What contacts are
Contacts are Dovetail’s built-in CRM: a single record for each person you talk to. A contact holds the standing facts about that person — things like their company, role, region, plan, or customer tier — so you don’t have to re-enter that context on every call. Each contact also has a profile that gives a full view of the person: a summary of who they are and a list of every call or piece of data they’ve appeared in. Attach the same contact across many touchpoints and Dovetail unifies them under one profile, which is what lets you ask things like “what have we heard from this person over time” or “when did we last talk to them.”Should your team use contacts?
The most important thing to know up front: contacts are a tool, not a requirement. As with most things in Dovetail, if they’re useful to your workflow, use them; if they’re not, you can carry on as though they don’t exist. Plenty of teams never touch them. The deciding question is simple: are you going to talk to the same people again? Contacts are worth it when you speak with the same people repeatedly and want to track and connect those relationships over time — recurring customer or partner conversations, an ongoing research panel, or any case where several teammates need visibility into one person’s history. This is exactly why Dovetail’s own customer-facing teams lean on contacts heavily. Contacts are usually not worth it when:- You rarely talk to the same person twice. For large, general-population research where the goal is a fresh participant every time, a contact record just adds noise and upkeep for little payoff.
- You handle sensitive PII or work in a regulated space (medical, pharmaceutical, finance) where storing person-level records isn’t appropriate.
- The attributes you care about change too often to maintain. If people’s roles, offices, or details churn constantly, the upkeep can outweigh the benefit — and there’s a specific reason why, covered below.
Your own team are users, not contacts
One rule applies to everyone: don’t make your internal teammates contacts. Mark them as Dovetail users (or placeholders). This tells the AI they’re the moderator or interviewer, not the research subject, so it won’t treat your own comments as participant data. The AI can still read what a user said when you explicitly ask (for example, “what did I promise to send after this call?”), but it won’t fold moderator talk into customer themes.If you use contacts, here’s how
Attaching a contact
There are two ways to connect a contact to data:- In the transcript — the easiest path. Map a speaker to a contact, and from then on every quote is attributed to that person and their fields flow through. The call also shows up on their profile.
- As a field on the data object — used when there’s no transcript to map a speaker to (notes, usage data, an unrecorded item). Applying the contact as a field still ties that data point to the person.
Contact fields vs. data fields
Attributes that are persistently true about a person (company, role, persona, tier) can live on the contact, while fields that describe a specific call (the feature discussed, the journey stage) live on the project or data object. If you commit to using contacts consistently, this is a real advantage: you don’t need to recreate “company” or “customer tier” fields on every project, which means fewer fields for contributors to fill in. Reserve your workspace fields for what the call is about. One important behavior to understand: editing a field on a contact updates that value everywhere the contact has ever been attached — there’s no point-in-time history. Changing someone’s role from associate to manager rewrites it across all their past data. That’s ideal for keeping a CRM-style record current, but it’s the reason teams who need “what was true at the time” should keep that attribute as a call-level field instead.Syncing from a CRM, or uploading a CSV
If you maintain contacts in a CRM like Salesforce, you can sync them so contact fields populate and stay current automatically. A few things to know:- The sync enriches the contacts you already have in Dovetail rather than importing your entire CRM — so you bring across a relevant subset, not tens of thousands of records.
- It’s one-way: the CRM is the source of truth, and edits don’t write back to it.
- You choose which fields map across (sync revenue and region, skip billing address, and so on), and it refreshes on a schedule (daily by default) so values like ARR update on their own.
- Email is the anchor. It’s the identifier that links data across sources and prevents duplicates, so a known speaker gets matched to their contact automatically.
Segments
Build a segment from any contact field or combination of fields — for example, company size plus a revenue range to define an “SMB” segment. Segments do two jobs: they help you find the people who match, and they let you filter your data to just those people’s calls. They also work in chat, so you can say “only look at data where the contact is in this segment,” which is a clean way to slice your analysis by who the feedback came from.Privacy and anonymization
Contacts have their own access controls, from no access (can’t open the database) through can-use, edit, and full access — so you decide who can see or change contact information. To anonymize, remember that a contact’s name is the same everywhere; you can’t mask it per project. The standard approach is to name the contact itself with a coded handle (consistent initials and a number, not “Participant 1, 2, 3,” which becomes impossible to tell apart) and keep the real name in a field, then lock down field and profile visibility so restricted users see only the handle. For truly sensitive data, redaction rules set at the workspace level can also strip names, IDs, and similar details.The short version
- Contacts are an optional internal CRM — use them only if you’ll talk to the same people again.
- Skip them for one-off audiences, sensitive PII, or fast-churning attributes; store one-off identifiers as project fields instead.
- Keep your own team as users, never contacts.
- Attach contacts in the transcript, let them carry person-level fields, and sync from your CRM (one-way, field-mapped, email-anchored) or upload a CSV.
- Use segments to filter data by who it came from, and lean on access controls, coded handles, and redaction when privacy matters.