Adobe announced the Import Leads agent for Marketo Engage at Summit a few weeks ago, and it's now rolling out to instances over the coming months. I had access to a Champion Sandbox, so I ran a messy CSV through it end to end. Here's what the workflow actually looks like.
For the broader take on where this fits in a full list operations workflow, see Marketo's Import Leads Agent: What It Does and What Production-Grade List Operations Look Like.
Finding Build with AI
On the My Marketo page, there's a new Build with AI tile. Clicking in takes you to the AI-powered Marketing Operations page. You can chat directly to get started, or scroll down to the Import Leads agent and click in from there.
Selecting the agent pre-populates a starter prompt and sends it through, so the agent gives you initial guidance on how to begin and points to other AI assistants you can call.
Uploading the CSV
I prepped a list of messy records — first name, last name, company, email, title, phone. Phone numbers in a variety of formats. First names with inconsistent capitalization. Companies a little jumbled. Emails with typos. The kind of file you actually get from a trade show.
Click the paperclip, attach the CSV, send. After a few seconds, the agent previews all the records.
Field mapping
The Mappings tab shows source columns mapped to target Marketo Engage fields with a confidence rate next to each. For obvious matches like email, the confidence is essentially 100. For less obvious ones — say a Company Name column instead of Company — the agent surfaces the confidence so you can verify before committing. That's a thoughtful piece of UX.
Data quality rules
Once mapping is confirmed, the agent prompts you to run data quality rules. In this sandbox, the available rules were pre-configured. When this rolls out, you'll be able to configure your own per instance.
I selected:
- Normalize email (lowercase)
- Normalize phone
- Remove duplicates
The agent also tells you why each rule matters. For folks who are less data-savvy, that nudge is genuinely useful — it keeps people from skipping a rule they should be running.
I hit send, and the agent applied the rules.
The history log
The History tab shows a clear log of everything the agent did: lowercased the emails, normalized all the phone numbers, removed the one duplicate row. Going back to Preview, the record count had dropped from 13 to 12, the phones were consistent, and the emails were all lowercase.
I also ran a Title Case Names rule, which fixed some funky capitalization (Norville Rogers, hello). I didn't have a rule available to clean up job titles, so I left those as-is — but it's the kind of thing that becomes possible as you build out your own rule library.
Importing into Marketo
Click the Import to Marketo tab. In this test instance, the only target was static lists. Future versions will support programs too.
I named the destination list, ticked the PII consent box, and hit Approve and import. The agent submitted the import, ran a verification pass, and flagged three warnings.
The warnings were all invalid email addresses: a double @, a missing .com, and one more malformed address. The right fix is to edit those three rows and rerun them through the import. The other 12 records went straight into the static list, which I confirmed by hopping over to the list view.
What I'd watch for
Two takeaways from running this end to end.
First, the rule library is going to be the thing that makes or breaks this for any given instance. The defaults in the sandbox are sensible, but the real value shows up when teams configure rules that match their actual data quality issues — persona mapping, job-level classification, regional firmographic cleanup. Adobe gave us a good chassis. The rules are where the fit-to-business work happens.
Second, the validation step is conservative — invalid emails get flagged rather than silently dropped or auto-corrected. That's the right default for anything touching the lead-to-revenue pipeline. Better to send three rows back to the marketer than to have a bad email creep into a nurture stream.
Let us know what other demos you want to see of the new AI agents Adobe is rolling out for Marketo — leave a comment on the YouTube video or reach out directly.
