Klenty

The price list settles this cleanly and in the vendor's own structure. The entry tier is described as ideal for teams scaling personalised email outreach and its enumerated contents are entirely mechanical: deliverability controls, custom domain tracking, sending randomisation, volume control, analytics, interface connections, spreadsheet upload, split testing, an engagement feed, merge fields, template logic, and preview, edit and approve before sending.

Not one model dependent capability appears on it. Everything with artificial intelligence in its name starts a tier higher, and the deepest of it, automatic execution of next steps after a conversation, objection detection, call scorecards and battlecards, is reserved for the top tier. A complete and saleable outreach product therefore survives removal of the models entirely, which is the test. The company began in 2016 as an email sequencing tool and the agent layer is recent and additive rather than foundational.

Oversight appears where it counts, in the shipped feature list rather than in a policy paragraph. Preview and edit, and approve before sending, are named capabilities on the entry tier, so a review step before a message leaves is a product feature a buyer can point at rather than a claim. The positioning reinforces it: the vendor runs a page arguing that reps should move from taking orders to approving them, framing the agent layer as producing work a person signs off.

That is an unusually explicit statement of intended human placement. Held off the top band by the other half of the product. Cadence generation composes entire multichannel sequences from a plain language brief, and the top tier feature is described plainly as automatic execution of next steps after a conversation, which is action taken without a stated approval gate. Nothing published describes an audit trail of what the agents did, a guardrail that withholds rather than displays a bad output, a confidence threshold, or an escalation path by risk.

The volume of model dependent features is high and the disclosure behind them is nil. Cadence generation, account research, subject line and opener writing, voicemail and menu detection, outcome detection, conversation tracking, objection detection, call scorecards, battlecards and adaptive roleplays are all marketed by name, and no provider, model family, version, inference location or evaluation method is published for any of them.

The omission that matters most on this product is detection accuracy, because two separate classifiers decide what happens to a live call: one judges whether a human or a recording answered, and one labels the outcome that then drives the automated follow up. An error in the first drops a representative into a greeting or leaves a person listening to silence, and an error in the second sends the wrong next message to a real prospect. No accuracy rate, false positive rate or confidence indicator is published for either.

Among the better evidenced vendors in this index and it clears the bar on named attribution rather than volume. Three testimonials carry a full name, a job title and a company, and two of them carry a quantified outcome: a 93 percent jump in outbound revenue attributed to a named president, and a move from 60 to 200 converted meetings attributed to a named head of business development. A dedicated customer stories section and at least one full case study sit behind them.

Independent corroboration is genuine rather than self reported: a listing in a major software marketplace carries a 4.7 rating across 386 ratings, which is a verifiable third party record rather than a curated quote. The company states more than 5,000 teams across 45 countries. The vendor also publishes what it describes as original cold calling research, which is a rarer thing than a blog and is at least an attempt at primary evidence.

Held off the top band because no measurement basis, baseline, period or methodology accompanies any of the quantified claims, the logo band is unnamed, and one case is stated as nine months in a testimonial and eight months in the linked story.

Five regulated surfaces run through this product, being email, telephone, text message, business messaging and the professional network, and no statute governing any of them is named anywhere on the surface read. There is no published sending policy, acceptable use policy or compliance page, no position on consent for calling or texting, and no abandoned call rate or pacing rule despite a parallel dialer marketed on multiplying dialing volume and opening ten or more conversations an hour.

What holds this above the floor is real but indirect. The contract prohibits unsolicited bulk sending and other prohibited content, and the privacy policy states that the vendor will access an account to identify a breach of those terms, so the prohibition has an enforcement mechanism behind it rather than sitting decorative. European data protection law is named and addressed in the privacy estate, and a processing agreement incorporating standard contractual clauses is offered. The gap is that all of it lives on the data protection side and none of it reaches the outbound conduct the product exists to perform.

Real substance, undermined by maintenance. The policy states its legal bases, works the controller and processor split correctly rather than hiding behind it, and says plainly that the vendor has no direct relationship with the individuals whose data it holds as client data and that each client is responsible for giving those people notice, which is the honest position for a processor and is repeated where deletion rights are discussed.

A processing agreement incorporating standard contractual clauses is offered, a data protection officer address is published, the absence of do not track support is disclosed rather than concealed, under sixteen collection is addressed, and testimonial consent is obtained in writing before posting.

Credential handling is described with unusual candour: the vendor states that it may hold either an access token or a username and password for a mailbox, that this permits reading inbox contents and sending on the customer's behalf, and that copies of both sent and received messages are stored. The problem is the date.

The policy is stamped as last updated on 24 May 2018, the day before European data protection law took effect, and has not been re dated in eight years, yet it demonstrably has been amended since, because it carries a clause about generalised model training that postdates that stamp by years. A buyer cannot tell which parts are current.

Alongside that, no California or other state privacy section exists beyond an older marketing disclosure provision, the access right extends only to name, password, email and payment card, and no portability, restriction or objection right is articulated.

Two distinct acquisition routes are disclosed and neither is sourced. The privacy policy states that third parties are used to collect publicly available information relating to the customer's prospects, which is an admission that the vendor enriches beyond what the customer supplies, with no supplier named and no method described.

Separately the top tier includes four thousand data credits covering telephone numbers and email addresses, so there is a contact data product being sold, and no provenance statement, coverage claim, accuracy rate or refresh cadence accompanies it. A tracking script installed on the customer's own website collects prospect usage information there, which is a third route.

Held at the middle band rather than lower because the controller position is correct, the prospects supplied by the customer remain the customer's responsibility and the vendor says so, and there is an express statement that neither customer nor prospect personal information is rented or sold in identifiable form.

Mixed, and the exposure sits on the channels the vendor does not own. Professional network selling is a priced tier feature and covers connection requests, direct messages and paid messages in a single click, executed from the seller's own account, and business messaging automation adds a second platform whose commercial policy is strict and frequently enforced. No method, activity ceiling, pacing rule or terms of service position is published for either.

Mailbox custody adds a third exposure: the vendor states it may hold a username and password rather than a revocable token where the older mail protocols are used, which is on the wrong side of the line the major providers have been moving toward, though it is disclosed field by field rather than concealed.

Pulling the other way, and genuinely: the vendor operates its own dialing infrastructure rather than reselling, and distributes through official listings and dedicated integrations with five major systems of record. Held at the middle band because there is no account rotation, no multi account mechanism, no marketing that sells evasion and no scraping product.

The training question is answered in writing for the most sensitive input this product touches. The vendor stores copies of both the messages a customer sends and the messages prospects send back, and the privacy policy states explicitly that the major mailbox provider's interfaces are not used to develop, improve or train generalised models, alongside adherence to that provider's limited use requirements. That is a specific negative commitment rather than a reassurance.

The cross customer question is also answered and, unusually, it is answered with a boundary drawn rather than a vague permission: the vendor reserves access to general statistical information such as reply, open and bounce rates for benchmarking that helps all customers, and confines the described use to that numerical layer rather than to content. Off the top band because the commitment covers one provider's interfaces and not the rest of the corpus.

Mailboxes connected by the older protocol route with a stored username and password fall outside it, as do call recordings, transcripts and the coaching material derived from them, and no tenancy statement, retention period or model provider is published for any of those.

Several mechanisms here shape what the person contacted believes and none carries a stated position. The connect rate layer matches the number shown on the recipient's screen to their own area code and rotates numbers out of the pool automatically, so the displayed origin is both manufactured and changing. Voicemail drop leaves a message that sounds spoken in the moment. Detection means a person who picks up may hear a pause while a classifier decides.

Generated openers and subject lines are written per prospect and presented as the sender's own words. Message open, click and reply activity is recorded against the individual. Professional network actions run from the seller's account under their real name. No disclosure of artificial origin is described on any channel and the European marking obligation is not addressed.

This sits at the same band as the other two dialers graded here for the same manufactured location tension, and unlike one of them there is no branded calling counterweight showing the recipient a real company name.

Depth where it matters for this category. Five major systems of record each have a dedicated integration page rather than a logo on a grid, synchronisation is bi directional rather than one way activity logging, and the described behaviour is specific: standard and custom fields mapped both ways, records routed automatically into lists, sequences triggered on arrival, and sales stage automation writing back.

A listing in a major software marketplace, a general automation platform connector, spreadsheet import and export, triggers and actions, a plugin and a documented interface complete the surface. Off the top band on three counts: the interface documentation lives in a support centre article rather than a developer portal, no webhook surface or event model is described, and no agent protocol server was located, which several vendors graded here already ship.

The residency clause reserves more discretion than almost any other graded here. The policy states that information may be transferred to and stored and processed in the United States or other countries that the vendor chooses to use, which names no region, commits to nothing and leaves the set open ended by its own wording. No data centre, hosting provider, processing location or sub processor register is published.

The gap is widened by a fact visible elsewhere on the site but absent from the policy: the product organisation sits in India while the corporate base is in the United States, so a second jurisdiction is plainly involved in operations and is never disclosed as a processing location.

What holds this at the middle band is that a processing agreement incorporating standard contractual clauses exists and is offered on request, which is a real transfer mechanism even though it is gated behind an email and describes lawfulness rather than location.

Attestations are named and, critically, named with their type. The pricing answers state a service organisation control report at type two specifically, rather than the bare assertion this index has repeatedly marked down, and the footer displays that alongside health information and European data protection marks. A responsible disclosure page is published as a standing footer item, which gives security researchers a route and is a commitment most vendors at this size do not make.

Off the top band on the access question rather than the existence question. There is no trust centre, no route to the report itself, no audit period, no auditor named, no penetration testing statement and no enumerated control set, and the privacy policy's own security section is thin enough to sit oddly beside the badges, describing only password protection and transport encryption while conceding it cannot guarantee security. Recorded as observed: the marks appear as images in the footer and an image is a claim rather than evidence, so the underlying reports were not seen.

Strong and specific. Three tiers carry real figures at 50, 70 and 99 dollars, a billing period toggle is offered with the discount stated, and the unit is made explicit where it changes: the entry tier is priced per month against published volume allowances of fifteen thousand contacts and seventy five thousand emails, while the two higher tiers are per user.

Consumption is quantified rather than implied, with calling minutes included per tier at two hundred and one thousand, and four thousand contact data credits named at the top. Every tier's contents are enumerated in full. The piece that most often breaks this axis is present: the load bearing dialer add on carries a published price of 45 dollars per user per month rather than being reserved for a quote.

Commercial terms are stated plainly, with cancellation at any time, no cancellation fee, no charge for the following period, and trial terms specified down to the message cap. Held off the top band because only the entry tier offers self service, while both tiers a real team would buy route to a demand for a demo, which is the pattern seen elsewhere here where the tier the buyer actually needs is the one gated. A separate dialer price list also exists and was not read, so the full bill is not visible from one page.

The commercial exit is clean and the data exit is not described. Cancellation is available at any time with no fee and no charge for the following period, stated in the pricing answers, and the service is described as pay as you go, so a buyer is not trapped by term. Spreadsheet import and export and automatic export are named features, and bi directional synchronisation means engagement history accumulates in the buyer's own system of record rather than only in the platform.

What is missing is the whole post termination picture. Account deletion is available only by emailing a request, no timeline or confirmation artefact is offered, the policy states that some information may remain in private records afterwards, and aggregated data derived from customer information may continue to be used after deletion. The stored copies of sent and received mail are nowhere addressed in the exit context, which matters because the vendor holds them by its own account. The published access right reaches only four account fields and does not extend to the operational corpus.

Sending discipline appears as named product features rather than as advice, and it appears on the cheapest tier, which is what earns the band. The entry plan lists a deliverability capability, custom domain tracking, randomised sending intervals and email volume control, so pacing and volume governance are shipped rather than sold up, and a dedicated deliverability feature page sits behind them. Split testing and an engagement feed give a buyer the feedback loop to act on.

On the voice side, automatic number rotation and number reputation management appear a tier higher. Off the top band on two specific gaps and one reservation. No sender authentication guidance, blocklist monitoring, complaint or bounce threshold, warmup process or inbox placement testing is named.

And the voice mechanism is described as rotating numbers out before they damage the connect rate, which is the same pattern flagged on the comparable dialer here: moving away from a degraded identifier treats the signal as an obstacle rather than as information about the calling behaviour that produced it.

The tier design is a genuine segment statement rather than a feature ladder. The entry plan is priced on volume rather than seats and is scoped to teams running email only, the middle plan is aimed at teams whose problem is keeping the system of record clean, and the top plan is described as being for mature development teams consolidating an outbound stack, which is a maturity dimension rather than a size one and is the more useful cut for this category.

Distinct solution pages address outbound, inbound and account based motions. The unusual element, and one almost nobody else publishes, is geographic segmentation as a product: a region based dialer line with a global variant and a dedicated variant for calling into India, which tells a buyer selling into that market something concrete about fit. Held off the top band by the framing sentence, which claims teams of ten through teams of ten thousand, and by the absence of any stated ceiling, headcount band or disqualifying condition anywhere.