From Badge-Scan Export to a Ranked Outreach Queue

You just got back from a show with a CSV. It has a few hundred rows across dozens of companies, some emails, some blanks, some fields in the wrong columns, and a decay clock already running against you. This guide is for founders selling their own product, account executives, SDR leads, and partnerships teams who need to turn that artifact into a ranked, safe-to-send outreach queue before the momentum fades. It carries one real list through resolution, dedup, fit-match, and ranking with the counts shown at each fork, including the wrong turns.

Most published follow-up advice starts from a clean ICP or a single named account. That is not what you hold. You hold a messy, multi-company registrant export, and the interesting work is the funnel math: how many rows survive each stage, which stage does the real damage, and how to defend the final number to whoever asks why the queue is smaller than the badge count.

What a badge-scan export actually contains

A registrant or badge-scan export reliably carries identity and event context, and reliably lacks verified contact and firmographics. That gap is the whole job.

Badge-scan and registrant exports from platforms like Cvent, Whova, and ATIV EventPilot typically carry full name, company, job title, ticket or attendee type, role, a unique QR or barcode value, and often email and phone. What they miss or corrupt is the part you actually send against: verified business email, direct dial, firmographics, and current-employer confirmation.

Several conditions degrade the data before you touch it. Registrants who used personal rather than business emails, small businesses with thin online presence, self-employed individuals, badges that belong to event staff, sprawling sub-company networks, and security-sensitive industries all resolve poorly. And column-shift corruption is real and common: raw badge CSVs silently place email in the last-name field or company in the title field, so a row that looks clean is not.

The practical read: treat the export as a list of identities to resolve, not a list of contacts to mail. Everything downstream is about converting identity into a deliverable, current, compliant, ranked contact.

The funnel and where it actually leaks

A single canonical drop-off percentage for event lists is not established publicly, so build your funnel from your own counts and label the illustrative stages as illustrative. What is established is which stage leaks most.

The biggest drop is usually the contactable-record filter, not the ICP filter. With B2B deliverable appends resolving at 25 to 60%, up to three-quarters of rows can fail enrichment before fit is ever judged. So when your final queue looks thin, suspect data resolution first and targeting second. The two are easy to confuse and expensive to confuse.

Here is the worked example this guide follows. I start with a raw export of 500 badge scans from a single B2B software conference. The counts below are illustrative, derived from the published match-rate and follow-up bands in the dossier, not a benchmark you should quote as fact. The point is the shape and the arithmetic, which you can reproduce on your own list.

Read the funnel as diagnosis. Dedup took 70 rows, a normal 14% for a multi-day show where people get scanned twice. Enrichment halved the survivors, which is the single largest cut and sits squarely in the published 25 to 60% band. The ICP filter then dropped fewer rows than enrichment did, because a conference audience is pre-qualified: 81% of trade show attendees have buying authority and 67% are net-new prospects. The compliance gate trims a handful more. If your enrichment stage is the leak, no amount of tighter targeting fixes it.

The decay clock that sets your ranking

Ranking event leads is a race against a measured decay curve, not a matter of preference. Reply rates fall fast enough that a slower, higher-fit row can be worth less than a faster, lower-fit one.

The dossier's clearest evidence is a reply-rate index tracked across BoothMaven customers at North American trade shows, with same-day set to 100. It falls to 92 within 24 hours, then collapses to 47 by 48 to 72 hours, roughly halving in two days.

Two consequences. First, the window is 24 to 48 hours; leads followed up within 7 to 10 days convert to opportunities at 20 to 30%, but at 5-plus days response rate falls to 2 to 5%, at which point you are cold-emailing a person you already paid to meet in person. Second, ranking must let a hot same-day row jump ahead of a higher-fit but slower row, because the curve halves faster than any fit advantage compounds.

The reply-rate index halves in two days, so speed is a scoring input, not a nicety.

The payoff for moving fast is real: it takes about 3.5 sales calls to close a qualified trade show lead versus 4.5 for a cold lead. You bought that advantage with the booth fee. The decay curve is how you keep it.

Enrichment: what resolves and what lies

Enrichment converts identity into a deliverable contact, and it is where you should expect the honest majority of your rows to fall away. Plan the whole queue around the match rate, not the raw count.

The bands vary by source and method. The table below is built from the vendor figures in the dossier, described abstractly because the method matters more than the name.

Two rules follow from this table. A vendor quoting 90% is counting matches, not deliverable matches, so verify deliverability after the append and track bounce rate. And a multi-provider waterfall lifts match rates 20 to 40% over a single provider, which is the cheapest way to move the biggest leaking stage.

There is a real ordering fork here. Some teams enrich first, then dedup on the enriched identity, because a verified email is the best dedup key. Others dedup before any sync to avoid paying to enrich duplicate rows. Both are defensible. Pick one, document which, and keep the choice consistent across shows so your stage counts stay comparable.

Before enrichment, resolve stale employers. Average tenure at high-growth firms is about two years, so the company on the badge is often last quarter's job. Enrichment tools flag when a contact is no longer at a company; CRM practice is to keep the old record, mark it "No Longer At Company," and create a new record at the current employer. A champion who moved is not a dead lead, it is a job-change trigger at a new account.

This resolve-and-find-the-mover step is exactly where a plain-English search beats a rigid append. Instead of guessing which badge companies went stale, you can ask directly.

Refolk resolves an identity to a current, reachable person across public LinkedIn, the public GitHub graph, and the open web, which is the friction this stage otherwise imposes one row at a time.

Fit and intent: scoring without manufacturing scarcity

Score each row on two axes: a fit score from firmographics and persona, and an intent score from what the person actually did at the event. A common split caps fit at 50 and intent at 50, then routes once the combined total crosses your threshold.

Event data is unusually rich on the intent axis because it fuses fit and intent in a single hour. One prospect can visit the booth, request a demo, exchange a card, and book a follow-up in sixty minutes, giving you a complete fit-and-intent profile that a digital lead never hands over. That is why event leads justify a lower qualification threshold than digital ones.

Weight the intent signals from weakest to strongest. A published model gives concrete numbers you can adopt directly.

The gap between a registrant-only no-show and a session attendee is the single most useful distinction on the list, and a traditional attendee export flattens it: both appear identical. Separate registrant-only from badge-scanned from session-attended before you score, or a registered no-show will score like a booth visitor.

On the fit axis, the failure to avoid is rigid title matching. Exact-string title filters manufacture false scarcity. In Refolk's index, an exact "VP Revenue Operations" query at VP seniority returned 0, while Director-level Revenue Operations in the US returned 389 real people. The same population hides behind variant titles, so filter on seniority plus keyword families, never one literal string.

Supply concentration also belongs in scoring, because it reshapes territory before a single email sends. US Director-level RevOps outnumber the UK 3.0x, which changes how you route the same job title across two geographies.

A US conference list will over-index New York, Austin, and San Francisco; a UK list will over-index London. Knowing that before you sequence lets you set realistic per-territory targets instead of expecting UK volume to match US volume.

The procedure, with a count at every stage

Run the export through nine stages in order, logging the row count and drop at each so the final number is defensible. Owners and rough times are given so you can staff it against the 24-hour clock.

Log every stage in one place, because the counts are your defense. When someone asks why 500 scans became a 108-row queue, you point to the 50% enrichment resolution, not a vague claim about quality.

Compliance is a routing rule, not a footer

A row is safe to send when its source is documented, the contact is relevant to their role, a postal address and one-click opt-out are present, headers are not deceptive, and no prior opt-out exists. Because the legal standard differs by recipient geography, the same queue needs two send policies.

Conference lists sit in a defensible zone. Industry conference attendee lists whose data sharing is covered in the event's privacy policy are legitimate sources for building prospect lists, because you are not obtaining data through deception or from private sources. For EU and UK recipients, GDPR usually leans on legitimate interest: you can send without prior consent if the outreach is relevant to the recipient's professional role, you document a genuine business purpose, you are transparent about data use, and you offer easy opt-out.

The split that matters operationally is timing. GDPR requires immediate opt-out processing and a documented legal basis; CAN-SPAM in the US allows emailing anyone unless they opt out and lets you honor opt-outs within 10 business days. So segment by recipient geography and apply the stricter standard to EU and UK rows. Treating a US opt-out-only policy as sufficient for an EU contact is a false-safe, and the downside is asymmetric: CAN-SPAM fines reach up to $51,744 per non-compliant email, while GDPR fines reach up to €20 million or 4% of global revenue.

recipient_geo: US | EU/UK        # sets which opt-out standard applies
source_documented: yes/no        # event privacy policy covers data sharing
role_relevant: yes/no            # outreach matches their professional role
postal_address_present: yes/no   # required for both regimes
opt_out_present: yes/no          # one-click; immediate for EU/UK, 10 days US
prior_optout_checked: yes/no     # suppress if previously unsubscribed
headers_truthful: yes/no         # no deceptive from/subject lines
safe_to_send: yes/no             # true only if all above pass

How this goes wrong

The failure modes below are the highest-value part of this guide, because each produces a queue that looks fine and is not. For each, I give the false positive it creates and the check that catches it.

• Column-shift corruption. Raw badge CSVs put email in the last-name field and company in the title field. The false positive is a "clean" import that scores personal provinces as companies. Check by eyeballing ten random rows against the QR source before enriching.
• Match-rate inflation. A vendor quoting 90% is counting matches, not deliverable ones. The false positive is a full-looking email column that hard-bounces. Check by verifying deliverability post-append and tracking bounce rate.
• Wrong-person append. Loose matching on name plus city hands you a different person's email. Check by requiring company-domain agreement, not just name similarity.
• Stale employer. Two-year average tenure means the badge company is often last quarter's job. The false positive is outreach to a champion who already left. Check by running job-change flags and re-resolving before you sequence.
• Registrant-only mistaken for intent. A registered no-show scores like a booth visitor if the model only counts "attended event." Check by separating registrant-only from badge-scanned and session-attended before scoring.
• Rigid title filter drops real fit. Exact-string title filters return zero where variant titles exist; the "VP Revenue Operations" query returning 0 is the proof. Check by filtering on seniority plus keyword families, not one literal string.
• Compliance false-safe. A US-style opt-out-only send to an EU contact breaches GDPR. Check by segmenting on recipient geography and applying the stricter standard.

The pattern across all seven: a stage that silently passes bad rows is worse than one that visibly drops them. Your logged counts are the instrument that makes silent failures visible.

Before you hit send

Run this checklist against the queue, not against the process. Every item is something you can verify by looking at the final file.

Keeping the queue current between shows

The queue is not a one-time artifact; it decays on two clocks and both are re-checkable. The reply-rate clock decays in hours, and the employer clock decays over roughly two years of average tenure. Build the re-check into your cadence rather than treating the export as finished.

For the reply clock, dispatch the top segment same-day and re-rank the remainder every morning of the follow-up window, because a row that was 48 hours cold yesterday is worse today. For the employer clock, re-resolve stale companies before every send campaign and treat a champion's move as a fresh trigger at the new account rather than a lost row. A plain-English search for the people who moved off your list, at their new employer, turns the weakest part of the export into new pipeline.

Finally, keep your stage counts across shows so you can tell whether a thin queue is a data problem or a targeting problem. If enrichment resolution keeps landing at the bottom of the 25 to 60% band, add a second provider before you touch the ICP filter, because the waterfall lift of 20 to 40% moves the leak you actually have.