2
3 Comments

We decided our CRM will never have custom objects. Here's the line we drew and what it costs us.

Last week I posted about reading 180 Attio reviews before building anything. The most useful comment asked a fair question: if setup is the problem, isn't the fix just better defaults on a flexible tool? Why give up flexibility at all?

We spent the week on that and landed on a rule. Columns yes, tables no.

Columns means custom fields. You can add any field you want to a lead, an account or a deal. Text, number, date, checkbox, single select, multi select. 6 types. Make any of them required. Your industry has a field we didn't think of, add it.

Tables means custom objects. A new record type with its own relationships, views and permissions. We won't build this. Not now, and not as some enterprise tier later either.

Here's why the line sits there and not somewhere else.

A new column is 1 decision, made once, by 1 person. Everyone else just sees a new box on the form. A new table is a different animal. It needs a name, relationships to the existing objects, its own views, a rule about who edits it, and every person on the team has to learn where it lives. On a team of 6 with no ops person, that second thing never finishes. It half happens, and half a data model is worse than none because people trust it.

The reviews back this up. The 3 happiest groups in 180 reviews were technical founders, VC firms and RevOps teams. All 3 have someone whose job is building the model. The unhappiest were non-technical founders and sales reps. They wanted a CRM, got a database kit, and were still assembling it 3 months later.

What this costs us, honestly:

• Every VC fund. They need a Portfolio object. Attio is the right tool for them, and I'll say so.
• Property, fleet, grants, anything with a domain object that isn't a lead, account, or deal.
• The "can it do X" demo moment. A flexible tool always wins that moment. We lose it on purpose and try to win week 6 instead.

The bet is that teams of 2 to 10 who want a working pipeline on day 1 outnumber teams who want to model their business, and that the second group already has 3 good options.

One thing I'm still unsure about: where "field" ends and "object" begins. A dropdown of 40 options is a field. A dropdown that needs its own contact info is an object trying to get in. We're drawing that edge now, and it won't move later, so if you've built or bought a CRM and hit this line, I'd like to hear where it bit you.

Still not live. Same as last time, not dropping a link.

on September 21, 2026
  1. 1

    Nice, this makes a lot of sense. What's been the most surprising part of it so far?

  2. 1

    This is useful. How are you finding your first users so far?

    1. 1

      Mostly through conversations and communities so far. We’re talking to small teams, founders, and people who’ve actually used CRMs and asking about where setup starts to become a problem.