Backstory

The matching layer is the product and it cannot be done by rules. Turning raw activity into a usable record means deciding that a given email thread belongs to a particular opportunity at a particular account with a particular stakeholder, across systems where the same person appears under different addresses and the same company under different names.

That is a probabilistic matching problem, the vendor holds a large patent portfolio around it per third party reporting, and its own material argues the capability cannot be replicated quickly. Strip the model and there is no product: what remains is unmatched activity, which is what the customer already had. The generative assistant layered on top is newer and separable, and the vendor's own framing distinguishes the matched data foundation from the answers built on it.

Held below the top band because the company dates to 2016 and the matching engine predates the current generative positioning, so the marketing has shifted further than the architecture. Ask what the matching accuracy is and how it degrades on accounts with complex corporate hierarchies.

The system observes and answers rather than acting, which limits what this axis can find, and the boundary is still unstated. Capture is continuous and automatic by design, running without rep involvement, but it writes activity records rather than taking actions in the world, and the assistant responds to questions a person asks. Nothing contacts a customer, sends on anyone's behalf or changes a deal.

The oversight questions that do apply are unaddressed: no published statement describes what an administrator can constrain, whether capture can be scoped or paused per person, what happens when the assistant is wrong about a deal a leader then acts on, or what controls exist over answers surfaced into third party assistants through the protocol server. The last is the live one, because once answers flow into an external assistant the vendor no longer controls their presentation. Ask what an administrator can disable per person or per team, and what governs answers delivered through the protocol server into external assistants.

Traceability is real and the models behind it are undescribed. The vendor's strongest transparency claim is provenance rather than model disclosure: every answer is stated to trace back to the meeting, stakeholder and figure behind it, qualification scores carry a line by line link to a real meeting or email, and pipeline status traces to the underlying activity.

For a product whose failure mode is a confident wrong answer about a deal, sourcing each claim to a record is the mitigation that matters most, and it is a genuine design commitment rather than a page. Against that, no model provider, family or version is named anywhere, no model card exists, and no accuracy figure is published for either the matching engine or the generated answers, despite matching accuracy being the foundation everything else rests on.

The vendor argues its answers are more defensible than a general model's, without publishing the measurement that would show it. Ask which models generate the answers, and for measured matching and answer accuracy.

Quantified outcomes in volume, undercut by a contradiction on the vendor's own homepage. Published figures include a seventy five to eighty five percent reduction in manual data entry, more than one thousand manager hours projected to be reclaimed, more than one thousand seller hours saved annually, and two to three times improvement in forecast and pipeline visibility, alongside named case studies at large technology and industrial customers and a third party report of twenty to thirty percent forecast accuracy improvement.

None states a population or period, and one figure is explicitly a projection rather than a measurement. The contradiction is the sharper finding: the same homepage states trusted by more than one hundred thousand sellers near the top and trusted by more than five hundred thousand go to market teams further down.

Those differ fivefold and the second counts teams rather than individuals, so they cannot both be right and the larger one is implausible against a review base of roughly six hundred and thirty. Ask which reach figure is correct, and for forecast accuracy improvement measured across a stated customer population and period.

The platform sends nothing, and the compliance question it raises is about capture rather than outreach. There is no sequencer, no dialer and no campaign delivery, so consent, suppression and unsubscribe obligations sit entirely with whatever systems the customer uses to contact people.

What this product does instead is ingest the correspondence those systems produce, which relocates the question rather than removing it: the emails and meetings captured contain counterparties at customer and prospect organisations who never agreed to have their interactions analysed, scored for engagement and surfaced to a sales leader. Nothing published addresses what those people are told or whether any notice reaches them.

The vendor does publish a do not sell route and a rights management page covering its own site. Ask what notice, if any, reaches the external participants whose emails and meetings are captured and scored.

The published surface names the right documents and none of their substance was reachable. A privacy policy, terms, a rights management route and a dedicated trust and security page are all published and linked from the footer, and a toll free number is given.

What could not be established on the routes taken is everything a privacy review turns on for this specific product: no transfer framework certification was located, no subprocessor list, no processing agreement, no stated retention period and no data protection officer. That matters more here than for most vendors because of what is captured.

The corpus is the complete correspondence and meeting history of a customer's sellers, which necessarily includes the personal data of external counterparties who are not the vendor's customers and have no relationship with it, captured continuously and retained to build a multi year record. Revenue Grid, the closest comparable capture product in this index, publishes a certified privacy management system, transatlantic framework participation and a named European representative. Ask for the processing agreement, subprocessor list, retention period and transfer mechanism, and specifically how counterparty personal data is handled.

No external corpus is licensed and the provenance question is entirely internal. Everything the platform reasons over comes from the customer's own systems, so there is no purchased database, cooperative or third party feed to trace, and the vendor's positioning rests explicitly on this being first party data rather than bought signal. What is unstated is what happens to that data in aggregate.

The vendor and third party reporting both describe a proprietary activity dataset accumulated over nine years as a durable advantage, which invites the obvious question of whether patterns derived from one customer's deals inform the models or benchmarks serving another, and nothing published answers it either way. For a product whose moat is described as an accumulated cross customer dataset, that is the provenance disclosure a buyer needs. Ask whether activity data from one customer contributes to models, benchmarks or matching rules used for others, and in what form.

Capture runs on sanctioned interfaces into the systems it reads, and the newest surface is built on an open standard rather than a private arrangement. Named connections cover three customer record platforms, two mail systems, two meeting platforms and a messaging platform, all through their own published integration routes, and the vendor states it connects to several customer record systems simultaneously. Nothing resembling credential storage or unsanctioned extraction appears.

The distinguishing choice is the protocol server, which publishes answers into third party assistants through an open interoperability standard, so the vendor works alongside those tools rather than seeking exclusive placement, and one named customer is described building its own assistant over several sources including this one.

Held below the top band because no stated conformance position against any specific platform's terms was located, and because deep continuous mail and calendar ingestion depends on grants those platforms control, with nothing published on what happens if that access narrows. Ask what permission scopes the mail and calendar connectors request, and whether any named platform conformance commitments exist.

The most sensitive corpus in this category and no stewardship position reachable on the public surface. What accumulates here is every email, meeting and call a customer's sellers exchange with the outside world, matched to named individuals and retained across years, and the vendor backfills two years of history at go live so the corpus is deep from day one.

Nothing published states whether that material trains or tunes any model, whether it crosses a tenant boundary, which providers process it when the assistant generates an answer, what retention applies, or what happens to a departed employee's captured correspondence. A trust and security page exists and its substance was not reachable on the routes taken this pass, so this grade reflects what a buyer can establish before contacting sales rather than a finding that no position exists.

Compare Revenue Grid, which answers the cross customer question architecturally through tenant isolation up to physical separation. Ask whether captured activity trains or tunes any model, whether it crosses tenant boundaries, and what retention applies to correspondence after an employee leaves.

Nothing generated by this platform reaches a buyer, which narrows the axis to the people whose data it holds. No message is sent, no content is authored for external delivery and no agent speaks to a customer, so the authorship and impersonation questions this axis normally asks do not arise. What does arise is the position of the external counterparty.

A person corresponding with a seller has their emails and meetings ingested, their engagement scored, their responsiveness tracked and their status as a stakeholder assessed and reported internally, and nothing published indicates they are told any of it happens. That is a disclosure gap about people rather than about content, and it is inherent to the product rather than incidental.

A rights management route exists on the vendor site but a counterparty would have to already know the vendor's name to use it. Ask what notice reaches external participants, and how a counterparty could discover and exercise rights over activity data held about them.

Genuine breadth on the ingestion side and an unusually open posture on the delivery side. Capture spans three customer record platforms, two mail systems, two meeting platforms and a messaging platform, with simultaneous multi system support presented as a specific capability for organisations running different systems across regions or subsidiaries, which is a real architectural claim rather than a connector count.

Delivery inverts the usual model: rather than requiring work inside its own interface, the platform publishes answers through a protocol server into external assistants and into the customer record system, and the vendor states explicitly that a customer does not have to work in its application to get value. A partners and integrations page and a help centre are published.

Held below the top band because no developer documentation or interface reference was reached on the routes taken this pass, so the programmable surface beyond the protocol server is unverified, and no integration inventory was counted. Ask for the interface documentation and the current verified connector list.

A dedicated page addresses the security buyer and the residency answer was not reachable. The vendor publishes a role page aimed at information technology leaders promising deployment their security team will approve, and a trust and security page, which indicates the questions are anticipated.

What could not be established on the routes taken is any of the answers: no hosting provider or region named, no residency election described, no tenancy model stated, and no recovery time or recovery point objective published. Deployment speed is quantified at two to four weeks, which is an implementation figure rather than an architecture one.

The gap is consequential given the corpus, since a European customer's sellers correspond with European counterparties and the resulting record is personal data at scale, and the closest comparable capture vendor in this index answers this at the maximum by offering on premises deployment and physically separate tenancy. Ask which regions data can be pinned to, what the tenancy model is, and whether any private or self hosted option exists.

A trust surface exists, is signposted twice, and no certification could be confirmed from it on this pass. The vendor links a trust and security page from the main platform navigation and the footer, describing data a buyer can trust backed by the security and compliance they need, and separately publishes a role page for information technology leaders.

Neither the page contents nor any certification, attestation, audit period or auditor was reachable on the routes taken, and no certificate, report or questionnaire response was located. This grade therefore records what a buyer can verify before a sales conversation rather than asserting that certifications are absent, and given the enterprise customer base named on the site it is likely the underlying documentation exists behind the page.

The distinction matters and is recorded plainly: this is a retrieval limit on this pass, not a finding that the vendor holds nothing. Ask which certifications and attestations are held, with audit periods and auditors, and whether the reports are released under a confidentiality agreement.

The most thorough explanation of a pricing model in this index, containing no price. The vendor publishes a dedicated page setting out a hybrid structure in real detail: a platform fee charged only for people whose customer activity feeds the system, consumption charged for answers actually delivered, and no charge at all for viewers, so managers, finance, operations and deal desk use the product without a licence. It states there are no feature tiers and everyone gets the full product.

It defines what counts as a billable answer and states that reading a dashboard is free. It explains that each connected person carries a pre packaged consumption allotment pooling across the organisation, and describes a forward adjustment mechanism where overuse is neither cut off nor back billed, with the term resized at renewal instead. Seven questions answer the mechanics. Every structural question a buyer could ask is answered except what any of it costs.

No platform fee per connected person, no price per answer, no package size, no minimum and no term appears, and the page routes to a quote form. Third party reporting clusters around fifty dollars per user monthly with a median annual contract near twenty three thousand dollars and seat minimums, though those figures describe the prior per seat model rather than the current one. Ask for the platform fee per connected person, the price per answer, the pre packaged allotment size, and the minimum term.

Terms are published and the exit terms within them were not reachable. A terms and conditions document is linked from the footer, so the contractual position is obtainable rather than hypothetical, but nothing on the public surface states export scope or format, retention or deletion timelines, or renewal and termination notice periods. The exposure is unusual in shape and worth naming precisely.

The asset here is not configuration or content but a matched activity history, built from the customer's own emails and meetings and then enriched with the vendor's matching, scoring and stakeholder mapping. The underlying raw material still sits in the customer's mail and calendar systems, so a departing customer does not lose the source data.

What they would lose is the matching and the derived structure, which is the part that took two to four weeks and two years of backfill to build and is the reason the product was bought. Ask whether matched activity records and derived engagement scoring export in usable form, and what the deletion timeline is for captured correspondence.

The platform operates no sending infrastructure of any kind. There is no sequencer, no dialer, no campaign delivery and no mail relay, so shared reputation, warmup, rotation, bounce handling and complaint thresholds have nothing to attach to here. The product reads mail rather than sending it, and the vendor is explicit that execution stays in whatever tools the customer already uses.

The only adjacent surface is the vendor's own newsletter, which carries a subscription route and a linked privacy policy. Nothing further is knowable because nothing further exists. Ask, if any notification or digest is ever delivered by mail from vendor infrastructure, what domain carries it.

The target buyer is stated with unusual bluntness, including who the product is not for. The vendor says directly that it is built for revenue leaders who own the number and answer to the board, and that it is not a rep productivity tool, which is a sharper exclusion than most vendors are willing to publish.

Five role pages address executives, revenue operations, sales leadership, information technology and marketing, so the security and operations buyers who gate an enterprise purchase are addressed by name rather than left out. The customer evidence supports the enterprise claim strongly, with named references across large technology, security, semiconductor, industrial and telecommunications organisations, and a review base of roughly six hundred and thirty.

Independent analysis adds the floor the vendor does not state: the platform assumes a customer record system deployment with administrative support, and teams without one are a poor fit. Held below the top band because no industry pages, no localisation and no stated smaller boundary were located, and reach figures on the site contradict each other. Ask what the smallest viable deployment is in connected sellers and administrative support.