A civil investigative demand arrives. Or a complaint is filed alleging a Section 1 conspiracy, describing parallel price movements among four competitors over eighteen months. Nothing in the pleading quotes an agreement, because there is usually no agreement to quote. What it offers instead are plus factors, and the most common is the opportunity to conspire.

Opportunity gets built from structured records. Call detail records, calendar exports, travel and expense entries, conference attendance rosters. Somewhere in the exhibit is a list: fourteen calls between executives at the defendant companies during the relevant period.

Fourteen calls is not the problem. The problem is that the fourteen came out of a production holding several hundred thousand rows, and nobody on your side has mapped the rest of it.

The risk is the part you have not looked at

Every antitrust defense team knows the contents of its own document production. Far fewer have mapped the structured records inside it as a network. That gap is where cases get lost, because opposing counsel is running the analysis whether you are or not.

The parties that matter are frequently not on anyone’s list. A distributor both companies were calling. An industry consultant retained by three of the four defendants. A former employee who moved between the two and stayed in contact with both. None of these appear by reading rows, because a spreadsheet shows you the calls you searched for, not the entities standing between them.

Some of what surfaces will help you. Contact that ran through a legitimate intermediary is the cleanest answer to a coordination allegation, because it explains the pattern without requiring anyone to have agreed to anything. Some of what surfaces will not help you, and that is precisely why you want it now, during your own analysis, rather than on a screen during a deposition.

You do not get to choose which by declining to look.

Where document platforms stop

Relativity, DISCO, and the rest of the review stack do their job well. They ingest custodial email, chats, memos, and attachments, and they let a team search, cull, tag, and produce at scale. If the allegation turns on something somebody wrote, review will find it or establish that it does not exist.

Antitrust rarely turns on what somebody wrote. It turns on the shape of contact between firms that were supposed to be competing. A hit report cannot describe that shape, nor can a call-record spreadsheet, which is why the version attached to the complaint is a narrative summary with a handful of selected rows to support it. Presented in isolation, fourteen calls look like coordination. Seen within the full network, they usually look more specific and far more arguable.

Start with their list. Then go past it.

The workflow is the same one an investigator runs, pointed at your own production.

Take the custodian phone numbers named in the exhibit and search them as a set. Load the full call detail record production behind them. Then ask the tool to return every path between those numbers, including the ones that run through parties you never searched for.

SentVi is built for this layer of the case. It stores data in Microsoft SQL Server and handles millions of entity and relationship records, so the whole production goes in rather than a sample. Get Links Between Selected surfaces the communication paths connecting any set of entities, including multi-hop paths through intermediaries. Social Network Analysis metrics such as centrality and betweenness rank which of those intermediaries actually sit at the center of the traffic and which are incidental. A data refinement filter layers subscriber and owner information onto the numbers, and entity replace swaps the numbers themselves for the people and companies behind them, so the finished chart names parties rather than digits.

Timeline analysis places a time-range slider across the whole network, so you can run the map for the exact conspiracy window the complaint alleges, then run it again for the 18 months before, which is frequently where a theory comes apart. Geospatial views place the same records on integrated online and offline maps, with Google Earth export, which matters when the alleged plus factor is co-location at an industry event.

What you do with the map

Findings from this exercise tend to land in three buckets, and all three are worth having.

The explanation. Contact between competitors was routed through a shared distributor, vendor, or trade association. That is the strongest possible answer to an opportunity-to-conspire argument, because it accounts for the communication pattern without any agreement.

The context. Contact volume between the parties held steady before, during, and after the alleged conspiracy period, or did not vary around the pricing events the complaint anchors to. A time-filtered network makes that visible in a way a table does not.

The surprise. A connection nobody on the defense team knew about. Finding it during your own analysis is a bad afternoon. Finding it during cross-examination is a different kind of problem entirely.

Getting the work out of the analyst’s hands

A licensed user can export any chart to an SVX file, which is a self-contained mini database. Anyone can open it with SentViReader, a free, view-only application, and drill into the metadata and notes behind every entity and relationship, without a license or a procurement cycle.

For a joint defense group, that is the difference between one shared analysis and six separate reconstructions of the same records. For mediation or a class certification hearing, it is a demonstrative that opposing counsel can open and examine, which is a stronger position than a static image they can dismiss as attorney-drawn.

Confidentiality, since the data is the sensitive part

Antitrust productions include competitive pricing, customer lists, and executive communications, sometimes shared among multiple parties under a common interest agreement. SentVi installs on your own hardware and can run completely disconnected from the internet. FMS never sees or touches the data. There is no vendor-side copy to disclose and no third-party processor to add to the protective order.

Where to start

Take the call detail records already produced in the matter. Load them, search the custodians named in the exhibit, and return every path between them. The parties you have never heard of are the reason to run it.

Book a Demo Today

author avatar
Jason Christy
Jason Christy is Chief Marketing Officer at SentVi. He spent 25 years in enterprise sales and go-to-market leadership, and writes about link analysis, investigative technology, and how legal teams work with structured data.