Yardstick Research tear-sheet / AI sales cohort

Methodology · how we score · rubric weights in plain sight · vendors received this sheet seven days before publication and could flag factual errors, never rankings

Lavender

Founded: 2020 (CITED — Crunchbase, TechCrunch; pivoted from a prior product called Sorter during COVID) Headquarters: New York, NY (CITED — Crunchbase; Atlanta origin per Hypepotamus 2022) Funding: ~$14.2M total across two rounds; Series A of ~$11–13.2M led by Norwest Venture Partners, Feb 2023; no publicly reported raise since (CITED — TechCrunch Feb 2023, Crunchbase, no contradicting source 2024–2026) Headcount: 79 (Latka 2024) vs. 16 (PitchBook) vs. 1–10 (Tracxn 2024). Latka figure most plausible given product surface area (CITED + ESTIMATED) Leadership: Will Allred (Co-founder & COO, public face) and William Ballance (Co-founder & CEO). One secondary source (sendtrumpet GTM Insider podcast write-up) introduces Allred as "Co-Founder and CEO" — likely a host-side title error rather than a real role change; flagged for vendor right-of-reply. Archetype: AI augmentation layer (specialized — email coaching). Not a sender, not a data provider, not multi-channel.

ICP (vendor-stated): Individual SDRs and AEs writing their own outbound or follow-up email; teams seeking real-time AI coaching on email quality; managers seeking team-level email-quality analytics without having to read individual rep emails.

Critical category-fit caveat for readers. Lavender's 0/4 scores on Deliverability infrastructure and Data accuracy are absent-by-design category caveats, not quality signals. Lavender is a coaching layer that scores prose; it does not own send schedules, sender reputation, domain warming, or a contact database. Buyers using this tear-sheet to compare against senders (Outreach, Reply, Lemlist) or data providers (Apollo, Cognism, ZoomInfo) should match category-to-gap rather than optimize blindly on total score.


Headline numbers

Observation Value Evidence
Total score (0–100) 60.0 (with category caveats on deliverability + data accuracy) Weighted sum of dimension scores per Yardstick v1 rubric
Cohort rank 3 of 12 Behind Apollo (67.5) and Lemlist (63.75); ahead of Reply.io (57.5)
Personalization grade 4/4 — top of cohort THIRD-PARTY (G2, MarketBetter Jan 2026, Reply.io 2026 review, Originality.ai Aug 2025) + THIRD-PARTY-VIDEO-OBSERVED (Lavender 3.0 Personalization Assistant in Videos #1, #2, #8 of demo dossier)
Deliverability infrastructure score 0/4 — category caveat CITED. Lavender is upstream of deliverability; absent-by-design
Cost-per-seat efficiency 3/4 CITED ($0/$29/$49/$69/seat published; one of the cleaner pricing pages in the cohort, no usage credits, no contract minimums)
Reply rate v2 — held-out test data forthcoming
Cost per booked meeting v2 — held-out test data forthcoming

Dimension scores (0–4)

Dimension Weight Score Weighted Evidence
Personalization quality 25% 4/4 25.0 THIRD-PARTY-VIDEO-OBSERVED. "Lavender's entire engineering surface is dedicated to scoring and improving the prose of one email … shows up consistently in third-party reviews as the strongest positive signal." Personalization Assistant pulls LinkedIn / News / Events / Job History / podcast summaries (corroborated in Videos #1, #2, #8 and Originality.ai Aug 2025 review). Lavender 3.0 (Oct 2024) added keyword search across News / Events / Job History, multi-email switching, and podcast-episode summary cards.
Deliverability infrastructure 20% 0/4 0.0 CATEGORY CAVEAT. Cited from lavender.md §2, §7: "Not a sequencer. It does not own send schedules, throttling, or domain warming … No deliverability signal. Lavender scores prose; it does not look at sender reputation, spam-filter risk, or domain warming health." 0 because absent-by-design — flagged as category-fit caveat, not a quality signal. A rep can write a 95-scored Lavender email that lands in spam.
CRM integration depth 15% 1/4 3.75 THIRD-PARTY. Salesforce, HubSpot, Pipedrive named explicitly in multiple reviews. Two-way data sync claimed for prospect context pulled into the sidebar (lavender.md §4). But Lavender is a coaching extension, not a CRM-system-of-record integrator — there is nothing material to sync because Lavender does not own opportunity, lead, or account state.
Cost-per-seat efficiency 15% 3/4 11.25 CITED. "Per-seat per-month is the only model. There are no usage credits, no per-send fees, no contract minimums published. This is one of the cleaner pricing pages in the cohort": $0 / $29 / $49 / $69 / Enterprise (~$89+) per seat. ~20% annual discount across paid tiers. Loses one point for the unusually low free-tier ceiling (5 emails / month — barely a trial).
Setup time 10% 4/4 10.0 THIRD-PARTY. "Cleanest 'low setup time' story in the cohort. No sequence to build, no domain to warm, no integration to configure beyond installing the extension and granting OAuth. Setup-to-first-coached-email is plausibly 5 minutes, the lowest in the test set." Will be MEASURED-stopwatched on May 5–18.
UI / UX 10% 4/4 10.0 THIRD-PARTY-VIDEO-OBSERVED. "Lavender 3.0 (Oct 2024) is a substantive UI refresh — left-rail layout … information density: low-to-medium … much less per-pixel information. This is a feature, not a bug" (demos/lavender.md). Visible in Video #2 (Jan 2025 Coach Walkthrough), Video #6 (Jun 2025 third-party live-rewrite), Video #8 (homepage 2:59 sizzle). Pre-Oct-2024 demo footage shows an obsolete right-rail UI; buyers researching Lavender on legacy YouTube content from before Oct 15 2024 are looking at a UI that no longer ships.
Data accuracy 5% 0/4 0.0 CATEGORY CAVEAT. Cited from lavender.md §2: "Not a data provider. It enriches by pulling from third-party sources at compose time; it does not maintain a B2B contact database." 0 because absent-by-design. Buyers needing a data layer must pair Lavender with Apollo, Cognism, ZoomInfo, or Clay.
Total 100% 60.0 with category caveats on Deliverability (0/4) and Data accuracy (0/4)
Reply rate (v2) v2 forthcoming
Cost per booked meeting (v2) v2 forthcoming. Methodology note: because Lavender does not send and does not own meeting attribution, the $/meeting metric becomes an apportionment exercise rather than a clean read. We will report it but flag the methodology when v2 ships.

Pricing tiers (CITED — triangulated across MarketBetter, SyncGTM, Reply.io 2026 review, Dimmo, AI Agent Square; Lavender's own /pricing URL returned 404 during the v1 evidence pass and must be re-verified live before publication)

Tier Monthly Annual (~) What's included
Free $0 $0 5 emails / month scored; basic coaching
Starter $29 / mo (some sources $27) ~$23 / mo Unlimited email scoring, AI coaching, basic analytics, Chrome extension for Gmail / Outlook
Pro $49 / mo ~$39 / mo Adds advanced personalization, communication-style matching, deeper analytics
Teams $69 / seat / mo ~$55 / seat / mo Team analytics dashboards, shared templates, manager visibility
Enterprise Custom (~$89+ / seat / mo) Custom Custom AI training on company emails, API access, dedicated CSM, compliance bells

Pricing footnotes for buyers:

Ora — autonomous agent, separate SKU, beta: ~$500 / agent / month, 1,000 emails/month per agent (~4,000 contacts/year), month-to-month, no annual lock-in, build phase free (VENDOR-CLAIMED on lavender.ai/ora; corroborated in the Aug 21 2025 launch coverage and confirmed by Will Allred on camera in Video #9 of the demo dossier). Ora is a separate SKU from the Lavender coach and is not blended into Lavender's score for v1 of this guide. If Ora exits beta during the test window, we will revisit; otherwise it gets a footnote.


Integrations (CITED — multiple third-party review sources; field-level confirmation deferred to the May 5–18 trial pass)


Editorial assessment

Lavender is the only tool in our 12-vendor cohort whose entire engineering surface is dedicated to making one email better, and that focus is the single most consistent positive signal across third-party reviews. Where every other vendor in the test set treats personalization as one feature among many — sequencing plus data plus reporting plus a generative-AI bolt-on — Lavender treats it as the product. The Personalization Assistant pulls structured recipient context from LinkedIn profile fields, recent news, recent X / social posts, podcast episode summaries with links, and job-history entries into a panel beneath the live 0–100 score, and the score updates in real time as the rep types. As Originality.ai's August 2025 review put it, Lavender's coach is "the surface where research-then-mention happens" — and that framing matches what we see in Video #2 (the post-Lavender-3.0 January 2025 Coach walkthrough) and Video #6 (the third-party June 2025 live-rewrite demo). The October 2024 Lavender 3.0 redesign moved the sidebar to the left rail for "improved left-to-right readability," refreshed the color palette (purples plus lime-green accents), added a Frameworks feature (customizable email foundations with suggested beats and rationales, replacing rigid templates), and shipped Start My Email (one-click AI generation from a recipient address plus a short prompt). Buyers researching Lavender on YouTube content from before October 15, 2024 are looking at an obsolete right-rail UI that no longer ships and should disregard pre-3.0 demo footage when evaluating the current product.

The 0/4 scores on deliverability infrastructure and data accuracy are not quality failures — they are category-fit caveats, and any reader treating them as quality signals is reading the rubric wrong. Lavender does not own send schedules, throttling, domain warming, sender reputation, spam-filter risk, or a B2B contact database. By design, Lavender is upstream of deliverability and orthogonal to data. The honest comparison is not Lavender-versus-Outreach (different categories) but Lavender-as-a-layer-on-top-of-Outreach-or-Apollo-or-Lemlist. This is also Lavender's strongest commercial story: because it does not send, it does not threaten existing sequencer investments. A team running Outreach plus Apollo plus Clay can layer Lavender on top with no rip-and-replace, which makes it the most-buyable tool in the cohort for risk-averse RevOps teams. The MarketBetter January 2026 review and the Reply.io 2026 review both call out the manager-side Coaching Dashboard as the differentiated reason to upgrade to the Teams tier, and the pricing — $0 / $29 / $49 / $69 — is one of the cleaner pricing pages in the cohort (no usage credits, no per-send fees, no contract minimums published).

The structural ceilings are real. Single-channel by design: a buyer running multi-channel outbound (LinkedIn DMs plus email plus dialer) gets coverage on one third of their motion, and buyers building toward an Outreach-replacement or 11x-replacement should not consider Lavender a candidate. The Chrome-extension stability complaint is recurring: multiple G2 reviews, Chrome Web Store ratings (4.4 / 5 across 37 ratings, 20,000 users as of December 2024), Reply.io's 2026 review, and Woodpecker's review aggregation cite extension crashes, slow loads, and the sidebar occasionally not rendering — and for a tool whose entire UX is a sidebar this is a structural risk, not a polish issue. The vendor's own sizzle reels (the Mux-hosted homepage 2:59 reel and the Coach Walkthrough) show snappy, instant scoring updates that consistently over-represent the perf-in-the-best-case state; the lived buyer experience in the wild is meaningfully more variable. AI suggestions plateau for experienced writers: Originality.ai's reviewer noted that "the same shorten this, rephrase that patterns repeat after the first weeks of use," and the product's sweet spot is converting bad-or-average writers to good ones, not turning good writers into great ones. Originality's AI-detection model also flagged Lavender's raw generated output as "100% AI-written" — i.e., the prose has the statistical signature of AI generation rather than human writing pre-coaching, and Originality recommends "blend AI output with personal writing for better results." The score-and-fix loop is what closes the gap; the unaided generate-from-scratch motion is not yet at the indistinguishable-from-human bar.

The Ora question deserves separate framing for buyers reading this tear-sheet. Lavender's autonomous-agent product Ora (ora.im) launched publicly on August 21, 2025 with Will Allred unveiling on camera (Video #9 in the demo dossier), positioned to compete in the same market as 11x.ai, AiSDR, and Regie.ai. Ora is a different SKU at ~$500 / agent / month for 1,000 emails, runs in OCD ("Optimize, Confirm, Deliver" — review-and-approve) or LFG ("Lead Flow Genius" — fully autonomous) modes, and ships proprietary risk-mitigation scanning (legal compliance, bias, hallucinations, "Vibe Check" interpersonal-EQ assessment) before any send. Ora has the strongest video evidence of any feature in the entire demo dossier — co-founder unveil, internal-dogfooding video, custom-campaign builder demo, plus an independent reviewer (Video #12) who published in April 2025, four months before the public launch with hands-on early-access footage. For the 2026 research report we are scoring Lavender, not Ora. They are different SKUs with different value propositions and different scoring exposure. If Ora exits beta during the test window we will revisit; otherwise it gets a footnote, the same way we treat Reply.io's Jason AI as a separate SKU from Reply Multichannel.

Strategic position. Lavender is the cleanest layered-on-top buy in the cohort: cheapest credible AI coaching layer at the Starter tier ($29 under most discretionary thresholds), most-buyable risk profile because it does not threaten existing sequencer or data-vendor contracts, and the only product whose engineering surface is single-mindedly oriented to making one email better. Buyers should not buy Lavender expecting deliverability, sequencing, data, or multichannel coverage; buy it for what it is, layer it on top of whatever already sends, and the v1 personalization-quality score (4/4, top of cohort) earns its keep.


Best for

Not a good fit for: - Teams replacing SDR headcount with autonomous AI (use Ora separately, or look at 11x.ai / AiSDR / Regie.ai) - Teams that do not write their own emails (fully sequenced templates) — coaching only fires on composed text - Buyers running multi-channel where email is one third of the motion and the other two thirds need coverage too - Procurement-heavy enterprise — $14M total funding and an ambiguous post-2023 raise will trigger vendor-risk review

Stage fit (per scoring rubric): - Foundation: yes — cheapest credible AI coaching layer - Pilot: yes — layers cleanly on Outreach / Salesloft + Apollo; manager analytics justify Teams tier - Scale: yes — best-of-breed coaching layer for mid-market SaaS already running a sequencer - Optimization: yes — coach + manager analytics differentiated


Right of reply

Lavender will receive this draft tear-sheet 7 calendar days before publication per Yardstick's standard policy. Likely correction surface area: - The exact current Starter price ($27 vs. $29) and the Enterprise ~$89+ figure (single source) - Customer count — Lavender may have a fresher 2026 number than the ~11,000 sellers cited from TechCrunch February 2023 - Will Allred's current title (one secondary source lists him as CEO; LinkedIn and Lavender's own copy list him as Co-Founder & COO — likely host-side title error, but worth a clean confirm) - LinkedIn-data attribution for the Personalization Assistant — LinkedIn API (unlikely without a partnership LinkedIn rarely grants), scraping, or third-party data provider — material for compliance-sensitive buyers - The headline of any vendor-claimed metric (we will quote them, but they may want to swap out which case study we feature) - Ora positioning — Lavender may push for Ora to be scored alongside Lavender; we will decline unless Ora exits beta before publication

Rankings are not subject to appeal. Only factual corrections accepted.


Sources

Vendor: - Lavender homepage - Lavender Coach product page - Ora — Lavender's autonomous agent - Lavender 3.0 launch blog (Oct 15 2024) - Lavender Ora — Healthcare client case study (Aug 21 2025) - Will Allred LinkedIn

Funding / corporate: - TechCrunch — Lavender lands $13.2M (Feb 2023) - Hypepotamus — Atlanta-based SaaS Lavender founder profile - Crunchbase — Lavender funding profile - Latka — Lavender revenue / headcount profile

Independent reviews (2024–2026): - MarketBetter — Lavender AI Pricing Breakdown 2026 - MarketBetter — Lavender AI Review 2026 - Reply.io — Lavender AI Review 2026 (note: competitor-authored) - Originality.ai — Lavender Review (Aug 2025) - Dimmo — Lavender Review 2025 - Dimmo — Lavender demo video (homepage 2:59 sizzle, embedded) - G2 — Lavender Reviews (G2 returned 403 to direct fetch in our pass; star rating UNKNOWN — re-verify before publication) - Lavender Chrome Web Store listing — 4.4/5 stars, 37 ratings, 20,000 users, last updated Dec 2 2024

Demo videos (full inventory in demos/lavender.md): - Video #1 — How to Use Lavender's AI Email & Personalization Coach (canonical product tour) - Video #2 — Lavender Coach Walkthrough, Jan 2025 (post-3.0 walkthrough) - Video #6 — Lavender AI Tool Fixes BAD Cold Emails, Jun 2025 (third-party live-rewrite) - Video #9 — Will Allred Introduces Ora, Aug 21 2025

Founder talking points: - GTMnow podcast — "Write Sales Emails That Stand Out with Will Allred" (May 17 2023) — source for the "20k+ active inboxes / millions of emails" data-moat claim


Tear-sheet compiled 2026-04-29 for the Yardstick Research 2026 Q2 Yardstick Report to AI Sales Agents. Every claim labeled per the report's evidence policy. The 0/4 scores on Deliverability infrastructure and Data accuracy are absent-by-design category caveats, not quality signals; readers using this rubric to pick a tool should match the vendor's category to their actual gap, not optimize blindly on total score.