---
brand: PayWise
date: 2026-06-03
participants: Phill Lewis (PayWise); Ian Alleyne (PayWise); Sylvannas Berkeley (PayWise); Pablo Esparza Jimenez (Coordenadas)
topic: Monthly higher-ups review — platform performance data, registration & bill-payment funnel, marketing alignment
source: "PayWise Wednesday Meetings - Higher ups - Jun 3.txt" (raw: raw/PayWise_Meeting_2026-06-03_higher-ups-monthly_raw.txt)
tags: [meeting, higher-ups, performance-data, analytics, activation, bill-payment, funnel]
---

# Summary

- **Fecha:** 2026-06-03
- **Participantes:** Phill Lewis (PayWise), Ian Alleyne (PayWise), Sylvannas Berkeley (PayWise), Pablo Esparza Jimenez (Coordenadas)
- **Marca:** PayWise
- **Tema:** Monthly higher-ups review — platform performance data, registration & bill-payment funnel, marketing alignment
- **Duración:** ~80 min (last timestamp [79:46])

> Nota del solicitante (contexto, no dicho en la reunión): esta sesión se pensó como
> una reunión mensual de management (nominalmente el día 5) y no se ha vuelto a
> realizar. Muchos de los ítems son de junio 2026 y ya están vencidos.

**Resumen ejecutivo**

- Coordenadas (Pablo) presentó el Marketing Hub, el Marketing Brain y el proceso de filtrado de contenido; management reaccionó positivo a usarlo como contexto para agentes de IA.
- Marketing por correo: ~10% de tasa de apertura (abril–junio); la meta de click-through es 1.4. Pablo tiene contenido programado hasta fin de mes y más en producción.
- Nuevas cuentas ~70–90/mes (Feb 17, Mar 70, Abr 56, May 77). La caída de abril se atribuye al fin de semana de Semana Santa. El cruce con el volumen de posts sugiere que los sign-ups vienen de contenido orgánico más que de pauta.
- Hallazgo central: el embudo de registro gotea. ~502 eventos de instalación (marzo); en Android 160 instalaciones y 90 desinstalaciones (mayo); ~533 intentos de registro en mayo con fuerte caída después; ~1,100–1,200 sesiones de login sin siguiente acción en 90 días.
- Advertencias de calidad de datos: "Ready to Pay" no cuadra con pagos completados (35 vs 13 en 90 días; 6 en los últimos 30 días); faltan métricas de iOS; el doble conteo de alta personal/negocio infla el número de usuarios nuevos.
- Dirección acordada: mejorar la activación post-registro y la guía al usuario, hacer visibles los métodos de pago (radio buttons en vez de dropdown), atar recompensas a conductas deseadas, y dar a marketing mejores datos (notificaciones de error, lista de emails de usuarios, reporte mensual).
- Coordenadas consolidará un reporte base con "valores reales" (Play Store, Search Console, social) y armará plantillas de reporte mensual; después se agregarán integraciones y se validarán contra ese baseline.

# Notes

## 1. Marketing Hub, Marketing Brain y agentes de IA

- Pablo mostró un hub interno que construyó para onboardear a su propio equipo (diseñadores, respuestas a mensajes): una tabla con recursos, fees, documentación y archivos de flyers ya hechos; todo mantenido actualizado.
- El hub es público y se lo compartió a Phill (ya lo conocía) e Ian.
- Pablo explicó que el **Marketing Brain** se usa para fact-checkear el contenido: cada pieza pasa por los "message blocks" para asegurar la comunicación correcta a la audiencia.
- Phill: muy bueno, especialmente las últimas páginas; sugirió que un agente de IA encima de eso podría usarlo como "repo" para responder preguntas.
- Pablo: están configurando agentes para recorrer estos procesos **mensualmente**, con triggers manuales para campañas más personalizadas.
- Phill: si el hub se mantiene como única fuente de verdad y se mantiene actualizado, la agencia casi siempre recibirá el contexto correcto.
- Pablo: un agente también podría revisar documentos de registro/KYC (business registration PDF) y hacer cambios en la plataforma siguiendo instrucciones, sin integración por API.

## 2. Email y social (abril–junio)

- Email marketing sostenido: **10% de open rate** (descrito como bueno); el click-through varía mucho y la meta es mantenerlo/mejorarlo en **1.4**.
- Phill: la entrega y la apertura se ven muy bien.
- Phill preguntó si las métricas mostradas eran solo de Instagram/Meta; Pablo aclaró que esa vista era solo social.

## 3. Contenido y foco de junio

- Foco actual de Pablo: contenido educativo para el canal de la empresa.
- Ya tiene suficiente contenido programado hasta fin de mes; el resto (lado visual) está en producción.
- Ángulo en desarrollo: **"easy to use"** primero, luego **"free to use"** como test.
- Pablo está armando **plantillas de reporte** que se actualizarán cada mes (impresiones, posición promedio, tendencia).
- Busca correlaciones como la que encontró para otro cliente (relación entre impresiones sociales y cierres de negocio: quien entra a social suele estar en el pipeline o negociando). Quiere replicarlo para PayWise recolectando datos.
- El reporte gráfico de Phill alimenta ideas: al ver caídas grandes en un paso, se puede crear contenido educativo que explique mejor ese paso.

## 4. Volumen de nuevas cuentas y correlación orgánica

- Ian aportó el número de nuevas cuentas: **Feb 17, Mar 70, Abr 56, May 77**. Atribuye la caída de abril al fin de semana de Semana Santa.
- Ian: el rango típico de sign-ups es **70–90**; ofreció dar a Pablo las **fechas** en que se crean las nuevas cuentas del último año, que Pablo quiere poder graficar.
- Pablo observó que bajaron los posts en marzo y los retomaron en abril/mayo, y el sign-up repuntó: su hipótesis es que los sign-ups vienen de **contenido orgánico**, no de pauta.

## 5. Mensajería social, bot y soporte

- PayWise recibe DMs por Instagram/Facebook; al primer mensaje se activa el bot y se **crea un contacto nuevo** automáticamente en la lista, con la fuente registrada.
- Si el bot llega a un "roadblock", manda una alerta a Pedro.
- Phill quiere entender, mensualmente, el volumen y la **naturaleza** de esas conversaciones (quejas, preguntas, soporte, ayuda con la API) para saber en qué enfocarse.
- Pablo: buscaría una solución con **Hermes** (los agentes de IA actuales no son tan listos); también recordó que hay un web scraper con el que había scrapeado Docs y una conversación profunda donde el usuario no sabía que hablaba con un bot.
- Volumen bajo: **16** contactos nuevos el mes anterior y ninguno en lo que va del mes; Pablo cree que sube cuando las ads están encendidas (ahora están apagadas).
- TikTok: se puede **postear** desde la plataforma, pero no responder mensajes de TikTok.
- Pedro manejaba la conversación de la plataforma; Pablo tuvo que tomar esa conversación.

## 6. Analítica de plataforma (login, registro, bill payment) — dashboards de Phill

- Phill compartió su tablero de tracking (últimos 90 días) sobre acciones tras el login: la mayoría va primero a **pagar facturas** o a "ready to pay"; algunos intentan **send money** (un caso 143 veces); hay caída en la etapa de completar.
- **~1,100–1,200 sesiones de login sin siguiente acción** en 90 días; 25 tras registrarse no hacen nada.
- Phill enfatizó que la *próxima acción después del registro* es un hueco grande ("huge leaky funnel").
- Aún hay trabajo de "tightening" del payload: algunas métricas se pierden y no se pueden coser; varios eventos no se ingieren bien en la base (trabajo con "Elish").
- Phill mostró una segunda versión del tablero, de "base": encontró **~356 usuarios nuevos** en 60 días, número que sospecha inflado por **doble conteo** (se registra como negocio y personal a la vez) y necesita business logic.
- Incluye errores de onboarding/acceso y top-up de wallet, con flagging en tiempo real.

## 7. Embudo de registro: instalaciones vs registros

- Phill: en mayo hubo **533 intentos** de registro, y una gran caída después de elegir personal/negocio. Sospecha que parte puede ser bug y quiere validarlo con Elish.
- Ian pidió a Pablo revisar **cuántas descargas** hubo, en Google Play y en App Store.
- Pablo (Play Store, mayo): **160 install events**, **90 uninstalls**, **128 daily devices installs** desde **105 users**, **215** (otra métrica). Es solo Android.
- Phill: vio **208** instalaciones Android y **287** iOS. Se discute si los distintos valores de "installs" se están sumando (128/215/240/368).
- Descargas de Android muy por encima del número neto (~70) que ve Ian: **~50% de caída** entre instalar y usar la app.
- Pablo: **160 install events** es lo más confiable porque viene directo de Google Play. Se comprometió a descargar la info manualmente y consolidarla.

## 8. Bill payment: completación y UX de métodos de pago

- Discrepancia central: "Ready to Pay" ~**56** en 90 días con **5** que se caen, contra **13** bill payments completados reportados por Ian en 90 días y **6** en los últimos 30 días.
- Ian: no tienen visibilidad de los bill payments hechos por usuarios; Phill aclara que su tablero cubre **toda la plataforma**, no solo la app, e incluye agentes. Al sacar agentes, "ready to pay" baja a 56 y 5 abandonos.
- Sylvannas: posible que el usuario haya seleccionado el vendor, pero no terminó la verificación/finalización.
- Phill sospecha que "Ready to Pay" puede estar contando mal (35 intentos vs 13 completados).
- Eventos que se trackean: `Bill pay`, `Bill pay ready to pay`, `Bill pay completed`.
- **Propuesta de Ian (UI):** el selector de método de pago es un **dropdown**; la opción por defecto es **wallet** (sin fondos), así que el usuario podría no ver tarjeta débito/crédito. Pide cambiar a **radio buttons** para que las opciones sean visibles, y pide a **Nathaniel** el cambio ("superficial cheat, high level UI change").
- Sylvannas sugirió poner "agregar tarjeta" primero; Ian explicó que el flujo actual obliga a agregar tarjeta durante la transacción. Phill recordó que guardar la tarjeta depende de regulación PCI y del retorno del procesador tras completar el pago.

## 9. Manejo de errores y seguimiento al usuario

- Phill trackea en tiempo real errores de onboarding (selfie fallida, subir ID múltiples veces, acceso, balance de Aircon, top-up de wallet).
- Pablo propuso: cuando a un usuario le ocurre un error, mandarle de inmediato un email de marketing reconociendo el error, disculpándose y ayudando manualmente o pidiéndole reintentar en una hora.
- Pablo propuso copiar la notificación de error de Phill a un **inbox separado** (creó un email para el CRM de PayWise) y correr una **automatización** que contacte al cliente o le avise a Pablo, según convenga.
- Phill: se puede hacer juntos — "spin up those journeys": que al detectar el error se dispare un email a Pablo con la información, y el bot escuche. Pablo: "let me put in the request properly" y tomar nota de qué información necesita.
- Phill aclaró que el valor de la recompensa hay que definirlo: **credits de PayWise**, "satoshi", etc., para poder anunciarlo.

## 10. Recompensas / incentivos de conducta

- Idea de Pablo: en la creación de cuenta (primeras 24 h), enviar un email con un primer hito a completar a cambio de "algo": por ejemplo conectar la cuenta bancaria. Hay que definir qué recibe el usuario.
- Phill: atar recompensas a conductas que además generen ingresos (bill payment, mantener saldo/cash en la plataforma, etc.); "tie it to rewards" para incentivar la conducta deseada.
- Pablo: comunicarlo no solo en social sino por email blast a la lista actual: "have you not connected your bank account? Do this and get extra [reward]. If you already did, click here to request it."
- Meta relacionada: enviar por el mes how-to guides y recordatorios de pago de facturas alineados al ciclo de facturación.
- Destacó como beneficio "pay with card" y "tap on page" para el nuevo contenido.

## 11. Integridad de datos y reporting

- Phill: algunos eventos no se ingieren bien por cómo se reciben de Elish; va "chipping away" cuando tiene tiempo.
- El stitching del funnel personal (trazar ID → email verificado → inicio → completación) no está armado y no ha tenido tiempo de resolverlo.
- Phill propuso **examinar el modelo semanalmente** para que alinee con el desempeño real; Ian lo apoyó como crítico para alimentar mercados y estrategia y disparar "next best actions".
- Phill: le gustaría que la data se pueda segmentar por plataforma y por rango de fechas (algunos filtros no funcionaban en la demo).
- Pablo cerró: consolidará todo en **un documento de reporte** como el que hizo para otro cliente, tomado de **exports reales** (Google Play, Google Search Console, social, etc.) sin integración; luego agregará integraciones y las comparará contra esos valores reales, y recién cuando las integraciones coincidan, se podrá confiar en ellas.

## 12. Información de usuarios y notificaciones

- Pablo pidió a Ian la **lista de emails de usuarios** (solo emails) y si son cuenta **business o personal**; con el teléfono podría empezar a hacer notificaciones de WhatsApp.
- Ian aceptó compartirla.
- Phill preguntó cómo hace Pablo los pushes de WhatsApp (¿API de WhatsApp u otro?).
- Pablo quiere además el **email list formal** para crear notificaciones; su socio lo ayuda con los pushes de WhatsApp vía **n8n**.

# Transcript

> Transcripción limpiada (muletilla/repetición eliminada; contenido conservado).
> El original sin tocar está en `raw/PayWise_Meeting_2026-06-03_higher-ups-monthly_raw.txt`.

**[0:09] Phill Lewis:** How are you, Pablo?

**[0:10] Pablo Esparza Jimenez:** All good, all good.

**[0:14] Sylvannas Berkeley:** Good evening, Pablo, what's up man.

**[0:17] Pablo Esparza Jimenez:** Good evening, all good, thank you.

**[0:49] Ian Alleyne:** [On] mesh all the different aspects of the service so that we can learn how to scale — that is, how to generate and encourage more activity on the platform. And then once we have understanding of that, then we can go hard in the paint, bring more people onto the platform knowing that with what we understand and know, once they come on we can start getting them to use the platform immediately. So it's an overview but it is very important — the rule that you're playing here, because it's the customer messaging.

**[3:04] Pablo Esparza Jimenez:** I also think these meetings will help me stay focused on what the real updates are, and I'll be able to make better content. The organic content was able to maintain the current sign-up rate. If anyone has a question or answer regarding that, that'd be cool.

**[5:24] Ian Alleyne:** I'll double check that, but from my impression I don't think there was any material change. In terms of my involvement, or the number of accounts I had to treat with, I did not [see a change].

**[5:59] Pablo Esparza Jimenez:** [Content updates —] more educational, with actual screenshots of the content. Those are the updates on my side. We've also kept doing email marketing. I have some metrics here for April to June: we have a 10% open rate, which is pretty good. And the click-through rate varies a lot, so the idea is to bring this up, keep it at 1.4, which is good. So that's what we keep working on. Any questions regarding that?

**[8:07] Phill Lewis:** Not from my side. The delivery and open [rate] seem really good. So this is a combination of Google Analytics, only Instagram like Meta?

**[8:34] Pablo Esparza Jimenez:** That was only social media. I'll show you right now what we're working with. This is something I also worked on — I forgot to show you. It's a hub that I created; this was the easiest way. This was something I built internally, primarily, then I decided to show you again because I made it public — Ian and you, Phillip, I shared this with you.

**[8:54] Phill Lewis:** Yes, I remember that.

**[8:54] Pablo Esparza Jimenez:** So you'll find various resources. This was just the easiest way for me to onboard my own team: photographers/designers find everything related to graphic design; the ones who reply to messages — people ask where to find information and everything is always on the website. So I decided to list this table and like anybody asks about fees, just go to the table, check out the fees and click the link to read all about fees. That's a good thing about this website: we have everything up to date. Files for flyers or things we've done; the PayWise documentation, which is really cool — I was going through it the other day.

**[12:28] Pablo Esparza Jimenez:** These are the updates. This Marketing Brain is what we're using — we fact-check all the content we now make with it. It has to go through some filters, which are basically the messaging blocks, always going through the messaging blocks to make sure that we communicate to our audience the correct way.

**[12:46] Phill Lewis:** Yeah, I think that's really very good — especially the last couple of pages you just showed, because I was just thinking: if you have an AI agent sitting on top of that, it could use that as a good repo to get some answers. I think that's why you set it up that way.

**[13:07] Pablo Esparza Jimenez:** You've got that with it, and we're trying to configure agents to basically go through these processes on a monthly basis, and we also run them; they have manual triggers for the on-and-off more personalized campaigns that we make.

**[13:33] Phill Lewis:** Very interesting, because if you have that as a context, you update that as the single source of truth, make sure that it's [right] and always accurate, and then 9 out of 10 times it could more or less rely on — the agency's going to be fed the right thing.

**[13:54] Pablo Esparza Jimenez:** [Users] have to be manually [verified]: look at their files — either building address or the business registration PDF. Those are things that an agent could do too, with the proper access: view the information and make the changes in the platform without any API integration, just following instructions.

**[14:44] Ian Alleyne:** So what's the main focus for June?

**[15:11] Pablo Esparza Jimenez:** Right now I'm working on creating that [educational] channel content.

**[15:15] Ian Alleyne:** Okay, the educational content.

**[15:16] Pablo Esparza Jimenez:** Correct. I already have enough content scheduled all the way through the end of this month, and more content is being created right now — the visual side.

**[15:37] Ian Alleyne:** And that content target — what specifically?

**[15:39] Pablo Esparza Jimenez:** That one is the "free to use." Sorry, no — the "easy to use" first. We're gonna go with "easy to use" and then we test out "free to use."

**[15:55] Pablo Esparza Jimenez:** The next thing I have pending for PayWise — one of my other priorities — is having this report ready. These are all templates that I'm creating. This will be updated every month with the next impressions, what the average position has been, if it's been improving or declining. For example, for this client it has improved from May 2025 — it was around 37th position and now it's around 25th. So it improved a bit. I want to find these kinds of things with PayWise — see if I find any correlating data. For some reason I found there was a relationship between social impressions and closed deals for this company. That told me something very particular: people who go into their social media are usually people who are closer to dealing, or negotiating with them — they're in the sales pipeline somehow. [I want to] find out with PayWise. So right now this is another focus I have: collecting data so I can have something like this and make those kinds of relationships.

**[18:28] Ian Alleyne:** Pablo, I put in the [sheet] the number of new accounts. You can see in February there was 17, in March there were 70, in April there was 56 — we dropped. In April I think the Easter weekend is the period; we had an effect. And [in May] it shot back up to 77.

**[19:12] Pablo Esparza Jimenez:** Interesting. You know what, I think that was all organic posting. In May, let's pull up the amount of social media posts — let's go back to this graph. We did lower the amount of posts in March; April and May we picked it back up. Maybe that tells me that the sign-ups are actually coming from organic content and not so much the paid content.

**[19:46] Ian Alleyne:** And these figures are not off. As I said, it's between 70 and 90 — that's typically where you see the number of sign-ups, although in April there was a period in which there were no new accounts, and I reason it was due to Easter. So what I can do: I can also provide you with the dates on which your new accounts are being created, over the last year.

**[20:38] Pablo Esparza Jimenez:** Yeah, that'd be useful, and we get that on a graph.

**[20:45] Ian Alleyne:** Yeah, you can do up a graph on that.

**[31:51] Phill Lewis:** [Looking at pipelines —] leads, leads and opportunity.

**[31:51] Pablo Esparza Jimenez:** Different pipelines — one for support, one for sales. This is actually a pretty good conversation to have.

**[31:51] Phill Lewis:** That's actually a good thing to do because there's a lot of business opportunity.

**[31:51] Pablo Esparza Jimenez:** [With] phone numbers, they should have a phone number registered that way. I could do the emails a bit more personalized — like "Hi Pablo, I noticed you...". That question, I'd have to ask my business partner: he's the one who actually helps me with the WhatsApp pushes and integrations; he does it through n8n.

**[31:51] Phill Lewis:** Okay, maybe back up a bit just so that I get caught up.

**[31:51] Pablo Esparza Jimenez:** Instagram, Facebook — that's it. It was 11 and 16.

**[31:51] Phill Lewis:** So the 16 right now.

**[31:51] Pablo Esparza Jimenez:** Meaning Instagram — that's the 16.

**[31:51] Phill Lewis:** And one person last month, nobody so far for this month. It'll be good — maybe you might not be able to say, but to see the groupings of those conversations on a monthly basis. Like how many of these things are complaints, questions, support related, help on the API layer. It'll be really interesting to know what those funnels look like.

**[31:51] Pablo Esparza Jimenez:** I agree. Let me find a solution for that using Hermes, because these AI agents are not as smart.

**[31:51] Phill Lewis:** How do you know what to focus on in? Because it seems like there are hundreds of messages in here.

**[31:51] Pablo Esparza Jimenez:** Maybe we focus on improving the structure. Because here you see it's all 71 messages from those 16.

**[31:51] Phill Lewis:** And I wonder, can this filter by platform? Actually no, nobody from May either. Although there's not a data point for that.

**[31:51] Pablo Esparza Jimenez:** There's another way to get the metrics and find the information someplace else — Reporting, I think. I have to go through it to actually check, but I remember back in the day [there was] integration. And Pedro was good. So I had to take over that conversation, but he kept asking questions I wasn't sure how to answer. This thing has a web scraper, so I scraped the whole Docs — which I have to redo now that I think about it. I went through the conversation; I remember I took a screenshot and shared it with Ian at some point. I'll look for it and share it with you, because it was having such an in-depth conversation and I don't think the user knew he was talking to a bot.

**[31:51] Phill Lewis:** Where is this user engaging with this bot?

**[31:51] Pablo Esparza Jimenez:** Through Instagram or Facebook, wherever the user decides to message.

**[31:51] Phill Lewis:** It would be an easy thing to answer if you know what to prioritize. If the answer is "we have like a hundred unique people per week..."

**[31:51] Pablo Esparza Jimenez:** No, it's not many. It's mainly because of the ads — when ads go on, communication goes up. Right now the ads are off, maybe we don't need to focus on this.

**[31:51] Phill Lewis:** From a data perspective — how many of these messages happen per month? Do you know, even not statistically? Let's make it up: on LinkedIn, or Facebook for example, and we have a bot that will reply to them.

**[31:51] Pablo Esparza Jimenez:** So if I go to Instagram and DM you, it works with DMs. The bot will start replying. We could have more than one.

**[31:51] Phill Lewis:** Do we have TikTok as well?

**[31:51] Pablo Esparza Jimenez:** I can post to TikTok from here, but I cannot reply to TikTok from here. One of the reasons we haven't opened TikTok: we have to coordinate creating the account from Trinidad, because it matters a lot where you create the TikTok account — it takes the IP of the location you open it from. So it's best that you create the account [from Trinidad] and then give me a password so I can connect it here and post to TikTok from here.

**[31:51] Phill Lewis:** So someone on Instagram would send a DM to PayWise, the bot gets activated and starts replying, and this platform consumes that whole conversation.

**[31:51] Phill Lewis:** It was only 16 unique contacts.

**[31:51] Pablo Esparza Jimenez:** That's 16 unique — new people who messaged the page — and it generates new contacts in the Contacts list. The contact is automatically created and it tells us the source.

**[31:51] Pablo Esparza Jimenez:** If I go to conversations, all the conversations through Instagram, email, Facebook fall here. The AI agent takes that conversation and starts replying. If for some reason it hits a roadblock, it sends an alert to Pedro.

**[32:05] Pablo Esparza Jimenez:** We could target them directly, not through ads, and maybe they will engage more because they actually have an account. I was planning on sending them how-to guides and reminders — by the end of the month, send them reminders to pay their bills or how to pay their bills around that date.

**[32:42] Pablo Esparza Jimenez:** Also big right now: the fact that they can pay with card — that's a big benefit, and "tap on page." Another sentence I've been using for the new content.

**[32:55] Phill Lewis:** I'm going to share my screen for a second. Let me know when something shows up.

**[33:48] Pablo Esparza Jimenez:** Yep, I'm seeing it.

**[33:49] Phill Lewis:** From the user: when they download the app, if they install the app and go through different screens — in the onboarding process, bill payments, money transfer process, push notifications — we try to log it. Small style to get good clean data. I think there's still some tightening up of the payload; sometimes we lose some of the metrics and aren't able to stitch things together correctly. What I found really interesting: when they use the app, this is data for the last 90 days — what action do they take right after login and immediately. You can see most people try to go to pay bills, to the "ready to pay" — still around bill payment on this side.

**[35:19] Pablo Esparza Jimenez:** Interesting. Is it only a counter, or do you get the recorded user ID or the user email who tried that action?

**[35:30] Phill Lewis:** Once they go through the registration process and select one of those, we start getting a little bit more. [It] creates what we capture for them, and then we get to see where things drop off. There's this second action after login: they try to pay their bill first, and then they get ready to pay, which kind of makes sense. Some people log in and try to send money 143 times. Then it's the completion piece. And on this side, here's what happens after — this is registration sessions without the next login.

**[36:56] Pablo Esparza Jimenez:** A question: what would be the best way if I wanted, for some reason, they haven't come back in a couple hours — we send them an email saying "Hey, don't forget to pay your bill, just click here and resume the action."

**[37:24] Phill Lewis:** With this, it's 100% possible to do that. We just need to configure the system to listen for those types of triggers. Maybe it's a job that runs every minute, asks questions of a series of things — who hasn't done this in the last [period] — and then we connect with them. And here, if they signed in with their phone or some people signed in with email, regardless of which one they choose we can capture that information off of it.

**[38:10] Phill Lewis:** What I was trying to show up here: look at how many login sessions without next action — in the last 90 days that was like 1,100, almost 1,200 sessions. Nobody does anything after they logged in. 25 of them, after registering, they don't do anything. This would be better to know out of how many. I just marked this up really quick, just to have a conversation. We don't have enough time to sit with this as a group and really come back and say how this can be further enhanced to tell a story.

**[39:08] Phill Lewis:** Look how many times [they clicked one of the ads], which is interesting that they would have done that. I wonder what they were trying to get to. One person [went to] their bank account, which is probably one of the right actions we want them to take. And Ian will get into this at the appropriate time, but one of the things we want to do is tighten up the next action after registration because there's a lot of holes. Look at friction, which I also find fairly interesting: requests money, same thing, almost 30% of them fall out.

**[41:22] Pablo Esparza Jimenez:** But people are paying bills.

**[41:24] Phill Lewis:** Yeah, most of them are paying bills.

**[41:30] Ian Alleyne:** What do you mean by "bill payment, ready to pay"?

**[41:36] Phill Lewis:** These different screens you go through give different messages. You select you want to pay a bill, then you get to it. You go to the Services screen on your app and tap on it — it's being selected. Then you select the bill you want. After you fill out all the details — in terms of how much, what percentage — [we see] does not finish, basically. According to this, after this starts, most people finish or want to finish, at least.

**[42:27] Ian Alleyne:** One second, that's not good. That's not good.

**[42:35] Phill Lewis:** I'm not disputing you at all, but that could just mean the way we're collecting this data needs to be verified and tightened up. Directionally it's something that is too good to be true as you're saying, so we should check it out. I'll check out this calculator.

**[43:05] Ian Alleyne:** What I'm reading that to mean is: the bill payment attempt — they attempt a transaction, but they don't go beyond that screen where they enter the information. Which is curious, because maybe the payment method option is not clear to them.

**[43:53] Phill Lewis:** That's what it's saying. So you're saying this number is too low.

**[44:17] Ian Alleyne:** No, I'm seeing the total. What this is suggesting is that you have 3,000 and 37 people [complete the function].

**[44:31] Phill Lewis:** Not 97, so like 3,300. Any [in the] last 90 days.

**[44:42] Ian Alleyne:** In terms of bill payments by persons from within the app, we're not seeing that number.

**[44:49] Phill Lewis:** Keep in mind, this is not only the app. This is the whole platform.

**[44:58] Ian Alleyne:** Oh, so this represents the agents then.

**[45:02] Phill Lewis:** Yeah, I'll just put this on everywhere. So I just took off the agents from the data set. As you can see, "ready to pay" [is] 56 in the last 90 days, and five of them dropped off.

**[45:46] Sylvannas Berkeley:** This is "ready to pay," Phil. Is it that they selected their vendor who they want to make a bill payment to, and selected an option, and they just haven't completed the verification and then finalize[d]?

**[46:01] Phill Lewis:** That's how I understand it, but I could verify that.

**[46:07] Phill Lewis:** [Last] 90 days.

**[46:09] Ian Alleyne:** I can tell you what's happening over the last 90 days. So that takes us back to March. [We] just [look at] activity.

**[46:57] Phill Lewis:** We could just look at this month or last. This is last.

**[46:59] Sylvannas Berkeley:** Do the last [period] contrasted and compare with what you see, because this is what Ian has seen on the back end. Maybe there's some sort of data duplication.

**[47:17] Ian Alleyne:** Okay, so we went live with the in-app bill payments using a card. Over the last 90 days we had 13 in-app bill payments.

**[48:16] Phill Lewis:** Okay, so there could be a problem with how "Ready to Pay" is being calculated. Maybe it's 35 [that] people tried to start it, and it's calling it that, and then to your point only 13 of them completed in the last 90 days. How many of them in the last month? Because I don't know when I turned this feature on, to be honest.

**[48:40] Ian Alleyne:** Last month... six.

**[49:05] Phill Lewis:** Six people completed it. The type of events there are: Bill pay, Bill pay ready to pay, and then Bill pay completed. That's the type of events we track.

**[49:39] Ian Alleyne:** But what we can do is — when you [decide], let's do the necessary modification. We could always study this back again [against] the transaction set.

**[49:54] Phill Lewis:** Yeah. And a key thing to keep in mind: several of these events still fail to get ingested into the database properly. That's because of how we get it from Elish — I've been working with him whenever I have time, which is not that often, to keep taking a look at why some shapes or the payload that his team created isn't working, and slowly but virtually we've been chipping away. There's one more thing I wanted to show you. I didn't like that, so instead of trying to enhance it some more, I created a second version of this. This one is a kind of a base version.

**[50:55] Phill Lewis:** As you can see, the error rate is still fairly high, in terms of the events that we're supposed to [get] when we're not doing anything with it. What I find interesting is how many active users we've found in the last 60 days according to this — it feels like there's 356 new users, and I need to figure out how it got [there]. These numbers are really wild. How is this still high?

**[51:26] Ian Alleyne:** Over here.

**[51:30] Phill Lewis:** We're still at 30 days though. I don't know where these people come from. Maybe I just need to do this. It's still fairly high. I don't know if it's "unknown" as well. This is still fairly high. Maybe what's happening is it's double counting. When people sign up as a business in the beginning, we don't really know who they are but we know that they went through the business flow, so we log it as a business. This in itself needs some business logic to just type it up, because it inflates these numbers. If I just do this for now — any person, only the personal side — it might be a little better, but still that's very high.

**[53:09] Phill Lewis:** But the point is, I find this piece really insightful here, because the stitching is not set up and I haven't had time to figure it out: how many people start and how we can trace them through this whole funnel — on the personal side, like the ID, then email verified. How is it growing? It needs to be checked. But you get to see how these people drop off over time, and the same thing when you build a funnel — you break out the different people as they go and complete it. So they started, they got to this stage, and then you see conversion. Same thing with send money and login. All in all, I think it's still directionally really good data. Daily active usage by people is kind of interesting to see too.

**[54:01] Pablo Esparza Jimenez:** The other graph you were showing me on the previous page definitely tells me what kind of content I could create — educational content, precisely: because if I see a big drop-off on something, maybe instructions are not clear and we just need to explain it more.

**[54:17] Phill Lewis:** That's right. And I think that's critical too. The point that Ian was making at the beginning — what we want people to do after they've registered is a big thing, because the majority of people we see try to get money as one of the first things they do. It's gonna be really interesting to see how we could guide them to a place we might want them to go.

**[54:49] Pablo Esparza Jimenez:** Well, I have an idea. You mentioned [connecting] a bank account, which is something we want them to do. What if, in the account creation, before some time — it's in the first 24 hours — they get sent an email, but the first milestone they have to completely do is this and they get something in return. That's something you have to do: just add an account, add your bank account.

**[55:22] Pablo Esparza Jimenez:** And now we have to define what that something they get — do they get PayWise credits, whatever the "satoshi" — whatever that is. So that I have that clear and I could advertise it, not just on social media but also send an email blast to our current email list, because whoever doesn't have a bank account — for example, we could also actually get that. Oh, we don't have it, that's fine. I could just make the communication around "have you not connected your bank account? Do this and get extra [reward]. If you already did, just click here to request your [reward]."

**[56:01] Phill Lewis:** Yeah, exactly. Or if you complete a bill payment, you get that too. That's one of the things to tie it to rewards. I feel like that would really incentivize the behavior we want them to do, and then we tie it to things that we also make money from. So it's kind of [aligned] with the cost situation, in other words. Something to think about — especially if we get people to bring cash onto the platform and hold a balance over a certain amount for a period of time. I'm making that up, but it's things like that [that] could kind of help.

**[56:26] Phill Lewis:** Here you get to see things like registration completed, bill payment, cash-in completed, all of them from the agent, bill payment. I think there's a lot of interesting information on this page. The goal was just to not have to leave the page. You see some errors: check an Aircon balance, onboarding access errors — this is when they drop off the onboarding.

**[57:18] Phill Lewis:** Not only that — they drop off because they got an error. Even when the errors happen, we flag it, so we're checking on those things. Sometimes when people are trying to top up their wallet, there may have been a card error or whatever happening and it creates an error. So we catch that too. This is where the majority of errors are happening, so it's worth probably drilling into this a little bit more. And other areas — I'm not sure what that means, but it's worth a double-click.

**[57:50] Phill Lewis:** So, there we go — last 30 days and 72 registration-completed events. That aligns with what you see, right? So 77 [for you]. I think if I were to [set] a custom range just really quickly... it's not working because I should have seen a date slice. Anyway, I'm gonna check this in an hour — stop talking. It'd be great if it says 77, but it does not. 72.

**[59:52] Ian Alleyne:** Scaling — I think this will be a very similar aspect of that. And then this could talk into what Pablo is doing: from end to end we can have data that speaks to the effectiveness of what we're doing.

**[59:52] Phill Lewis:** Agreed. It could also trigger next best actions.

**[59:52] Ian Alleyne:** What we can do is examine this model that you have on your fingers, let's say on a weekly basis going forward, to make sure it aligns with the actual performance. Because this would be critical for us.

**[59:52] Phill Lewis:** I know 100%. You could see how this information would feed the markets and strategy — what areas you should really lean into to try and enhance the education pieces. Pablo's making [content] — we get this information sent to us in real time, so I know right away when there's a [system] failure. I also see the email come through, but if I didn't, I could see that here: someone was trying to take a selfie and it bombed out, or they tried multiple times to upload an ID card [in the] registration process. You see all of that in real time.

**[60:31] Pablo Esparza Jimenez:** To me, one thing I would try to solve: those people who get those errors — if you actually get our email notification, [it] would be nice for [marketing] to send them a message just acknowledging the error and apologizing, and either helping them manually if possible or asking them to try again in an hour or something.

**[61:01] Pablo Esparza Jimenez:** You said you get an email notification. Let's see if we could get that email notification to another inbox — copy to another receiving [address] we create. There is an email I created for the PayWise CRM; I used it to create the CRM account. We could use another one, and then I just connect that inbox, and whatever subject line I see that has that error, then I run automation to contact the customer, or alert me to contact the customer — whichever is more convenient or possible at this point.

**[61:40] Phill Lewis:** Well, Pablo, we could do all of that together. We could spin up those journeys where you say: whenever you detect this error, send me an email with this bits of information and to this account, and the bot just listening to know what to do. And I should be able to answer.

**[62:12] Pablo Esparza Jimenez:** All right, let me put in the request properly, then I'll take a note of this right now, request that information and tell you what bits of information I would need so that I could just follow up on that.

**[62:34] Ian Alleyne:** I think, guys, let's work on that bill payment issue, whereby persons are reaching the "ready to pay" [stage] but nothing happens. One suggestion I have: the payment method allows them to see wallet, card, debit card or credit card, but it's a drop-down box. Maybe they don't click on that to see the other options. So instead of having a drop-down box, maybe we can have some radio buttons where it is visible to them, so they can immediately see [they can] compare with a debit card or credit card, which I may have in my pocket. What I will do is I will just ask [Nathaniel] to make that change, because it's a superficial, high-level UI change. That's all it is.

**[64:12] Ian Alleyne:** What I'm saying is, in what I'm seeing in the data there — [they're] stuck between "ready to pay" — meaning the entering, the selecting, below entering information. What is going on on that page? We're not too sure what they may be doing on the page. One of the things that maybe [is] obscure to them [is] how to pay, because the default is wallet, and the default wallet will have no money in the wallet. So they don't have money to pay. They may not click on it to see what the other options are. So what I'm saying is to surface the other options — have radio buttons, [because] radio buttons will require that the options are visible to the user.

**[65:30] Sylvannas Berkeley:** Well, yeah, I have a suggestion. What about [ordering] the option — the first option being to add a card, and then if you do have...

**[65:36] Ian Alleyne:** No, it doesn't work like that. The flow is that you have to add a card [during] any transaction itself. The flow is not [starting with] a card. Which, I mean, is something we can consider, but the flow is that when you make the payment with the card, then you get the option to see if the card [is saved].

**[66:08] Phill Lewis:** Well, that's because of PII, right? Because of the PCI [DSS] regulations, that's why we have it done [that way]. Once the payment has been completed and we get [the return] from the payment processor, that's when you could save it.

**[66:37] Ian Alleyne:** Yeah, possibly. But in terms of surfacing the options — a way to show people, give them the option to choose. You're presenting them with a choice as opposed to obscuring the choice. We've got the choices there, but it takes maybe a bit more thinking to realize "let me click on this to see what other options there are." Whereas if we have radio buttons where you have to show all the options, they can see it immediately. So I would ask Nathaniel and ask him to make that change.

**[67:31] Phill Lewis:** So I just saw something very interesting, and I'm not discounting that it could be a bug, but it's in the chart here. I probably need to ask Elish this in a little bit more detail, but for the month of May, when someone opens the app and starts the registration process, there are 533 attempts at that in a month. I don't know if it's all unique; I probably have to work with them to get it more unique, but I find it very interesting that it's so large, even if it's half.

**[68:21] Ian Alleyne:** Okay, we could confirm that, Pablo. Can you tell how many downloads there were?

**[68:30] Pablo Esparza Jimenez:** I could check, probably on analytics.

**[68:35] Ian Alleyne:** Yeah, if you can check how many downloads there were.

**[68:39] Pablo Esparza Jimenez:** Let me check on Google Analytics and I'll come back to you. I should be able to check on [the] Play [Store].

**[68:47] Ian Alleyne:** And please go to [the App Store] too — the Apple Store.

**[68:48] Pablo Esparza Jimenez:** Absolutely. Let me check.

**[69:25] Phill Lewis:** Wow, there are 502 events for app installs in March.

**[69:42] Pablo Esparza Jimenez:** Oh, [it's] the wrong account. I remember you sent me the invitation to my other email, hold on.

**[71:41] Phill Lewis:** If you look at the chat that I pasted, it's still 502 of them, and then 533 attempts to do a registration. And look at what happens after that: it drops off significantly, by the time it comes to choose if you're a personal or business.

**[71:47] Pablo Esparza Jimenez:** I was able to find the downloads report for the Play store. For the App Store, I won't be able to login right now — I don't have an Apple device with me, and I need to get the code to an Apple ID [on an] Apple phone or device. But from here, for again, what I can see for the month of May alone: we had 160 install events, 90 uninstalls. So [net] 116 installs... Actually: 160 install events, 90 uninstalls.

**[72:22] Ian Alleyne:** How many uninstalled?

**[72:24] Pablo Esparza Jimenez:** [90] uninstalls. Let me see what else we see here. We have daily devices installed: 128. Daily devices... user installs: zero. Daily user installs — what is this? Android 5. Let me see. We have daily device installs and daily user installs. 215. 128 devices from 105 users.

**[73:19] Phill Lewis:** So, so, I am seeing — this is Android you're looking at.

**[73:23] Pablo Esparza Jimenez:** Yeah, exactly, that's only Android.

**[73:28] Phill Lewis:** I'm seeing 208 installs for me here, and for Android 208, and 287 for iOS.

**[73:44] Pablo Esparza Jimenez:** Maybe they're adding up. Look — install events plus daily user installs is 240; plus daily device installs is 368. So maybe something like that is going on — the two different install values are being added up.

**[74:10] Phill Lewis:** Those are the ones that are known. But even so, there's still a big drop-off, because even if my numbers [are] inflated — which is possible, because I don't know exactly what Elish is doing to trigger the event, maybe multiple types of things get the event called. But as it may be, I feel like we still have a huge leaky funnel, because people get the app and don't even finish registration.

**[79:41] Pablo Esparza Jimenez:** So that's 70 — that's the difference.

**[79:41] Phill Lewis:** Net. So if it's 17... let's assume it's about right, but that assumes there's nobody on iOS. 50% drop-off from installing to [using] the app.

**[79:41] Pablo Esparza Jimenez:** I'll download this information manually and put it in somewhere.

**[79:41] Phill Lewis:** All right, Pablo, as they say in Star Trek, make it so. This is really interesting.

**[79:41] Pablo Esparza Jimenez:** It's truly [interesting]. Thanks to you, honestly, I've been very focused on data in the past year. Without data you can't really make decisions.

**[79:41] Phill Lewis:** You really can't.

**[79:41] Pablo Esparza Jimenez:** I'll see you guys next month, Wednesday. Thank you guys for inviting me.

**[79:41] Pablo Esparza Jimenez:** I'll consolidate all of this info in that one document report that I showed you that I had for the other client. So we have a base — not any integration but taken from actual exports of Google Play, Google Search Console, social media, etc. And then we could do integrations and compare with the real data. And once our integrations are correct with the real data, then we can trust that integration and continue with it. This is for this year — August.

**[79:41] Phill Lewis:** But I feel like we should be doing this kind of thing. It's all good.

**[79:41] Pablo Esparza Jimenez:** I'll start with the Play Store, which I have really easy access to right here from my Chrome. I'll put all of this up, organize it immediately, I'll use [the] cloud for that, and we could have a place where we know it's the true values. And then that's something that you can use to compare data as you pull it.

**[79:41] Ian Alleyne:** But also, Pablo, if you can do [it] for iOS...

**[79:41] Phill Lewis:** Even if it's 50/50 — because I think in China it's a little bit more iOS than Android. Even if it's 50/50, we still have [drop-off].

**[79:41] Pablo Esparza Jimenez:** And from those 160, where is the right here? It would be this one, that's 90, right here down below.

**[79:41] Phill Lewis:** Yeah. Pablo, thanks, and message me when you have some time to come up with [this].

**[79:41] Pablo Esparza Jimenez:** One thing we can trust is these two values here: 160 install events, taking at least for Android directly from Google Play itself.

**[79:41] Phill Lewis:** [We have] events that we have, and like some kind of English definition, and you could tell me which ones you want fired. You could create some rules — just tell me some rules and I'll put them in Play, and send emails when they happen. When they happen, we could send an email, and then your bot could pick it up. We obviously don't want to overkill people, so we should probably try it out first.

**[79:41] Pablo Esparza Jimenez:** Correct.

**[79:41] Phill Lewis:** Especially if it's our own retargeting, I would think that's good. Carefully [done] retargeting — we can do that here.

**[79:41] Phill Lewis:** All of this to say, Pablo, I feel like the data is [the] root — what we're looking at, what we're trying to verify. But as a root for the strategy that we should [pursue].

**[79:41] Phill Lewis:** You know what? What events you want triggered, I could send you the [definition].

**[79:43] Pablo Esparza Jimenez:** Talk soon. Bye, take care.

> [Note: the last ~15 minutes of the source timestamps collapse to `[79:41]` in the export; the speaker order above preserves the source sequence.]

# Action Items

| # | Acción | Responsable | Lado | Prioridad | Fecha objetivo |
|---|---|---|---|---|---|
| 1 | Armar las plantillas de reporte mensual de PayWise y consolidar todo en un documento (instalaciones/descargas de Play Store + App Store, Search Console, social y data de plataforma), actualizado cada mes | Pablo Esparza Jimenez | Coordenadas | Alta | — |
| 2 | Construir el dataset base de "valores reales" desde exports (Play Store, App Store, Search Console, social) y luego agregar integraciones y reconciliarlas contra ese baseline | Pablo Esparza Jimenez | Coordenadas | Alta | — |
| 3 | Descargar/procesar las métricas de descargas e instalaciones de Android (Play Store) para el reporte base | Pablo Esparza Jimenez | Coordenadas | Alta | — |
| 4 | Incluir las métricas de instalación de iOS (App Store) en el reporte base | Pablo Esparza Jimenez | Coordenadas | Media | — |
| 5 | Desarrollar una campaña de email/lifecycle a la base actual: recordatorios de pago de facturas alineados al ciclo + how-to guides | Pablo Esparza Jimenez | Coordenadas | Media | — |
| 6 | Especificar formalmente la solicitud de notificaciones de error (qué bits de información se necesitan) y montar el seguimiento/automatización al usuario que erró | Pablo Esparza Jimenez | Coordenadas | Media | — |
| 7 | TikTok: coordinar la creación de la cuenta desde Trinidad y, una vez creada, conectarla en la plataforma para postear | Pablo Esparza Jimenez + Sylvannas Berkeley | Coordenadas + Cliente | Media | — |
| 8 | Analizar el volumen y tipo de conversaciones/mensajes en social (quejas, preguntas, soporte, API) para informar el contenido | Pablo Esparza Jimenez | Coordenadas | Baja | — |
| 9 | Definir la recompensa/incentivo de activación (créditos PayWise u otro) para poder anunciarla | Phill Lewis / PayWise | Cliente | Media | — |
| 10 | Cambiar el selector de método de pago de dropdown a radio buttons (visibilidad de wallet/tarjeta) | Ian Alleyne → Nathaniel | Cliente | Alta | — |
| 11 | Investigar discrepancias de métricas de bill payment ("Ready to Pay" vs completados) | Ian Alleyne, Phill Lewis | Cliente | Alta | — |
| 12 | Arreglar la ingesta de eventos en la base de datos y el doble conteo (business/personal) | Phill Lewis (con Elish) | Cliente | Alta | — |
| 13 | Mejorar el embudo de registro y reducir el drop-off post-registro | PayWise | Cliente | Alta | — |
| 14 | Enviar a Pablo las fechas de creación de cuentas del último año | Ian Alleyne | Cliente | Media | — |
| 15 | Enviar a Pablo la lista de emails de usuarios (con flag business/personal y teléfono) | Ian Alleyne | Cliente | Media | — |
| 16 | Proveer acceso/credenciales de iOS (Apple ID / dispositivo) para poder jalar las métricas | PayWise | Cliente | Media | — |
| 17 | Crear la cuenta de TikTok desde Trinidad (geo-bloqueo de IP) | Sylvannas Berkeley / PayWise | Cliente | Media | — |