Lead Speed Plan
Where leads wait longest, from data intake to closing, and how we measure it. Click a stage, press "Next event" to watch one lead move.
The system writes the dates
Every time a lead moves to a new stage, an automation records the exit time and entry time. Nobody types a date.
One row per lead per stage
Rows are kept forever and never overwritten. Failed leads, send-backs and repeat cycles are all kept.
The slowest stage shows itself
Average days and "waiting now" per stage and per group (P1-P6, P7, P8-P12, grade A/B/C). The slowest stage is the bottleneck.
Goal: find where leads wait longest, from data coming in to closing, with measurement and not opinion. Fix the slowest stage first, then drill down week by week.
Words used on this page (click to hide)
One step of the journey, like Research or Deal Review. Nakul calls them phases.
The date and time written when a lead enters or leaves a stage.
One line in the log: this lead, this stage, in at this time, out at that time.
The time a lead sits between two stages, finished with one and not yet picked up by the next.
Leads that have a priority but no researcher yet. They are waiting in a pile.
A lead a researcher has been given to work on.
The priority level of a lead. P0 (foreclosure, urgent) comes first, then P1, P2 and so on.
The score a researcher gives a lead when research ends.
The number of times a property starts over, for example for the next heir.
The stage where leads wait the longest. That is what we are looking for.
The journey of a lead
Click any stage to see who owns it, when its clock starts and stops, how a lead can fail, and what is missing.
P0 is a special lane: foreclosure leads
P0 means a foreclosure with a deadline: the auction. It is the most urgent tier, so speed matters most here. We can already measure it.- 97 leads carry P0 today. All are Travis County, "Mortgage Foreclosure". They come from the separate foreclosure scraper and are tagged P0 by hand when pushed. The sale date field exists and is filled.
- The deadline is the clock. For P0 we also measure "days left before the auction", not only days in a stage.
- First look: all 97 P0 leads have been researched, in about a day each (median 1.1 days). 94 of 97 finished on or before the auction date.
- Watch now: no P0 lead is open. 17 still have an auction date today or later. Only 27 of the 97 passed research: 46 failed the smell test and 24 failed research.
- One gap: the SOP tier formula cannot produce P0 (the foreclosure flag is off in every county script), so P0 depends on someone tagging it. Fix: turn the flag on so P0 is automatic.
Do the 28 draft steps fit? Yes, all 28.
Checked by script: every draft node maps to one of Nakul's 9 phases, or to the one loop.
| Phase | Draft nodes | Count |
|---|---|---|
| 1. Data Cleansing & Filtering | 1.1-1.4, 2.1-2.4 | 8 |
| 2. Research & Genealogy | 3.1-3.4 | 4 |
| 3. Deal Review | 4.1, 4.2 | 2 |
| 4. Lead Manager | 4.3, 4.4 | 2 |
| 5. Relationship Management | 5.1-5.4 | 4 |
| 6. Acquisitions Discovery | 6.1 | 1 |
| 7. Due Diligence | 6.2 | 1 |
| 8. Offer & Negotiation | 6.3, 6.4 | 2 |
| 9. Transactions & Closing | 7.1-7.3 | 3 |
| Loop: next heir | 7.4 | 1 |
| Total | 8+4+2+2+4+1+1+2+3+1 | 28 |
Data assignment. Priorities wait in inventory until a researcher takes them. No draft node covers this, yet Nakul named it as the key delay.
Fix: add it as its own stage between phase 1 and 2 (stage "1b" above), stamped by a new "Date Assigned to Researcher".
• Data pulling has no phase of its own, so it becomes the start of phase 1.
• Phases 6 and 7 have no dates at all. Add stamps, or merge them until sales confirms they are separate.
How the automation works
Press the green button and watch one lead move. The log fills in by itself. Try the other two cases too.
| Stage | In | Out | Days | Result | Waited before |
|---|
An example lead, not a real one. "Waited before" is the gap since the lead left the stage before.
Edge cases and how each is handled
The situations that break most tracking systems, and what stops them breaking this one.
Who has to do what
The most common worry is "does this give me extra work?" The plan says no.
| Who | What changes for them |
|---|---|
| Researchers | Nothing new. Update status as today. The system stamps the time. |
| Deal Review and Lead Manager | Nothing new. Same status changes as today. |
| Relationship Manager and Acquisitions | Nothing new. Move cards in GHL as today. We pick the dates up. |
| Nakul | Gets a weekly report: slowest stage, and who is waiting longest. |
| Harsh | Builds it and checks every day that no stamp is missing. |
• Not a grade for any person. It measures the process.
• Not extra data entry for the team.
• Not a change to how anyone works yet. First we measure. Then we decide what to fix.
First numbers, from live data
Average days in each phase, over leads that have both dates. Whole days.
| Phase | P1-P6 | P7 | P8-P12 | A | B | C |
|---|---|---|---|---|---|---|
| Research | 11 | 13 | 17 | 14 | 15 | 14 |
| Deal Review | 30 | 6 | 8 | 9 | 27 | 4 |
| After Deal Review | Fewer than 5 leads with both dates. Not measurable yet. | |||||
Deal Review counts only 30, 11 and 22 leads: a lead to follow, not a conclusion. A / B / C is the researcher's lead grade (confirmed). P0 (foreclosure): 97 leads researched in about a day each, 94 of 97 on or before the auction.
Waiting now: open leads and their age. Averages of finished leads hide the stuck ones.
Wait between stages: where the backlog builds.
Leads counted: so nobody trusts a number built on 2 leads.
Failed vs passed: in research, 3,100 leads failed and 1,128 passed.
Rollout: one stage at a time, approval first
Nothing is built before Nakul says yes.
Every date field: which stage, who or what sets it, how often filled. Read-only.
Nakul approves the stages and the log design.
Log and automations for P0 (97 leads exist), then P1: list management, assignment, research, Deal Review.
GHL webhook for Lead Manager through Closing.
Weekly matrix from the log. Daily health check for gaps.
Review with Nakul. Drill into the slowest stage.
Questions you will be asked
Click a question to see the answer.
Why not just ask people how long things take?
People remember differently and count differently. A stamp is exact and the same for everyone, so we can compare stages fairly.
Does this give my team extra work?
No. Nobody types a date. The system writes it when a status changes. People keep working the way they do today.
Is this a performance review of people?
No. It measures the process, not people. If a stage is slow, we look at the stage and its steps first, like the Tesla example: the slow step turned out to be the procedure itself.
What if the numbers look bad?
That is the point. Bad numbers show where to improve. We pick the slowest stage and fix it, then measure again.
What about leads that already went through?
Where dates were stored, we use them: research and Deal Review have history. Dates that were never stored cannot be rebuilt, so the other stages start counting from go-live.
When do we see results?
Research and Deal Review show now, with a small count in Deal Review. The new stamps start when we go live, and Nakul wants that next week. The first weekly report follows, and a trustworthy average needs a few weeks of data.
Do we need new software?
No new tools. It uses Airtable automations and one new log table. The sales side uses the GHL connection we already run.
Why is P0 different?
P0 is foreclosure, so there is a hard deadline: the auction. We measure days left before the sale as well as days in each stage. Today 94 Travis leads carry P0. The tier formula cannot compute P0 yet, so someone tags it by hand. Making it automatic is one of the decisions below.
How do we know the numbers are right?
Every average shows how many leads it counts. A daily check looks for missing or odd dates and raises an alert. Bad dates are flagged and left out until fixed.
Decisions needed from Nakul
- Approve Nakul's 9 phases plus the added stage "assigned to a researcher".
- P0 is foreclosure (97 leads, sale dates exist). Approve turning the foreclosure flag on so the tier formula produces P0 automatically, and measuring P0 first?
- Calendar days or business days?
- OK to add: Date Assigned to Researcher, sent to skip tracing, reconciled for Airtable, and an Exit Reason?
Build status
What is done and what is checked, as of 27 Oct 2026.
| Piece | Status | Evidence |
|---|---|---|
| Stamp rules (stages, outcomes, cycles, edge cases) | Done | 19 tests pass; 5 deliberate breakages all caught |
| Run on a copy of today's live data | Done | 1,635 lines: 910 waiting, 236 in research, 489 in Deal Review. A second sweep changes nothing. Two runs at once are repaired |
| Log table in Airtable (Phase Time Log, 15 fields) | Done | Created, empty |
| Automations: hourly sweep + lead changed | Built, OFF | Waiting to be tested once and switched on |
| Nightly check against MASTER | Done | Read-only. Says NOT STARTED until the log has lines |
| Weekly report built from the log | Done | 7 tests pass. No data yet |
| Existing leads at go-live | Handled | Marked baseline and low trust, never averaged in |
| Lead Manager to Closing (GHL feed) | Not started | Needs an approved way to write |
| First real review with Nakul | Target Nov 16 | Needs the automations on by about Nov 4 |