Field Collections • Executive business case

Scale field collections selectively, with a gated hybrid model and a July 1 CDMX benchmark.

This memo combines the pilots, the delta workbook, the next-step sizing sheet, the raw Kobra operations feed, and Snowflake cure analytics into one executive story. The structure stays decision-first: what to approve, what the pilots proved, where the economics work, and how to expand only if the first gates are hit.

Decision in one sentence

Approve a staged hybrid rollout: launch CDMX 31-60 as a matched internal-versus-Kobra benchmark on July 1, continue Kobra where the vendor case is already proven, and do not approve broad 91-120 expansion until the next benchmark cycle is complete.

3 / 3 Pilots showed positive treatment cure uplift versus baseline.

MXN 300k Monthly payroll for one internal 10+1 team.

10-12/day Internal utilization range that makes CDMX economically viable.

4-7 days Most attractive revisit window in the raw Pilot 3 funnel.

1. Executive recommendation

Recommendation: staged hybrid, with internal bias only where density is proven.

The evidence still does not support a simple binary answer. In-house is not yet proven as a broad operating model; Kobra is not attractive as a broad answer at current pricing. The strongest path remains a hybrid, but the near-term design should become sharper: a CDMX 31-60 head-to-head benchmark, a narrow Jalisco 91-120 Kobra Pilot 4, and selective continuation in Chihuahua 91-120.

Go now: CDMX 31-60 benchmark on July 1 Go now: Chihuahua 91-120 with Kobra Test now: Jalisco 91-120 Kobra core Next internal replication: Jalisco 31-60 Expand later: selective Estado de México 31-60 Do not approve: broad CDMX 91-120 Do not approve: broad 91-120 Kobra scale

Key decision message: field collections appears to be a viable incremental channel, but only selectively. The right approval is a gated rollout with one matched CDMX benchmark, one narrow vendor expansion question, and explicit stop-losses on utilization, conduct, and incrementality.

Why this recommendation still holds

Positive pilot evidence exists, but it is uneven by geography and bucket.
Broad Kobra expansion destroyed economics in several states despite real cure uplift.
Internal economics look attractive only if productivity stays near 10-12 visits per day and route quality is high.
Jalisco is the strongest repeat-positive geography across pilots and projections.
Chihuahua is the strongest current vendor-backed 91-120 case.
CDMX has enough density to justify a benchmark, but not enough evidence to justify broad scale in 91-120.

2. Pilot results and proof of concept

The pilots proved that field visits can move cure. They did not prove that scale should be broad.

The proof is shown in two layers. First comes the gross field view: field recovery, field spend, and the cost of one collected peso. Then comes the net decision view: treatment recovery minus calls & notifications baseline minus field spend. Cure is shown as the share of clients with payment > 0 during the pilot window from Snowflake.

Pilot / bucket	Field recovery	Field spend	Gross field result	Price of 1 collected peso	Recovery from field-treatment group	Recovery from calls & notifications baseline	Treatment cure rate	Baseline cure rate	Cure lift	Final delta	Decision read
Pilot 1 91-120	MXN 357,609.16	MXN 220,000.00	MXN 137,609.16	MXN 0.62	MXN 331,964	MXN 115,559	3.73%	1.57%	+2.16 pp	MXN -3,595	The channel worked, but economics were only roughly breakeven after baseline subtraction.
Pilot 2 91-120	MXN 541,143.74	MXN 400,000.00	MXN 141,143.74	MXN 0.74	MXN 558,693	MXN 300,495	2.14%	1.05%	+1.09 pp	MXN -141,802	Cure uplift remained positive, but broad late-stage Kobra scale is not supported.
Pilot 3 31-60	MXN 580,145.42	MXN 326,012.00	MXN 254,133.42	MXN 0.56	MXN 1,037,916	MXN 731,912	13.54%	9.87%	+3.67 pp	MXN 38,400	Best combined proof: strongest cure lift and the clearest case for selective earlier-bucket rollout.

Source note: Field recovery and field spend / consumo now use the Kobra closeout file as the source of truth. Calls & notifications baseline recovery and final delta still come from the attached delta workbook, which is why the gross and net views should be read as two linked but distinct layers.

All regions, with cure rates and full delta construction

Pilot	Region	Field recovery	Field spend	Gross field result	Price of 1 collected peso	Baseline recovery	Treatment cure rate	Baseline cure rate	Cure lift	Final delta	Read
Pilot 1	Ciudad de México	MXN 175,368.56	MXN 102,976.00	MXN 72,392.56	MXN 0.59	MXN 13,667	3.85%	0.98%	+2.87 pp	MXN 15,456	Positive, but still much weaker than Jalisco on cost efficiency.
Pilot 1	Estado de México	MXN 45,084.17	MXN 43,359.00	MXN 1,725.17	MXN 0.96	MXN 14,417	1.77%	2.75%	-0.98 pp	MXN -26,942	Gross barely covers spend and net remains unattractive.
Pilot 1	Jalisco	MXN 109,730.00	MXN 42,822.00	MXN 66,908.00	MXN 0.39	MXN 8,508	5.19%	1.56%	+3.63 pp	MXN 46,850	Best Pilot 1 region by both delta and gross efficiency.
Pilot 1	Nuevo León	MXN 27,426.43	MXN 30,843.00	MXN -3,416.57	MXN 1.12	MXN 4,142	3.64%	1.89%	+1.75 pp	MXN -10,671	Cure moved, but both gross and net stayed unattractive.
Pilot 2	Chihuahua	MXN 70,397.60	MXN 37,157.33	MXN 33,240.27	MXN 0.53	MXN 9,098	3.02%	0.76%	+2.26 pp	MXN 51,890	Cleanest current vendor-backed case.
Pilot 2	Ciudad de México	MXN 192,443.97	MXN 170,517.67	MXN 21,926.30	MXN 0.89	MXN 46,523	1.52%	1.01%	+0.51 pp	MXN -80,688	The cure lift exists, but cost and baseline drag wipe it out.
Pilot 2	Estado de México	MXN 5,406.20	MXN 5,662.00	MXN -255.80	MXN 1.05	MXN 0	4.88%	0.00%	+4.88 pp	MXN -256	Tiny sample, not decision-grade for scale.
Pilot 2	Jalisco	MXN 138,243.79	MXN 82,380.00	MXN 55,863.79	MXN 0.60	MXN 27,615	2.93%	1.00%	+1.93 pp	MXN 31,300	Still positive, but not as clean as Chihuahua on vendor economics.
Pilot 2	Nuevo León	MXN 86,628.00	MXN 61,388.00	MXN 25,240.00	MXN 0.71	MXN 19,040	1.81%	0.45%	+1.36 pp	MXN -28,608	Gross-positive, but net remains unattractive.
Pilot 2	Puebla	MXN 28,650.00	MXN 28,591.00	MXN 59.00	MXN 1.00	MXN 22,093	2.50%	1.96%	+0.54 pp	MXN -16,466	Gross-positive only; baseline absorption is too large.
Pilot 2	Querétaro	MXN 19,374.18	MXN 13,857.00	MXN 5,517.18	MXN 0.72	MXN 5,907	2.17%	3.23%	-1.06 pp	MXN -389	Weak actual proof; keep only as a suggestion, not as a proven case.
Pilot 2	Chiapas	MXN 0.00	MXN 149.00	MXN -149.00	n/a	Not attached	n/a	n/a	n/a	Not in delta workbook	Too small to matter and not part of the decision set.
Pilot 2	Morelos	MXN 0.00	MXN 149.00	MXN -149.00	n/a	Not attached	n/a	n/a	n/a	Not in delta workbook	Too small to matter and not part of the decision set.
Pilot 2	Quintana Roo	MXN 0.00	MXN 149.00	MXN -149.00	n/a	Not attached	n/a	n/a	n/a	Not in delta workbook	Too small to matter and not part of the decision set.
Pilot 3	Ciudad de México	MXN 250,975.37	MXN 137,676.00	MXN 113,299.37	MXN 0.55	MXN 283,739	13.77%	9.81%	+3.96 pp	MXN 420	Right answer for a narrow proof, not for broad citywide scale.
Pilot 3	Estado de México	MXN 160,853.54	MXN 104,449.00	MXN 56,404.54	MXN 0.65	MXN 281,935	12.75%	9.78%	+2.97 pp	MXN -79,996	Operationally promising, but the baseline is too heavy for wave 1.
Pilot 3	Jalisco	MXN 168,316.51	MXN 83,887.00	MXN 84,429.51	MXN 0.50	MXN 125,260	14.26%	10.12%	+4.14 pp	MXN 38,413	Best region to replicate after the CDMX proof.
Pilot 3	Sin información de domicilio	MXN 0.00	MXN 0.00	MXN 0.00	n/a	Not attached	0.00%	n/a	n/a	Not in delta workbook	Included only for completeness; not a launchable geography.

Reading rule: price of 1 collected peso = field spend / field recovery. Cure uses cure_report, not payment > 0. For Pilot 3 the threshold is order_dpd_mature <= 60; for Pilots 1-2 it is order_dpd_mature <= 120. Field recovery and field spend come from Kobra's closeout file; baseline recovery and final delta come from the delta workbook where available.

Operational funnel from the raw Kobra visit feed

The raw Kobra workbook shows what happens on the ground between visit, localization, access, direct debtor contact, and PTP. These metrics are per visit, not per client. Reported contact often includes third-party contact, so direct debtor contact is the stricter field-quality measure.

Pilot	Region	Visits	Localized	Access	Reported contact	Direct debtor contact	PTP	Revisit requested
Pilot 1	All regions	1,147	87.62%	59.37%	87.62%	6.10%	2.27%	23.54%
Pilot 1	Ciudad de México	502	86.65%	51.59%	93.23%	7.57%	2.59%	32.87%
Pilot 1	Estado de México	258	81.78%	58.14%	80.23%	5.04%	1.94%	16.67%
Pilot 1	Jalisco	234	90.60%	66.67%	85.04%	4.27%	1.28%	7.26%
Pilot 1	Nuevo León	153	96.08%	75.82%	85.62%	5.88%	3.27%	29.41%
Pilot 2	All regions	3,218	87.07%	60.04%	90.27%	4.88%	1.46%	21.44%
Pilot 2	Chihuahua	256	92.58%	73.44%	80.08%	3.12%	0.39%	80.47%
Pilot 2	Ciudad de México	1,535	82.87%	47.95%	91.53%	4.69%	1.37%	26.25%
Pilot 2	Estado de México	38	73.68%	57.89%	89.47%	7.89%	5.26%	13.16%
Pilot 2	Jalisco	620	91.13%	72.42%	87.74%	5.48%	2.26%	3.55%
Pilot 2	Nuevo León	412	93.69%	75.24%	89.81%	7.28%	1.21%	7.77%
Pilot 2	Puebla	261	87.74%	65.13%	96.55%	0.38%	0.00%	8.43%
Pilot 2	Querétaro	93	89.25%	60.22%	98.92%	9.68%	4.30%	0.00%
Pilot 3	All regions	2,171	85.67%	63.89%	93.09%	5.62%	1.66%	25.38%
Pilot 3	Ciudad de México	911	88.14%	61.03%	95.50%	5.60%	1.65%	49.40%
Pilot 3	Estado de México	700	77.14%	62.29%	87.14%	4.29%	1.43%	12.43%
Pilot 3	Jalisco	560	92.32%	70.54%	96.61%	7.32%	1.96%	2.50%

Pilot	Visit timing bucket	Visits	Reported contact	Direct debtor contact	PTP	Read
Pilot 1	1st visit	1,032	86.72%	6.01%	2.42%	Baseline for the first wave.
Pilot 1	Revisit 8d+	98	94.90%	8.16%	1.02%	Some direct-contact improvement, but weaker PTP signal.
Pilot 2	1st visit	3,019	91.09%	4.74%	1.36%	Large-scale baseline.
Pilot 2	Revisit 8d+	172	79.07%	6.40%	3.49%	Delayed revisits still improved conversion quality.
Pilot 3	1st visit	1,627	91.09%	3.26%	1.54%	Starting point in the earlier bucket.
Pilot 3	Revisit 4-7d	127	100.00%	21.26%	4.72%	Strongest revisit window in the raw feed; supports fast revisit discipline.
Pilot 3	Revisit 8d+	406	98.77%	10.34%	1.23%	Still better than first visit on direct contact, but weaker than fast revisits.

Method for the raw funnel: the Kobra operations workbook was grouped into three pilot windows by visit-date clusters. Funnel metrics are per visit. Reported contact is vendor-recorded and appears to include third-party interactions; direct debtor contact is the more conservative field-quality measure.

Strong enough to support scaling decisions

Jalisco is still the strongest repeat-positive geography. Chihuahua is the cleanest vendor-backed 91-120 case. Broad 91-120 rollout through Kobra is not supported. Borough and city packs are better decision units than full states.

Strong enough to support operating changes

Revisits still matter. The raw vendor feed suggests the best revisit window is fast follow-up, especially in Pilot 3 where 4-7 day revisits showed meaningfully higher direct debtor contact and PTP than first visits.

Important caution

Not every positive funnel movement becomes positive incremental cash. The CRO decision should still anchor on matched-cohort delta and cure lift, not on raw vendor operational metrics alone.

3. Economics and operating model

Gross economics tell us where to look. Final delta tells us whether to scale.

The current materials contain two different economic views. The pilot readout lets us calculate final delta because it includes treatment-group recovery, remote-only recovery baseline, and field spend. The next_steps sizing sheet gives projected field recovery, projected spend, projected baseline recovery, and expected delta by cluster, which is enough to compare suggested launch candidates explicitly.

Model	Cost logic	Where it fits	What can break it
In-house 10 collectors + 1 supervisor	MXN 300k monthly team cost. Cost per visit = 300,000 / (10 × 22 × visits/day).	Dense, repeatable 31-60 pools with tight routing.	Productivity below 10/day, thin inflow, or omitted all-in costs such as transport, devices, insurance, and shrinkage.
Kobra	Working benchmark price in the materials: MXN 185 / visit.	Proven 91-120 pockets and matched benchmarks.	Vendor price reset, diluted geography, or weak address quality.
Hybrid	Internal fixed-cost cell for dense pools plus variable vendor cost where external flexibility is still useful.	Best fit to the evidence.	Governance complexity if one economic dashboard is not enforced.

CDMX internal sensitivity, shown with all parts of the delta

This is the clearest monthly internal proof in the materials because it explicitly shows all three parts: projected field recovery, remote-only baseline recovery, and field spend.

Visits / day	Monthly visits	Cost / visit	Projected field recovery	Projected calls & notifications baseline recovery	Field spend	Final delta
8	1,760	MXN 170.45	MXN 398,606	MXN 158,552	MXN 300,000	MXN -59,946
10	2,200	MXN 136.36	MXN 498,257	MXN 198,189	MXN 300,000	MXN 68
12	2,640	MXN 113.64	MXN 597,909	MXN 237,827	MXN 300,000	MXN 60,081

Risk note: the internal base case still excludes several cost lines. It is useful for directional comparison, but it is not yet a final investment-grade cost model.

Internal team cost breakdown

The internal base case uses one field cell with 10 field agents + 1 supervisor. The current model reflects payroll and labor burden only.

Cost component	Formula	Monthly amount	What it means
One field agent	(MXN 10,000 fixed + MXN 12,000 variable) × 1.20 burden	MXN 26,400	Per-agent monthly cost including the labor burden used in the current model.
Ten field agents	10 × MXN 26,400	MXN 264,000	Total monthly cost for the collector team.
One supervisor	MXN 30,000 compensation × 1.20 burden	MXN 36,000	Monthly supervisor cost in the base case.
Total internal 10+1 team	MXN 264,000 + MXN 36,000	MXN 300,000	Base monthly payroll cost used in the internal economics.

If internal only matches Kobra's CDMX 31-60 performance, it is still economically better to do the work ourselves

This benchmark keeps the same recovery performance and changes only the cost model. That is why the July 1 CDMX design should be a matched benchmark: internal does not need to beat Kobra on effectiveness to beat it economically.

Visits / day	Monthly visits	Field recovery if performance only matches Kobra benchmark	Calls & notifications baseline recovery	Internal spend	Kobra spend @ MXN 185 / visit	Internal delta	Kobra delta	Internal advantage
10	2,200	MXN 498,257	MXN 198,189	MXN 300,000	MXN 407,000	MXN 68	MXN -106,932	MXN 107,000
12	2,640	MXN 597,909	MXN 237,827	MXN 300,000	MXN 488,400	MXN 60,081	MXN -128,318	MXN 188,400

Debt and credit-limit insight: bigger is not always better

Snowflake confirms the user hypothesis, but in a nuanced way. Higher debt and higher limit usually bring higher absolute recovery, yet the strongest field-vs-control uplift is usually concentrated in the mid-to-upper bands, not only at the very top.

Pilot	Best total-debt band	Best credit-limit band	Read for targeting
Pilot 1 91-120	20k-30k +MXN 300 avg recovery +5.12 pp cure lift	10k-20k +MXN 334 avg recovery +3.00 pp cure lift	Do not frame the answer as “highest limit wins.” Mid-high balances were stronger.
Pilot 2 91-120	30k+ +MXN 269 avg recovery +2.26 pp cure lift	20k-30k +MXN 137 avg recovery +1.78 pp cure lift	Late bucket is less stable, but it still supports focusing above the floor rather than staying broad.
Pilot 3 31-60	20k-30k +MXN 502 avg recovery +2.12 pp cure lift	20k-30k +MXN 512 avg recovery +3.34 pp cure lift	Best evidence to prioritize 10k-30k debt and 10k-30k credit limit, not only 30k+.

Targeting implication: keep the floor at MXN 5k+ total debt, but prioritize 10k-30k debt and 10k-30k credit limit first. Higher absolute pesos in 30k+ do not consistently translate into the best field uplift.

What makes the model unattractive

The economics are sensitive in a few predictable ways. If these move the wrong way, the page should be read as a warning, not as a green light.

Utilization risk

The internal model only looks strong when one team stays near 10-12 visits/day. Below that, the fixed payroll cell becomes hard to justify.

Baseline absorption risk

The biggest destroyer of delta is not always weak field recovery. Often it is that the remote channel would have recovered too much anyway, leaving too little truly incremental cash.

Address and route quality risk

Poor address quality or fragmented geography raises cost per productive visit very quickly. This is why address enrichment and prioritization should be prerequisites.

Potential internal launch candidates

These rows come from the attached next_steps sheet. Team-months are shown as first-pass workload months at 10 visits/day, assuming one field team can cover 2,200 visits per month.

Cluster	Status / type	Current clients	Current total debt	Benchmark cure rate used	Expected field recovery	Expected calls & notifications recovery	Expected field spend	Expected delta	Price of 1 collected peso	Team-months @ 10/day	Read
CDMX carve-out	Yellow • pilot-backed	4,739	~MXN 30.55M	13.77%	~MXN 2.87M	~MXN 1.95M	~MXN 0.92M	~MXN 0.00M	MXN 0.32	2.15	Use as the July 1 head-to-head proof, not as a broad scale claim.
Jalisco cluster	Green • pilot-backed	2,847	~MXN 17.09M	14.26%	~MXN 1.44M	~MXN 0.75M	~MXN 0.46M	~MXN 0.23M	MXN 0.32	1.29	Best candidate for the next full internal regional cell after CDMX proof.
Querétaro cluster	Yellow • weak proof	903	~MXN 6.15M	2.17%	~MXN 0.19M	~MXN 0.06M	~MXN 0.14M	~MXN 0.00M	MXN 0.74	0.41	Not a priority. More placeholder than scale thesis.
Monterrey	New • proxy only	679	~MXN 4.68M	14.26% proxy	~MXN 0.34M	~MXN 0.18M	~MXN 0.11M	~MXN 0.05M	MXN 0.32	0.31	Interesting as a shortlist, but still proxy-based and not wave 1.
León	New • proxy only	941	~MXN 5.33M	14.26% proxy	~MXN 0.48M	~MXN 0.25M	~MXN 0.15M	~MXN 0.08M	MXN 0.31	0.43	Shortlist only until the internal model is proven in CDMX.

Potential Kobra / vendor candidates

These are the vendor-side candidates in the same sheet. The CDMX Kobra benchmark is not separately sized there, so it should be treated as a design choice inside the July 1 benchmark, not as a pre-sized rollout row.

Cluster	Status / type	Current clients	Current total debt	Benchmark cure rate used	Expected field recovery	Expected calls & notifications recovery	Expected field spend	Expected delta	Price of 1 collected peso	Equivalent team-months @ 10/day	Read
Jalisco conservative case	Green • pilot-backed	1,537	~MXN 11.78M	2.93%	~MXN 0.32M	~MXN 0.06M	~MXN 0.19M	~MXN 0.07M	MXN 0.59	0.70	Good first-cut Pilot 4 case if the goal is a conservative vendor test.
Jalisco upside case	Green • pilot-backed	1,537	~MXN 11.78M	5.19%	~MXN 0.56M	~MXN 0.05M	~MXN 0.24M	~MXN 0.27M	MXN 0.43	0.70	Best upside vendor case if Jalisco performs like Pilot 1 rather than Pilot 2.
Chihuahua	Green • pilot-backed	473	~MXN 4.02M	3.02%	~MXN 0.16M	~MXN 0.01M	~MXN 0.06M	~MXN 0.08M	MXN 0.38	0.21	Cleanest region to continue with Kobra.
Aguascalientes	New • proxy only	289	~MXN 2.60M	2.93% proxy	~MXN 0.06M	~MXN 0.01M	~MXN 0.03M	~MXN 0.01M	MXN 0.50	0.13	Possible later experiment, not a first-choice launch.
León	New • proxy only	463	~MXN 3.28M	2.93% proxy	~MXN 0.10M	~MXN 0.02M	~MXN 0.06M	~MXN 0.02M	MXN 0.60	0.21	Borderline economics; keep behind Jalisco and Chihuahua.
Cuernavaca	New • proxy only	184	~MXN 1.48M	2.93% proxy	~MXN 0.04M	~MXN 0.01M	~MXN 0.02M	~MXN 0.01M	MXN 0.50	0.08	Too small for priority unless there is a strategic reason to test it.

Approximate monthly and yearly internal run-rate in the strongest regions

This normalizes the best regional packs into a run-rate view. For 31-60, the anchor is the pilot-backed Jalisco internal cluster already sized in the planning sheet. For 91-120, there is still no true internal pilot, so the estimate is a proxy: we keep the Jalisco pilot-backed recovery and baseline assumptions and replace vendor spend with an internal team cost of MXN 300k per team-month.

Bucket / region	Evidence type	Pack workload	Pack field recovery	Pack calls & notifications recovery	Pack internal spend	Pack delta	Approx monthly field recovery	Approx monthly internal spend	Approx monthly delta	Approx yearly field recovery	Approx yearly internal spend	Approx yearly delta
31-60 • Jalisco	Internal candidate from planning sheet	1.29 team-months	~MXN 1.44M	~MXN 0.75M	~MXN 0.46M	~MXN 0.23M	~MXN 1.12M	~MXN 0.36M	~MXN 0.18M	~MXN 13.40M	~MXN 4.28M	~MXN 2.14M
91-120 • Jalisco Conservative proxy	Vendor-backed recovery benchmark, internalized cost	0.70 team-months	~MXN 0.32M	~MXN 0.06M	~MXN 0.21M (0.70 × MXN 0.30M)	~MXN 0.05M	~MXN 0.46M	~MXN 0.30M	~MXN 0.07M	~MXN 5.49M	~MXN 3.60M	~MXN 0.86M
91-120 • Jalisco Upside proxy	Vendor-backed recovery benchmark, internalized cost	0.70 team-months	~MXN 0.56M	~MXN 0.05M	~MXN 0.21M (0.70 × MXN 0.30M)	~MXN 0.30M	~MXN 0.80M	~MXN 0.30M	~MXN 0.43M	~MXN 9.60M	~MXN 3.60M	~MXN 5.14M

How to read this: monthly and yearly figures are simple run-rate normalizations, not guaranteed inflow. They assume the bank can continuously replenish the same eligible volume in the same geographies and keep the team utilized. The 31-60 Jalisco row is the strongest current internal anchor. The 91-120 Jalisco rows are useful only as a proxy range until an internal late-bucket benchmark exists.

4. Geographic / state-level recommendation

Prioritize geographies where pilot proof, density, and manageable field execution overlap.

The map below is intentionally conservative. It ranks places not only by modeled upside, but by whether the supporting evidence is pilot-backed or merely inferred from current opportunity sizing.

Wave 1: scale now

Chihuahua 91-120 with Kobra, especially Chihuahua and Juárez pockets.
Jalisco 31-60 remains the strongest next internal replication candidate.

Wave 1: prove now

CDMX 31-60 benchmark in the prioritized borough pack.
Run internal and Kobra under the same targeting rules.

Wave 2: conditional

Estado de México 31-60 in adjacent dense metros.
Only after productivity and conduct controls are stable.

Test only / stop

Jalisco 91-120 Kobra: Guadalajara + Zapopan core, San Pedro Tlaquepaque optional.
Do not scale: broad CDMX 91-120, Nuevo León 91-120, Puebla 91-120, Querétaro 91-120.

CDMX 31-60 launch pack: prioritized areas

This is the recommended first internal pack because it combines good field-fit evidence with attractive gross pack economics. Municipality-level cure is not attached in the source, so the table shows localized and direct contact as field-fit proxies.

Municipality	Projected field recovery	Projected field spend	Gross field result	Price of 1 collected peso	Localized	Direct contact	Action
Benito Juárez	MXN 186,981	MXN 112,581	MXN 74,400	MXN 0.60	97.9%	8.3%	Start
Coyoacán	MXN 266,975	MXN 160,745	MXN 106,230	MXN 0.60	89.9%	5.1%	Start
Cuauhtémoc	MXN 284,643	MXN 171,383	MXN 113,260	MXN 0.60	89.2%	6.5%	Start
Venustiano Carranza	MXN 213,973	MXN 128,833	MXN 85,140	MXN 0.60	93.2%	5.4%	Start
Gustavo A. Madero	MXN 425,492	MXN 256,188	MXN 169,304	MXN 0.60	88.6%	4.9%	Reserve / add after proof
Iztacalco	MXN 189,435	MXN 114,058	MXN 75,376	MXN 0.60	96.1%	3.9%	Reserve / add after proof

Why CDMX 91-120 is not in wave 1

The attached evidence is consistently weak for CDMX 91-120. The state-level pilot result is negative net, and every CDMX 91-120 municipality in the planning file is already gross-negative before subtracting the calls & notifications baseline.

Municipality	Projected field recovery	Projected field spend	Gross field result	Price of 1 collected peso	Localized	Direct contact	Action
Miguel Hidalgo	MXN 57,691	MXN 68,565	MXN -10,873	MXN 1.19	88.2%	1.8%	Do not start
Coyoacán	MXN 81,660	MXN 97,051	MXN -15,391	MXN 1.19	92.8%	2.9%	Do not start
Gustavo A. Madero	MXN 154,959	MXN 184,164	MXN -29,206	MXN 1.19	83.9%	7.5%	Do not start
Iztapalapa	MXN 186,173	MXN 221,262	MXN -35,089	MXN 1.19	76.0%	7.5%	Do not start

Simple answer for the CRO

We are not recommending CDMX 91-120 because the bank already has direct CDMX 91-120 evidence, and that evidence is negative after subtracting both the remote baseline and field spend.

If CDMX Kobra is launched anyway

It should be framed as a matched 31-60 benchmark against internal, not as an implicit approval for broad vendor scale in CDMX.

Targeting inside geographies

The Snowflake cut supports prioritizing debt 10k-30k and credit limit 10k-30k. That should be the day-one suppression and prioritization rule.

5. Improvement levers before or during scaling

The best levers are the ones that improve route quality, target quality, and true incrementality.

Not every idea should be a prerequisite. The right sequencing is to enforce the few levers that most directly change wasted visits, cannibalization risk, and conduct quality before headcount expands.

Lever	Likely value	Feasibility	Should it be a prerequisite?	Recommended stance
Address enrichment bureau + living address confidence	High	Medium	Yes	Highest-value prerequisite. It directly reduces wasted visits and should be built into dispatch gating.
Address prioritization algorithm	High	High	Yes	Implement early. Use the same mindset as third-party-number prioritization.
Image sharing already cleared by infosec / compliance	Medium	High	Operationally yes	Include in rollout. Strong governance and QA benefit, even if cash impact is indirect.
Rapid revisits	High	High	Yes	Operational evidence is strong. Put a 4-7 day revisit SLA into the launch design.
Visiting-hour adjustment	Medium	High	No	Quick win. Shift harder accounts to late afternoon and evening.
Geo-location check + silent pushes	Medium	Medium / low	Not yet	Interesting, but needs tighter legal and conduct design before becoming a proceed/stop gate.
Credit limit and total debt targeting	High	High	Yes	Part of rollout. Current evidence supports mid-high balance prioritization from day one.
Recent contact status in last 7 days	Potentially high	Medium	Needs testing	Important missing analysis for incrementality. Use as a suppression test before broad scale.

6. Next steps and expansion path

Immediate next steps, and what happens if the CDMX launch works

The rollout should be framed as a sequence, not as a one-time national decision. The safest pattern is: launch a July 1 CDMX 31-60 matched benchmark, run Kobra Pilot 4 in a narrow Jalisco core, deploy the highest-value improvements, then expand only if the gates are hit.

Step 1

Launch CDMX 31-60 on July 1 as a matched benchmark: internal in the core prioritized boroughs and Kobra in a matched carve-out under the same targeting rules.

Step 2

Run Kobra Pilot 4 in Jalisco 91-120 with Guadalajara + Zapopan as the core. Add San Pedro Tlaquepaque only if week-1 volume is insufficient.

Step 3

Put improvements live from day 0: address enrichment, address prioritization, image sharing, fast revisit SLA, visiting-hour adjustment, and recent-contact suppression test.

Step 4

Set a hard decision gate after the first operating cycle and expand only if the matched results prove productivity, conduct stability, and positive economics.

If the CDMX benchmark works, then expand in this order	Why this is the next move	What has to be true first
1. Add reserve CDMX boroughs Gustavo A. Madero, then Iztacalco, then later-wave areas selectively	Lowest execution risk. Same city, same team, same oversight model, and the cleanest way to keep the Kobra comparison honest.	Internal sustains roughly 10+ visits/day and at least matches Kobra on cure while beating it on unit economics.
2. Extend into adjacent Estado de México dense metros Ecatepec, Nezahualcóyotl, Naucalpan, Tlalnepantla, then adjacent clusters	Best way to deepen utilization using contiguous urban density rather than opening a totally new operating cell immediately.	CDMX proof is stable, address quality holds, and the wider footprint does not break routing or conduct controls.
3. Open the next full internal regional cell in Jalisco 31-60	Jalisco is still the strongest replication candidate once the internal model itself is proven.	The bank is comfortable funding a second internal cell and the first cell is operationally stable enough to replicate, not just to manage locally.

Clean answer on expansion sequencing: if the July 1 CDMX benchmark goes well, do not jump immediately to broad national rollout. First expand within CDMX reserve areas, then into adjacent Estado de México dense metros, and only then replicate the full internal model into Jalisco as the next regional hub.