mobile-appender

Mobile Appender

Overview

This skill produces a calling list: one row per person with a mobile number in E.164, its line type, a suppression check, a time zone and a dial_status that says whether a rep dials it now or a person reviews it first. A short report shows the measured hit rate, the cost, and a funnel from inputs to dial-ready numbers.

It prevents three expensive mistakes:

1. Paying twice for one person. Mobile Finder bills $0.25 for every number it finds; send a LinkedIn URL and an email for the same person and two hits cost $0.50 for one human.
2. Budgeting blind. ScraperCity does not publish a hit rate, so a list run without a pilot can cost anywhere from a few dollars to the full ceiling of $0.25 per input.
3. Loading numbers labeled "mobile" that are office lines, VoIP numbers or shared reception numbers, so reps burn dials on the wrong line.

When to Use

• "Find mobile numbers for these 300 LinkedIn profiles so my SDRs can dial them."
• "Append cell phones to this list of emails."
• "We need direct dials for the call blitz next week."
• "Add mobiles to our open opportunities."

Use a different skill when:

Prerequisites

Authoritative Foundations

Details and the routing table: references/framework-notes.md.

Workflow

1. Scope the job

Before any paid call, record in the report's Call Plan table: the dialing method, the suppression list used and its row count, the countries in the list, and whether line-type lookups will run on every number. Any automated or power dialer that drops calls on non-mobile lines needs a lookup on every number; a click-to-call team can skip it and let reps mark wrong lines.

2. Normalize and dedupe inputs

• One row per person, one input per person per run. If a person has a LinkedIn URL and a work email, the first run gets the URL; a second run sends the work email only for people whose URL found nothing. Misses are free, so the second pass adds coverage and cannot bill anyone twice.
• LinkedIn URLs: lowercase, https://www.linkedin.com/in/<slug>, strip query strings, locale suffixes and trailing slashes. Drop company pages (/company/), Sales Navigator links (/sales/) and post links; they do not identify a person.
• Emails: trim, lowercase, reject syntax failures. Work emails only; Mobile Finder's catalog description names social profile URLs or work email addresses.
• Remove people already holding a mobile in the source, and anyone on the suppression list.

Details: references/phone-normalization-and-line-type.md.

3. Pilot 50 and measure the hit rate

Run find_mobiles on 50 random inputs. Cost is at most 50 x $0.25 = $12.50. Hit rate = numbers found / 50. ScraperCity does not publish a hit rate, so this pilot is the planning number. At 50 inputs the standard error near a 50% rate is about 7 points (NIST/SEMATECH formula), so set the low and high scenarios at least 14 points either side. The method is in references/cost-and-hit-rate-model.md.

4. Estimate and confirm

expected cost = remaining inputs x hit rate x $0.25
ceiling       = remaining inputs x $0.25

Show low, expected, high and ceiling scenarios. Run check_wallet. Wait for a go-ahead on the ceiling before the main run.

5. Run the append

Phone ladder, cheapest first: the mobile already in the source; on Growth and Scale, a Lead Database people call with hasPhone: true for a direct or company line at $0 wallet (written to phone, never to mobile); then find_mobiles with URLs; then the work-email second pass for URL misses.

Send many inputs per run, never one run per row. Split the list into batches when the user wants to stop early: after each batch, compare its hit rate with the pilot's low scenario, and if it falls below, stop and look at the inputs (old export, wrong country, company pages). Poll check_run_status, report, and check again later instead of looping (the CLI poll command defaults to every 15 seconds); then download_results. Keep the runId per batch in the run log.

6. Post-process every number

1. Normalize to E.164. US and Canada: +1 and 10 digits, area code and exchange not starting with 0 or 1.
2. Flag shared numbers: one number on two or more people is an office line or a reception desk. Hold.
3. Line type: a number's prefix does not prove a mobile once it has been ported. When the plan calls for lookups, run each number through a carrier lookup the user holds (for example Twilio Lookup Line Type Intelligence). Keep line_type.
4. Suppression: match every number and email against the one suppression list. Record suppression_checked_at.
5. Time zone: from the person's location, not the area code (mobiles travel). Store tz so the dialer places each call inside the window the team sets in the prospect's local time.

7. Assign dial status and deliver

Deliver the CSV and the report from templates/output-template.md; run scripts/check-output.py.

ScraperCity Implementation

Tools and prices (from references/scrapercity-catalog.md)

Calls

{"tool": "find_mobiles", "args": {"inputs": [
  "https://www.linkedin.com/in/priya-natarajan",
  "https://www.linkedin.com/in/marcus-bell-it",
  "[email protected]"
]}}

CLI: npx scrapercity mobile-find https://www.linkedin.com/in/priya-natarajan [email protected], then npx scrapercity poll <runId> and npx scrapercity download <runId>. REST: POST https://app.scrapercity.com/api/v1/scrape/mobile-finder with {"inputs": [...]}, then GET /api/v1/scrape/status/{runId} and GET /api/downloads/{runId}.

Mobile Finder is also reachable through scrape_website_finder with autoFindMobiles and through find_emails options in the CLI, but the catalog prices only the standalone tool. Use find_mobiles directly so every charge is $0.25 per number found.

Worked cost estimate

340 rows received; after dedupe and dropping 6 company or Sales Navigator links, 326 inputs. Pilot 50: 18 found (36%).

Ask for a go-ahead on the $69.00 ceiling (the expected spend is $24.75). Cost per dial-ready number is total spend divided by the numbers that pass the shared-number, line-type and suppression checks; report it from the run's own counts. Carrier lookups are billed by the lookup provider, not ScraperCity.

Errors

Output Format

1. mobiles-[segment]-[date].csv: one row per person with first_name, last_name, company_name, input_used, mobile (E.164), line_type, suppressed, suppression_checked_at, tz, dial_status, source = mobile_finder, pulled_at.
2. mobiles-[segment]-[date]-hold.csv: every hold_* row with its reason.
3. A report from templates/output-template.md: input summary, call plan, cost estimate with the hit-rate scenarios, append funnel, line type and suppression results, dial rules, output schema, sample rows, next steps.

• What the reader receives is the deliverable only, in product names (Lead Database, Email Validator, Email Finder, Mobile Finder, Website Finder): no tool ids, JSON tool calls, run IDs, skill names, file paths, reference section numbers or runbook IDs; a downstream step is written out as the checks or commands the team runs, never handed to another skill by name; no tables of checks that were not run (one "Not checked:" line instead); and no repeated identical rows where one template plus a list will do. scripts/check-output.py fails a report that carries a tool id, skill name or path.

Quality Check

• The call plan was recorded before the pilot.
• One input per person per run; the email pass held only people whose URL missed.
• The pilot hit rate and the low, expected, high and ceiling scenarios appear, and the user approved the ceiling.
• Every delivered mobile matches E.164 (+ then 8 to 15 digits, first digit not 0).
• No callable row is on the suppression list or shared by two people.
• The funnel adds up from inputs to dial-ready, and the dial rules sum to the numbers found.
• scripts/check-output.py report.md prints PASS.

Common Pitfalls

Execution Artifacts

Related Skills

• cold-calling for scripts, cadence and call windows.
• crm-enricher to write the numbers back to the CRM with audit columns.
• multi-channel-outreach to combine calls with email and LinkedIn touches.
• decision-maker-finder to pick whom to call at each account.
• skip-tracer for consumer or property-owner phones.
• email-list-cleaner for the email side of the same list.
• scrapercity-mcp for MCP setup.