How Buy-Side Firms Handle Expert-Network Pre-Call Materials: 7 Structural Models

Pre-call materials are the least-discussed artifact in the expert-network workflow. Expert-supplied resumes, redacted deal decks, sample analyses, and industry references arrive by email attachment or platform upload before a call ever starts, which places them outside the recorded-call boundary that compliance teams normally rely on. Buy-side firms have converged on seven structural models for handling these documents, and each carries different implications for MNPI exposure, retention obligations under FINRA Rule 4511, and downstream ingestion into research and AI systems.

Why Pre-Call Materials Are a Distinct Compliance Problem

A recorded expert call is a controlled surface. The moderator is trained, the platform logs the engagement, and the network's compliance team can review the transcript against the expert's disclosed affiliations. Pre-call materials break that model in three ways. They arrive before the moderator is present, they often move through email rather than the network's audit-logged environment, and they can contain artifacts, redacted decks, org charts, sample outputs, that the expert generated inside a current or former employer's information environment.

The SEC's 2011 enforcement action against Primary Global Research and the surrounding cases established that written materials passed around expert engagements are treated as evidence in the same way testimony is. That precedent, combined with the 2010 SEC rule text on large trader reporting and related recordkeeping and FINRA's retention framework, means a PDF attached to an expert introduction email is a regulated record from the moment it lands in an analyst's inbox. Compliance officers routinely cite pre-materials as a top-three MNPI risk vector precisely because analysts often do not think of them as records at all.

The seven models below are the structural responses buy-side firms have built.

1. No Pre-Materials Permitted

The simplest and strictest model. Some hedge funds ban all pre-call attachments from experts outright, forcing every piece of context into the recorded call itself. The reasoning is defensive: if no document leaves the expert's environment before the call, there is no artifact to scan, retain, or later produce in response to a subpoena. MNPI exposure is compressed into the moderated call window, where the network's compliance layer is designed to operate.

The cost is analyst preparation time. Without an expert bio or sanitized background document, the analyst enters the call cold, and the first ten minutes of a sixty-minute engagement are spent on context that a two-page brief would have carried. Firms that adopt this model tend to be smaller, single-strategy shops where the compliance function is centralized and the number of expert calls per week is manageable.

2. Network-Mediated Pre-Materials

The default model at the largest networks. Guidepoint publishes its compliance framework publicly, and GLG and AlphaSights operate similar structures: the network vets the expert's biographical material and any sanitized backgrounder, then delivers the artifact through its own platform, logging it in the engagement record alongside the transcript.

The artifact never leaves the network's audit boundary in the technical sense, though buy-side analysts can typically download a copy for local reference. This model shifts the first-line compliance review to the network, which is the party best positioned to know the expert's employment history and any pre-clearance conditions attached to the engagement. It is the model most large buy-side firms treat as the baseline before layering their own controls on top.

3. Analyst-Requested Targeted Artifacts

A more granular model in which the buy-side analyst requests a specific document, an anonymized org chart, a publicly presented conference deck, a sample of the expert's published work, and routes each request through compliance sign-off. Nothing arrives unsolicited. Every artifact has a documented request, a compliance approval, and a retention tag.

This model is common at fundamental long-only shops and mid-sized hedge funds where the compliance team knows the analysts personally and can turn requests around in hours rather than days. It produces a cleaner audit trail than model 2 because every artifact has an explicit business justification attached, but it slows down the research cycle and depends on a compliance function willing to operate at analyst speed.

4. Standardized Expert-Questionnaire Pre-Fill

Rather than accept freeform attachments, the network sends the expert a structured questionnaire before the call. The expert fills in fields, current and prior employers, scope of responsibility, product lines covered, geographic experience, and returns a machine-readable prep document. The buy-side analyst receives a consistent format for every call, and the compliance team can scan the fields against pre-defined MNPI flags.

The structural advantage is that the artifact is bounded by design. An expert cannot accidentally attach a slide from a current employer's internal deck when the response surface is a text field with a character limit. The trade-off is depth: a questionnaire captures what the network knows to ask about, not what the expert knows to volunteer. Firms using this model typically pair it with model 3 for calls that require specific documents.

5. Read-Only Viewer With No Download

A newer technical model in which pre-materials are delivered through a browser-based viewer that prevents local storage. Third Bridge, Dialectica, and others have moved in this direction, keeping the artifact inside the network's audit boundary for the full lifecycle of the engagement. The analyst can read, annotate inside the viewer, and reference during the call, but cannot download the file to a local drive or forward it as an email attachment.

This model addresses the retention problem directly. Under FINRA Rule 4511, the firm is responsible for retaining external written communications received by its personnel; if the artifact never leaves the network's environment, the network carries the retention obligation and the buy-side firm's exposure narrows to the annotations and notes the analyst produced. The friction is real, analysts accustomed to downloading a PDF and marking it up in their own tools have to change habits, and it does not eliminate the need for local note retention.

6. Post-Call-Only Materials

Some funds invert the sequence. Nothing arrives before the call. The moderator and analyst conduct the engagement with only the expert bio in hand, and any supporting documents, a follow-up chart, a public reference, a sample output, arrive after the transcript is complete.

The compliance logic is that materials arriving after the call carry the context of what was actually discussed. If the analyst asked the expert about a specific public benchmark and the expert follows up with a public report referencing that benchmark, the artifact's business purpose is documented in the transcript. MNPI ambiguity is reduced because the artifact is anchored to an on-the-record exchange. This model is most common at firms where the research process is call-first and documents serve as reference rather than preparation.

7. Segregated AI-Ingestion Pipeline

The most technically involved model, and the one that has emerged specifically in response to the buy-side's growing use of internal AI systems on research corpora. The largest quantitative and multi-manager platforms route pre-materials through a separate compliance-scanned bucket before any of the content touches the research corpus that analysts, portfolio managers, or internal models can query.

The structural premise is that pre-materials carry a different risk profile than transcripts, which have already passed through the network's moderation layer. Treating both as equivalent inputs into a research index conflates two different compliance regimes. The segregated pipeline runs MNPI classifiers, entity-recognition scans against restricted lists, and retention tagging before releasing artifacts, and in some cases holds pre-materials in a quarantine tier that never becomes queryable by generative systems at all. The operational cost is significant, a second ingestion path, a second compliance queue, a second retention policy, and it is a model available only to firms with dedicated engineering capacity for their research infrastructure.

How Firms Choose Between the Seven Models

Strategy type is the strongest predictor of model choice. Single-strategy fundamental shops tend toward models 1, 2, and 3, because their compliance function can operate at analyst speed and the volume of expert calls is bounded. Multi-manager platforms with dozens of pods running hundreds of calls per week gravitate toward models 4, 5, and 7, because analyst-by-analyst discretion does not scale and centralized technical controls do. Model 6 appears across firm types and reflects a specific research philosophy rather than a compliance posture.

A firm rarely uses only one model. The common pattern is a default, typically model 2 as the network-supplied baseline, with model 3 layered on for calls that require specific documents, model 5 or 7 applied at the platform level for how those documents are stored and queried, and model 1 reserved for calls flagged as high-sensitivity. Compliance officers describe the choice as a portfolio of controls rather than a single policy.