A real audit, start to finish.
A pre-revenue medtech company, 8 people.
Runs on partnerships and is mid-fundraise. Attio for about a year, set up by the CEO herself. We read the workspace over two days with Expert access.
This is a real audit. The company is not named and the quote is attributed by role, but every number below is the real measured figure.
“It has a lot of positives, but if it doesn't have the right setup it can be a bit of a disaster. I look at it and then I turn it off, and I waste so much time trying to fix it that I'm not getting anything done.”
What we found.
The problem is not the data she cares about. It is volume and signal-to-noise.
Fill rates are good: 92% of people have an email, 94% are linked to a company, 84% have a LinkedIn. Better than most workspaces we see. But the workspace holds 7,539 people and 5,246 companies for an 8-person company. 55% of those people have never had a single interaction. Almost 4,000 are LinkedIn contacts scraped in and never engaged. So when she opens Attio, 3 out of 5 records are noise, and the ~500 people who actually matter this year are buried in it. That is the "I turn it off" feeling, measured.
This is a cleanup and wiring job, not a rebuild. Nothing needs re-modelling.
7,539
People in the workspace
55%
Never had one interaction
64%
Arrived in a single month
~500
People who actually matter
What she is already getting right.
- Fill quality is high where it counts. 92% email, 94% company link, 84% LinkedIn, 70% job title. The raw material is there.
- She modelled the real business. Deals carry a Fundraise Stage separate from the sales stage, plus Capital Type and Investor Type. Events for conferences, Grants, Programs, Clinical Sites. That is a thoughtful object model, not a default one.
- She has been fighting entropy the right way. Dedup lists, a Possible Duplicates triage, duplicate-check fields. The instinct is correct. It just does not scale by hand.
The findings, ranked.
Ranked by what each one is costing her, not by how easy it is to fix. Each has the fix attached.
55% of people and 68% of companies are dead weight.
4,155 people and 3,575 companies have never had one interaction. About 3,970 of those people are LinkedIn contacts synced in and never engaged. This is why the workspace feels unusable: the live signal, the ~500 people she has actually talked to, is drowned by ~7,000 cold records. Every filter, every list, every search has to wade through it.
The fix. Segment, do not delete blindly. Build a Cold / Unengaged holding list and get it out of the default views. The workspace stops feeling like noise the moment you do this.
Fundraising lives in lists, not in the pipeline she already built.
There is an Investors list (254 people), a Startup Fundraising list (189 companies), and a Pre-Seed Fundraising deals list (2). But the Deals object, which already has a Fundraise Stage status, Capital Type, and Investor Type, holds exactly 2 records. So the thing she most wants to run is spread across three lists with no stage and no next step. She cannot answer who is in diligence, who owes her a reply, who has gone cold.
The fix. Stand up an Investor Pipeline on the Deals object, one deal per investor conversation, seeded from the 254-person Investors list. Highest-leverage single move.
Two of five automations are down, and a third barely runs.
The revalidate-on-invalid workflow has been failing for 23 days, so the stale-data guard is broken. Draft LinkedIn follow-up tasks for unanswered outreach, the one automation that would directly fix "I forget to follow up," is paused and failing. A weekly company lookup has fired exactly once.
The fix. Repair the two failing workflows, decide keep or kill on the paused legacy ones.
The duplicates are real, but not the ones she thinks.
Attio enforces unique email addresses, so two records with the same email are impossible here. Hard duplicates left: about 4, plus the 28 already flagged. The ones that slip through are the same human as two records, a LinkedIn record on a personal email and an email-sync record on a work email, with a slightly different name spelling. Exact-match tools cannot catch those. Realistically this is tens of records, not thousands. The felt duplicate problem is mostly junk and staleness wearing a duplicate costume.
The fix. Fuzzy matching on name, company, and LinkedIn. Then merge.
Conference capture is built but empty.
The Events object exists with 4 conferences in it, and every one has zero contacts linked. So the exact thing she wanted, tag a conference and see everyone met there, cannot work, because nobody is attached. There is no card-scan path either, so conference contacts only arrive if the person happens to connect on LinkedIn.
The fix. Wire Events to People so tagging is one click, and add a phone-photo to contact flow. Replaces the scanner she misses, at no per-scan cost.
Schema bloat: 83 fields on People, 55 on Companies.
Most of it is import residue and half-abandoned experiments, which is why editing a record is overwhelming. Two Persona fields. Four separate company fields. A text field titled "Work Email Address (PRIMARY)" competing with the real email field. Around 8 overlapping triage selects including trash, keep, and ARCHIVE.
The fix. Consolidate to one Persona, one company link, one status and priority pair. Collapse People 83 to about 40 fields. Half-day job.
The runbook we sent back.
This week
Low effort, unblocks the trip she is about to take
- Stand up the Investor Pipeline on Deals from the Investors list.
- Repair the two failing automations.
- Create the Cold / Unengaged holding list and drop it out of default views.
This month
Medium effort
- Fuzzy-merge the personal and work email pairs, clear the flagged duplicates.
- Wire Events to People and set up the business-card photo flow.
- Slim the schema. People 83 to ~40 fields, Companies 55 to ~30.
Later
As she scales
- Keep titles and companies fresh on the live segment.
- Retire the shared-login members, decide keep or kill on the empty objects.
We will do this for your workspace, free.
Same read, same ranking, same runbook. Back in two business days, and yours to keep whatever you decide to do next.
Book the audit call →