all the rest.