Methodology

What we actually look for.

The Attio Audit runs 50+ analyzers grouped into four categories. Here is every analyzer, what it flags, and why it matters. No hidden heuristics.

Want to see the output before committing? See a sample report.

Category 1 · Foundation

Schema & Structure

Is the Attio data model serving you, or fighting you? We look at the shape of your objects, attributes, relationships, and pipelines.

  • Object bloat

    Custom objects with fewer than 10 records — usually abandoned experiments.

  • Attribute count outliers

    Objects with wildly more (or fewer) attributes than peers — tech-debt heatmap.

  • Suspicious relationship cardinality

    Multiselect attributes nobody uses that way, single-value links dense enough to suggest N:N.

  • Select option sprawl

    Status/select attributes with too many options to be meaningfully filtered.

  • Pipeline stage sufficiency

    Deal pipelines with too few stages to separate healthy vs stuck work.

  • Pipeline modeling ambiguity

    Status attributes where the stage names don't line up with the workflow Attio can see.

  • Versioned object names

    "Companies v2" patterns that suggest a migration you never finished.

  • Required but empty paradox

    Fields Attio says are required, but records still manage to have unset.

  • Archived but populated

    Archived attributes with live data — hidden but still drifting.

Category 2 · Hygiene

Data Quality

The kind of mess that silently breaks forecasts, outbound, and dashboards. We focus on duplicates, junk, and key-field population.

  • Duplicate companies (by domain + fuzzy name)

    Pairs and clusters sharing a root domain or near-identical names.

  • Duplicate people (by email)

    Same contact imported twice, usually by two different pipes.

  • Duplicate deals on the same company

    Two reps logged the same opportunity — pipeline double-counted.

  • Junk company candidates

    Companies with no deal, no linked person, and a missing name or domain.

  • Orphaned people

    People with no company, no deal, and a placeholder-looking name.

  • Test & demo artifacts

    Records whose name / domain / email screams "leftover from testing."

  • Attribute density heatmap + graveyard

    Which attributes are filled in, and which are effectively dead weight.

  • Key-field population

    Per-object report card on the fields used for reporting and outbound.

  • Company segmentation completeness

    Categories, industry, and location — the fields ICP filters depend on.

  • ICP contact completeness

    Missing job titles, phones, LinkedIn URLs, business emails on deal-linked contacts.

  • Zombie deals + stale record distribution

    How much of your pipeline and workspace hasn't moved in 60 / 90 / 180+ days.

  • Stale-by-status

    Records stuck in the same status for 90+ days, even without SLA configured.

  • Stage distribution anomaly

    Pipelines where >40% of open deals sit in a single non-terminal stage.

  • Ownership vacuum

    Records owned by nobody, or by a deactivated workspace member.

  • SLA violations

    Records past the target_time_in_status the workspace itself configured.

  • Invalid emails & phones

    Unparseable or missing contact points.

  • Name hygiene

    Name equals email, null last names, placeholder literals ("Unknown", "N/A").

  • Cluster orphans + domain typo clusters

    Groups of records that probably belong together but don't link up.

Category 3 · Team usage

Adoption & Team

Who's actually using the workspace, who's bottlenecked, and where activity is slipping.

  • Single point of failure

    When one person creates >70% of the activity across an object.

  • Owner load distribution

    P50/P90 deal count per owner — spot rep burnout.

  • Workspace member roster

    Active vs deactivated, including stale offboarding.

  • API app vs human actor ratio

    How much of the workspace's activity is automations vs actual reps.

  • Note frequency decay

    Are people still writing notes, or did adoption fall off a cliff?

  • Task debt + completion latency

    Overdue tasks, median close time, assignees who've left.

  • Meeting recording coverage

    Active-stage deals whose meetings have no recording attached.

  • Meeting writeup latency

    Time between meeting and post-meeting note.

  • Empty lists

    Lists created and never populated — usually failed rollouts.

  • Unresolved old comments

    Comment threads open for more than 30/60 days.

Category 4 · Integrations

Automation & Integration

The connective tissue to the rest of your stack. We check what's firing, what's broken, and what's silently drifted.

  • Webhook inventory

    Every webhook configured, what events it listens to, and where it's pointed.

  • Broken webhooks

    Endpoints failing >50% of recent deliveries — data isn't arriving downstream.

  • Integration fingerprinting

    Which third-party tools are creating records, and how much of the workspace they own.

  • File storage provider distribution

    Attio-native vs Google Drive / Dropbox / other — consistency check.

  • SCIM readiness

    Provisioning health indicators so IT can onboard and offboard cleanly.

How the score works

How is the health score computed?

Each category starts at 25 points, for 100 points total. Findings deduct based on severity: critical −10, high −5, medium −2, low −0.5, info 0. Per-category totals are clamped at 0.

Why four categories?

They're the smallest set that cover the lifecycle of CRM health: the shape of the data (schema), the state of the data (quality), the humans using it (adoption), and the systems around it (automation). Every finding cleanly lives in exactly one of these.

How are findings sorted inside a category?

Critical → high → medium → low → info. Within a severity, analyzer-defined order. The filter bar on the report lets you narrow to a single severity or a specific object (companies, deals, etc.) across every category at once.

Run this against your workspace.

Connect via Attio OAuth, read-only. Four minutes from start to a private, shareable report.

Start free audit →

Or see a sample first.