20 Customer Voices in 7 Days for £8,400: How a Multi-Strat Hedge Fund Got Its Q1 Read on a Listed BPO

There is a particular kind of stress that lives inside a Q1 earnings calendar at a multi-manager platform. Six tickers in scope. Roughly thirty trading days to refresh views. A sector under live debate — in this case, the AI-disintermediation question hanging over the listed BPO and digital-operations names. Bulls argue the vendors are embedding agentic AI into mission-critical workflows and trading FTE revenue for higher-margin non-FTE revenue. Bears argue the seat-count compression is running ahead of the AI revenue ramp. The 10-Q won't settle it. The sell-side won't settle it. The only people who actually know are the customers — the procurement leads, GBS heads and process owners who sign the renewal or quietly start an RFP eighteen months out.

A multi-strategy hedge fund running a fundamental long/short sleeve across TMT and Business Services had a position open on one of the listed pure-plays. The analyst needed a customer-side read before the print. Not "industry views." Customers.

They came to Woozle.

The Brief

Fifteen minutes. That was the briefing call.

The analyst described the thesis in plain terms: pressure-test the AI-disintermediation narrative on a NYSE-listed BPO/digital-operations large-cap, and use the same calls to set direction on three or four read-across tickers in the same basket — the Indian IT services majors, the analytics-heavy pure-plays, the consulting-led incumbents. The fund didn't want generic sector colour. They wanted the contract-side counterparties.

We locked the screener on the call. A respondent had to be a current or recent (within eighteen months) buyer-side decision-maker with direct visibility on the relationship with the target vendor. Seniority gates set at procurement category lead, GBS director, COO or functional process owner. Workflow coverage spread across finance and accounting, supply chain, customer experience, analytics and broader digital operations. Mix targeted as follows:

• Six to eight procurement and category leads for IT and business services — own the contract, see the price card, run the RFP.
• Four to six shared services heads and GBS directors — own the in-house side of the relationship, see headcount, scope, KPIs and any AI-led seat reduction directly.
• Four to six COOs and functional process owners — F&A controllers, supply chain directors, claims VPs — the people who decide whether the vendor is mission-critical or replaceable.
• One to three former employees of the target now sitting customer-side — the calibration layer that understands both sides of the table.
• One or two CIOs or heads of enterprise AI — specifically for the agentic-displacement question.

The discussion guide was built by a Woozle analyst against the thesis. Seven question areas: demand health and pipeline tone for FY26; pricing dynamics on the FTE book versus the non-FTE book; whether AI cannibalisation was real or theoretical inside the customer's own workflows; retention and switching, including who else was on the RFP shortlist (TCS, Infosys, Accenture, WNS, EXL, Cognizant, captive build, hyperscaler-led options); the split between net-new logos and share-of-wallet expansion; the competitive read-across into the rest of the analyst's basket; and the question of mission-critical stickiness — what could and couldn't be ripped out inside a contract cycle.

Fifteen minutes of the analyst's time. That was the input.

The Execution

Day zero to day one: discussion guide drafted, screener locked, recruiting brief out.

Day one to day five: custom recruiting. Not a database recycle. Fresh respondents profiled for fit against the screener before we put anything in the diary, with conflict and compliance checks run at the recruit stage. This matters more in this sector than most. The top of the BPO peer set has a concentrated customer base — top twenty logos drive a disproportionate share of revenue at the listed pure-plays — and the same handful of customer-side voices tend to recur across the shared databases of the major networks. We went looking for voices the rest of the buyside wasn't already calling.

Day two to day seven: interviews conducted, recorded, transcribed and structured against the questionnaire. Every call was moderated by a Woozle analyst. The fund's analyst did not sit a single interview.

Day seven: delivery. Twenty completed transcripts and recordings, indexed against the structured questionnaire, ready to consume ahead of the Q1 read.

The client did not write the discussion guide. The client did not vet a single profile. The client did not schedule, reschedule, or chase a transcript. The client did not moderate. The client did not synthesise. The 14 hours a month that a typical analyst loses to expert-network logistics — brief-writing, profile review, scheduling, sitting calls, chasing recordings, writing up — went to zero for the duration of the project.

What the Research Revealed

The kind of finding a twenty-call programme on this peer surfaces, drawn against the public commentary on the sector:

Renewal mood was healthier than the bear case suggested — but not uniformly. The bulk of customers had renewed in the last twelve months; a smaller cohort had run an RFP; one or two were mid-process. The renewals themselves split roughly sixty/thirty/ten between flat, expanded and descoped. The descoped cases were the ones that mattered — they sat almost entirely inside F&A workflows where the customer had stood up internal AI capability.

Pricing was rational on the FTE book, stable-to-up on non-FTE. Customers described low-single-digit rate uplifts pushed through on the FTE side over the 2025 cycle, with no meaningful pushback. On transaction- and outcome-priced contracts — the non-FTE book the peer group is steering towards (the target peer has disclosed 46% non-FTE today, 70% the medium-term target) — unit economics were stable or slightly improved. The peer CEO's "rational pricing environment" framing from the Citi conference last September stood up.

AI cannibalisation: the bear thesis was partially falsified, not confirmed. Customers were seeing seat reductions of 10–20% in pockets of F&A — but those same customers had signed agentic AI add-ons with the same vendor. The "non-FTE revenue replaces FTE revenue" mechanic is real. It is also lumpy. Some customers had paid; others were running pilots they hadn't yet scaled.

Pilot fatigue is the bear case in disguise. Several customers described three or more agentic pilots running concurrently with only one in production. A useful counterweight to the bull narrative — and not something that shows up in a 10-Q.

Mission-critical stickiness was confirmed. Close-the-books for F500 finance functions and supply-chain orchestration were described, repeatedly, as effectively un-RFP-able inside a contract cycle. The vendor's "last-mile execution" framing held.

The read-across. TCS, Infosys and Wipro showed up as the named alternative vendor in most accounts — but typically for adjacent IT services scope, not like-for-like. Accenture appeared where there was a consulting-led transformation programme on the front end. WNS and EXL came up as the direct overlap on analytics-heavy F&A and insurance workflows. Captive build and hyperscaler-led options were mentioned as a long-term threat, not a twelve-month one. That set the direction for the analyst's view on three other tickers in the same basket going into Q1 prints.

The Deliverable

Twenty transcripts. Twenty recordings. Each structured against the questionnaire so a reader could move horizontally across responses to a single question — renewal behaviour across all twenty interviews, or pricing posture, or AI deployment status — without re-reading every transcript end to end.

The analyst held the position into the print with conviction on the mix behind the headline number, a defensible view on the AI question, and a refreshed read on three read-across names in the basket. One brief. One delivery. No internal logistics drag.

What This Looks Like in Practice

The contrast with how this brief would have run at a legacy network is structural, not promotional.

A traditional expert network would have approached twenty customer-side completes as twenty separate per-call transactions. Sourcing windows on sector-specific buyer-side voices typically run five to ten business days per profile. The analyst writes the discussion guide. The analyst sits each call. The analyst takes the notes. The analyst synthesises. The deliverable is a transcript.

The pricing is well-documented. GLG charges clients $1,500–$2,000 per hour in 2025 according to an Expert Network Calls platform review published last November. Independent industry commentary puts sticker pricing at the major networks at $1,000–$1,200 per call, and notes that "if 40% of calls are useless, and each useful call requires 2–3 hours of analyst time beyond the interview itself, your cost per useful insight is closer to $2,000."

Run the math. Twenty calls at the legacy rate is £24,000–£32,000 at the low end and £30,000–£40,000 at the GLG-equivalent rate, before a single hour of analyst time is costed in. Add three to five hours of analyst time per call for vetting, prep, moderation and write-up, and you're looking at sixty to a hundred hours of analyst time absorbed across the programme — call it eighty hours, the equivalent of two working weeks of senior research capacity.

And the calendar. A twenty-complete programme at a legacy network on a customer-of-target screener — the recruiting-intensive end of the spectrum — typically lands three to four weeks after brief. After the print, in this case.

This isn't a discount. It's a different model. Per-call brokerage networks make money on volume of calls multiplied by margin per call; the unit economics depend on the client doing the moderation and synthesis work. As Marc Rubinstein noted in his October 2021 analysis of GLG's withdrawn S-1, GLG's gross margin is around 74% — "one of the highest take rates of any platform company, anywhere" — and that take rate is only possible because the labour-intensive parts of the workflow are pushed to the client. A fixed-fee, done-for-you engagement inverts the incentive. Faster, tighter and higher-hit-rate is better for both sides.

The savings on this engagement landed in three places. Roughly £20,000–£30,000 on the cash line. Around eighty hours of senior analyst time recaptured for the rest of the Q1 calendar. And — the one nobody costs but every analyst feels — the administrative and cognitive load of running the project simply did not exist. The analyst briefed once and read transcripts on day seven.