Follow Up Boss tags: exclusion rules that don’t backfire

Three tags in your Follow Up Boss account probably contain the word lender. Only one of them is on a lender.

The other two are a buyer who needs one, and a batch of leads a lender partner sent over that imported with the partner's name stamped on every record. All three match the word. Exclude on it and you have pulled a live buyer and a whole referral batch out of your follow-up — silently, with no error and nothing in the activity log to tell you.

Nobody warns you about this when you set up contact exclusions, whether in a smart list, a drip tool, or an AI layer reading the same database. The rule is right for the contacts you had in mind. It is the ones you did not that cost you.

Why the rule exists in the first place

Teams run more than one business through one CRM. The lender you co-market with, the title rep, the photographer, the agents you are recruiting, the staff accounts: all of them sit in the same list as your buyers. We covered the audit side in Follow Up Boss Database: Not Everyone In It Is a Lead. The damage is not cosmetic. A study of roughly one million CRM customers at a home-appliance business analyzed the impact of each customer-profile data-quality dimension on purchase loyalty and on promotional-campaign success rate (Suh, 2023, IEEE Access). The contacts you are not selling to move the averages you decide on. So you write a rule. Here are the three traps.

Trap one: the tag describes a need, not a role

Your tag vocabulary was not designed. It accumulated. Agents tag what is going on with a contact, so a large share of your tags describe a relationship to a role rather than the role itself, and to a keyword rule those two read identically.

Ace applies the test that separates those two columns before any built-in tag pattern runs. A tag opening with a word like needs, wants, looking for, referred by, imported, sent to, working with or assigned is read as a need, a provenance marker or a workflow step, and never disqualifies anyone through the built-in patterns. That is why needs lender and Lender get different answers from the same word. (If you deliberately add such a tag to your own rules below, your rule still applies — it is the automatic guess the guard holds back, not your instruction.)

Trap two: a role that fits thousands of contacts is not a role

Count before you exclude. It takes two minutes and catches the most expensive mistakes.

If a tag reads like a role but sits on a large share of your database, it is not describing anybody's role. It is a workflow marker that happens to contain a role word. A tag like lender [name] on tens of thousands of leads is not tens of thousands of lenders — it is the assigned lender on tens of thousands of buyers. Exclude on it and most of your pipeline stops being worked.

Ace can set a tag like that aside so it stops acting as a role signal, and any tag currently set aside shows read-only in Contact roles. But that is a reconciliation we run against an account, not something that happens on its own the moment a tag gets popular — so on any given day the thing actually protecting you here is your own count, not ours. Do step two below.

The lesson outlives any one tool: a category that applies to a third of your database is not a category.

Trap three: nobody decided which rule wins

Sooner or later two rules disagree about the same person. Your account rule says the stage Partners is not a client. An agent has tagged that contact as one to work anyway. If you cannot say which answer the system uses, you do not have exclusion rules. You have exclusion suggestions, and they will surprise you the first time it matters.

Ace resolves it with a fixed order. The first check that returns an answer is the answer.

Two consequences worth keeping. A contact carrying ace-include is worked even when their stage is on your exclusion list and even when a built-in pattern would have caught them: the per-contact decision beats the population rule. And ace-ignore beats ace-include, because an explicit "leave this person alone" should never lose to a broader rule.

Both overrides are ordinary Follow Up Boss tags, so you can apply them in bulk from a smart list, including our curated Follow Up Boss smart lists. That is deliberate. There is no second list somewhere else for you to keep in sync.

Where to set this in Ace

An admin opens Back Office → Recruiting → Contact roles. Four lists, all exact-match and case-insensitive:

• Tags that mean "not a client." Suggested from the vocabulary Ace has seen in your CRM — Follow Up Boss publishes no endpoint for listing an account's tags (API reference), so what has come through on real contacts is the only honest source. You can still type one it has not seen.
• Stages that mean "not a client." Your side-business stage, your partner stage, whatever your team named them.
• Lead sources that mean "not a client." For when an entire feed is partner traffic.
• Your own email domains. The cleanest way to keep staff and test accounts out of client-facing numbers.

Saving re-checks the contacts you already have, not only the ones that get CRM activity afterwards, so a rule you add today reaches the quiet records too. The re-check runs on Ace's own copy of the classification: it writes nothing back into Follow Up Boss and retags nothing in your CRM.

Two switches under Advanced turn off the built-in stage and tag recognition. Leave them off. With them on, only the exact strings you typed are excluded — Trash, Do Not Contact and your partner tags stop being recognized until you list every one by hand.

The 20-minute test, on whatever tool you already run

None of this is specific to us. Run it against your smart lists, your drip tool, or any AI layer reading your database.

1. Write down the tags and stages you would exclude on. Longhand, before you touch any settings. Our guide to segmenting a Follow Up Boss database by lead type covers building the lists.
2. Count each one. Build a smart list per tag and look at the total. Anything matching a large share of your book is a workflow marker — strike it.
3. Read ten records at random from each surviving list. Not the first ten. If one is somebody you would happily take a call from tomorrow, the rule is wrong.
4. Check for the prefix words. Needs, referred by, imported, sent to, working with, assigned. Any of them at the front of a tag means it is not a role. Match tags exactly, never as a substring.
5. Pick your per-contact override and put it in your onboarding doc. A rule without a documented escape hatch gets worked around in ways you will never find.
6. Put it back on the calendar for 30 days out. Tags keep accumulating, and so does the gap between your rules and your database.

What this does not do

Two limits worth knowing before you rely on any of it.

It governs what Ace works, not what your CRM does. Excluding a contact keeps them out of Ace's suggestions, the seller scan, the model-score sweeps, the rules engine and Autopilot's outreach. It does not stop your own action plans, your dialer, or an agent who opens the record and sends a text.

An unclassified contact is not a protected contact. The gates named above are fail-open by default: a record Ace has not yet reached a verdict on is treated as workable rather than held back, because silently freezing every contact nobody has looked at yet would be the worse failure. The marker tags work the same way — ace-ignore takes effect the next time that contact is classified, which for a tag you just applied is the tag itself arriving. Your account rules are the other lever, and they behave differently: saving a change in Contact roles re-checks the book against what Ace already holds.

The rest of what the classifier will and will not do for you — how conservative the built-in patterns are, why they read only what is already on the record, and why a DNC registered phone tag is not treated as an instruction — is covered in the audit piece. And if your wider problem is duplicates and stale statuses rather than roles, start with cleaning up CRM data in Follow Up Boss. Exclusion rules written over a messy book inherit the mess.

Common questions

Should I delete non-prospects from Follow Up Boss instead?

Usually not. Your lender, title rep and recruiting targets are contacts you want to keep and message — just not as clients. Excluding them from client-facing scoring and outreach keeps the relationship and fixes the numbers. Deleting throws away history you will want later.

What happens to a contact I exclude by mistake?

Nothing destructive. They stop appearing in Ace's client-facing queues and scans, and stay in Follow Up Boss untouched. Two ways back: remove the rule that caught them, and saving that change re-checks your existing contacts straight away; or tag them ace-include, which lands when the tag reaches Ace.

Does excluding a contact stop automated outreach to them?

It stops the Ace side of it. Ace's rules engine checks the classification before any automated action fires, and Autopilot will not select an excluded contact, so nothing reaches the automated reply flow either. It has no effect on action plans, drips or dialers running in your own CRM or another tool.

Why is a quiet contact still showing an old classification?

Ace re-checks a contact when they get CRM activity, so a record nobody has touched in a year carries the answer from the last time something happened. Saving a change in Contact roles re-checks the whole database, which is the fastest way to bring dormant records current. For how quiet contacts resurface once they do move, see the lifecycle engine.

Can an agent set this, or is it admin-only?

Account rules are admin-level: they change what every agent sees. The two per-contact overrides are ordinary tags, so any agent who can tag a contact can apply them.

Start with the count

If you do one thing from this, do step two. Build a smart list for each tag you were about to exclude on and look at the total. The one that comes back far larger than you expected is the tag that would have cost you, and you will have found it in about ninety seconds. The settings, step by step: which contacts Ace works, and how to exclude the rest.

One thing we would like an answer to: when a rule and an agent disagree about whether somebody is a client, which one does your team let win — and has that ever gone badly?