How to automate conference follow-up with AI

Clean the badge export first

Scanner exports arrive dirty in consistent ways: personal Gmail addresses on enterprise buyers, company names entered as ticker symbols, titles like “decision maker”.

Prompt: “Open the sheet ‘Booth scans SaaStr’. Clean it into a working copy: split full names, normalise the company name to its trading name and find the company domain, and standardise the title into a role family and a seniority level. Where the scanned email is a personal address, keep it but add a column flagging it. Do not delete any row and keep the original values in columns beside your cleaned ones so I can check your work.”

Keeping the original columns pays off later. When a row turns out to be assigned to the wrong company, you can tell whether the scanner or the cleaning produced it.

Which of these do we already know?

Deduplicating before enrichment saves money and embarrassment.

The worst follow-up you can send is an introductory email to someone mid-pilot with you.

Prompt: “Match every cleaned row against our CRM by email, then by domain plus surname. Split them into four lists: existing open opportunities, existing contacts with no open opportunity, known companies with a new person, and genuinely new companies. For anyone attached to an open opportunity, tell me the owner and the last activity date, and do not enrich or draft anything for them. Those go to their rep.”

That fourth list is what the sponsorship was for. Knowing its true size within a day is the most honest event ROI number available.

Fill the gaps without guessing an address

Now spend enrichment effort, and only on the rows that survived.

Hunter finds and verifies addresses. Apollo fills in the firmographics and the org chart around a person.

Prompt: “For the new-company list: use Hunter to find the work email where the scan gave us a personal address, and verify every address we intend to use, keeping the verification status in a column. Use Apollo to enrich each person with confirmed title, seniority, and company headcount, and to tell me whether anyone more senior in the same function also attended. Where Hunter cannot find or verify an address, write ‘unverified’ and leave the row. Never construct one from a naming pattern.”

The verified-status column is what lets you send in two batches and protect your sending domain on the first one. Same discipline as our lead enrichment work, under a deadline.

What did you actually say to them?

No system holds your booth conversations.

A companion prepares everything around them and hands you a checklist.

Prompt: “Give me the new-company list as a checklist, one row per person with company, title, and their scan time. Leave two empty columns: what we discussed, and tier. Tier is A for a real conversation with a next step, B for a genuine chat, C for a badge scanned in passing. I will fill them in now.”

Filling that in takes about fifteen minutes while the event is fresh, and everything after it is assembly.

Draft three tiers, send none

Prompt: “Create the CRM records for the new companies and people, with the source set to the event and my tier and notes in the record. Then draft one Gmail follow-up per person, matched to their tier: for A, reference exactly what we discussed and propose two specific times from my calendar next week; for B, one paragraph on the thing they asked about plus a relevant link, no meeting ask; for C, three lines saying it was good to meet, with a single link and no ask at all. Under 120 words each, no ‘it was great connecting’, no mention of the booth number. Leave every one as a draft.”

Tier C is the discipline. Sending 200 people a meeting request because they walked past a scanner is what makes everyone else’s follow-up unwelcome. The ongoing version of this shape is the sales prospecting workflow.

How do you work the long tail afterwards?

Prompt: “Two weeks after the event, list everyone from tiers A and B who never replied and never got a meeting, and check whether anything new happened at their company since. Draft a second note only where something genuinely did. Then every Monday for six weeks, tell me which event contacts have had no activity at all, so I can decide whether to keep them or let them go.”

The second half of that prompt is the part teams skip. Deciding out loud that 140 contacts are done beats a list that haunts a CRM for two years.

Frequently asked questions