Docs Core concepts Archetypes & lookalikes
Core concepts · 03 of 04

Archetypes and lookalikes — scaling the briefing motion.

Briefing one account is research. Briefing 50 a week requires a pattern. Archetypes are Mama's pattern-recognition layer: clusters of closed-won accounts grouped by signal-mix + persona shape. Lookalikes are the 'next 5' the rubric surfaces.

Time: 6 min·Updated: 2026-05-25·Audience: experienced Mama users·Deep dive: long-form essay

TL;DR

Archetypes group your closed-won accounts by the signals that preceded each win and the persona shape of who bought. Each brief inherits its archetype from the source account — and the lookalike rail at the bottom of every brief surfaces the 5 next-best accounts in the same archetype. The Reply Loop trains archetypes on what actually replied, so the matcher gets sharper every week.

01What an archetype is

An archetype is a cluster of past wins that look alike. Specifically: same signal pattern that preceded the win, same persona shape that bought, same firmographic envelope. When Mama detects these clusters in your closed-won data, it turns each cluster into a named archetype.

Example archetype (hypothetical): "Series-B SaaS with new VP Data hire + Snowflake adoption". Five of your 30 closed-won accounts fit this shape. Mama gives the archetype a name, stores the signal-fingerprint, and uses it as a target shape for new accounts.

02The 4-step mechanism

How an archetype gets created and used:

StepWhat happensWhen
1 · ClusterMama groups your closed-won accounts by signal-mix + persona shape using a similarity scoreNightly batch
2 · NameEach cluster ≥3 accounts becomes a named archetype (auto-generated, you can rename)Nightly batch
3 · MatchEvery new account in Mama gets matched against all your archetypes — top match becomes the brief's archetypeOn brief generation
4 · TrainReply behavior on emails sent to that archetype updates the archetype's confidence scoreReal-time (Reply Loop)

03What a populated archetype looks like

Sample archetype card (from your "Archetypes" screen in the app):

Archetype: "Series-B SaaS · VP Data + Snowflake"
Members: 7 closed-won, $890K total ACV
Signal fingerprint: Recent funding (88% of members) · VP Data hire in last 6 mo (100%) · Snowflake in tech stack (100%) · Headcount 80-300 (86%)
Persona shape: VP Data (economic buyer) + Senior Analytics Engineer (technical champion)
Reply rate (current): 14.2% (vs. workspace avg 6.8%)
Confidence: 0.81 — high, based on 31 sent emails

The reply rate matters: an archetype with high reply rate becomes a higher-priority lookalike target. Archetypes with low reply rate get de-prioritized in lookalike ranking.

04How lookalikes get ranked

Lookalikes appear at the bottom of every account brief as a "next 5" rail. They're ranked by composite score:

  • Archetype match strength (40%) — how closely the candidate matches the briefed account's archetype fingerprint
  • ICP rubric score (30%) — the candidate's standalone fit
  • Signal freshness (20%) — how recent the candidate's active signals are
  • Untouched-ness (10%) — accounts your team hasn't already emailed in 90 days get a small boost

Click any lookalike to generate a fresh brief — the new brief inherits the same archetype context, so the matching templates and playbooks surface immediately.

05How Reply Loop trains archetypes

The Reply Loop is what makes the system get smarter over time. When a prospect replies to your outbound, the loop:

  1. Classifies the reply (interested · objection · unsubscribe · referral · noise)
  2. Logs the outcome against the archetype that generated the brief
  3. Updates the archetype's confidence + reply-rate metric
  4. If a new pattern emerges (a sub-cluster within the archetype that replies at 2× the parent), proposes a new archetype

This is the feedback loop that makes archetype matching defensible. The more your team uses Mama, the smarter your archetypes become. More on Reply Loop →

06Why this is the wedge

This is the part most users don't realize until month 3. Archetypes are not portable. They're built from your closed-won data, trained on your reply behavior, scored against your ICP rubric. A competitor importing the same signal data has no way to replicate them.

This is also what compounds over time. Month 1, archetypes are rough. Month 6, they're sharp. Month 18, they're a meaningful competitive advantage that the team would lose if you switched tools.

The moatThe signal-data moat is real but copyable — competitors can buy the same Crunchbase + LinkedIn feeds. The archetype moat is yours alone. It's why we built Mama, not just a signals feed.

07Common mistakes

Expecting archetypes to be useful in week 1
You need ~15 closed-won accounts in Mama to get meaningful archetype clusters. Backfill closed-won via CSV import on day 1 — don't wait for new wins to accumulate.
Ignoring low-confidence archetypes
A 0.4-confidence archetype is telling you it's still learning. Don't disable it — the more emails you send into that archetype, the faster confidence climbs.
Manually editing every archetype name
Auto-generated names like "S-B-SaaS-VPData" are uglier but informative. Rename only the archetypes you reference often.
Not pushing closed-won status to Mama
If you close deals in HubSpot but never push the status back to Mama, the archetype matcher doesn't know what won — no training signal. Enable the CRM sync.
Was this page helpful?