Architecture call · decision sheet
Philip Engle and Travis Marshall. Built from your Functional Wireframe v1.1.
The call
Thursday, August 13
2:30 PM Travis · 3:30 PM Philip.
Thank you for doing this in writing first.
Trading documents instead of talking our way to clarity has already saved us a whole call, and it means Thursday can be decisions rather than discovery. The faster we close this part out, the faster we get to actually building the thing.
Your section 7 listed eight items to lock. Two of them are really one conversation, so they are one decision here, which makes seven. I added three: how the marketing engine actually gets built, one flag on the brand, and the asset list. Ten total.
Tap any decision to open it, then tap the option you want. It records as we talk, and at the end one tap emails the locked set to both of us so nothing gets re-litigated next week.
Read this first. It is why three of the ten decisions exist at all.
What those numbers actually mean
A conventional shop would tell you this does not fit. Here is why it does, and where the real edge is.
Your 9.1 puts Develop through December and January is Deploy, so the building window is the twenty weeks from now to New Year, not to the end of January. My build days are half of Tuesdays and all of Thursdays, which is 30 on paper.
I am also still an exec pastor, and November and December are Thanksgiving, Advent and Christmas services at my church. So the honest number is not 30. It is about 25 actual build days for 21 items. I would rather show you that arithmetic now than discover it together in December.
Two things change the answer.
First, the items are not equal weight. Your own build-status tags already sort them:
| Weight | Count | What it means |
|---|---|---|
| PORT-IN | 4 | Already works. Lead-gen assessment, team health platform, scalable org builder, lead intelligence filtering. Days each, not weeks. |
| PARTIAL | 4 | A real template or draft exists. Team leader breakdown, master playbook, metrics dashboard, resource assets. Half-builds. |
| NEW | 13 | Net-new. This is where the whole calendar goes. |
Second, I do not build the way a conventional shop does. I built ChurchOps.AI in five months while learning the stack, and I work with agent orchestration now, which is a genuine multiple on output rather than a marketing line. That is why I said yes to this scope instead of negotiating it down.
But a multiplier applied to a bad sequence still produces a bad January. So:
The one thing I need from you
Everything in your wireframe is in scope and stays in scope. I am not asking to cut anything and I am not asking for more money. What I am asking is that we set the build order together in Discover, on the wireframe, before anything is coded.
Then if something is still unfinished at rollout, it is still mine to finish inside this $15,000, on a date we set together in November. Phase 2 is for the new things you dream up afterward, not for anything on today's list. Nothing here becomes billable later.
Where the load actually sits
One item is far heavier than the rest, and naming it is the most useful thing in this document.
| Item | Est. days | Share of the 25 |
|---|---|---|
| Marketing engine at full spec | 30 to 46 | 120 to 184%. More than the whole window, on its own. |
| Marketing engine, trimmed first version | ~16 | ~64% |
| Everything else, 20 items | ~9 left | What remains after the trimmed version |
That is not an argument against building it. It is why decisions 5 and 6 exist, and why shaping that one item is worth more than shaping any other five combined.
Discover cannot finish without them, and Design starts in September.
Where the platform lives
One DNS record, and the address every church will see.
Wireframe 6What is actually true
Functionality is identical in all four, so this is purely an address decision. And Option A is not a dead end. Moving to B later is the rebuild your own doc describes, but the data, the logins and every tool come with you. Only the shell changes.
Part one, the base address
A door beside Drü and Renée's site. One DNS record is the whole ask of them. Fastest to stand up, still reads as focus412, and neither property can take the other down.
B rebuilds the portal and public site as one property later. C stands the platform on its own domain. Both are real options. If you want either, I will come back with what each costs in days before Design starts rather than guess on the call.
Part two, optional add-on
milestone.focus412.com, cwc.focus412.com. Sits on top of whichever base you pick. Strong invitation, and worth automating past about thirty churches. Pick a base address first.
The full tool list, and what governs the order
Your six additions are accepted. This is only about sequence.
Wireframe 7 item 2 · 9.1What is actually true
You were right that my list was partial. The six you added are the actual Essentials methodology, not extras: Culture ID Builder, Team Leader Breakdown, Meeting Planner with Leader Night Rhythm, Leadership Pipeline Visualizer, Health Metrics Scorecard, and the Weekend Experience Evaluation. They are in. That is not a question on this call.
Twenty-one items and roughly 25 build days means the order matters more than the list does.
Nothing is cut. We sequence on the wireframe together, biggest leverage first, and you approve that order before a line is coded. You always know what is landing when, and anything still open at rollout is finished inside this build, not billed again.
Same commitment, but you mark what has to be working by the January rollout versus what can land in February. Gives you certainty on the rollout story. Costs you the option to change your mind in October.
The role model
Confirming Super Admin sits above Coach. Thirty seconds.
Wireframe 5.1 · 7 item 5What is actually true
Your four-role model is better than the three levels in my scope of work. Three of your four already exist in the Team Health Assessment platform, so most of this ports in. Coach is the one we add, and it is the one that makes the model work at your size.
Coach scope is already settled in your own 5.1 table, assigned churches only. The only thing your section 7 flags as open is Super Admin sitting above Coach, so this really is a yes or no.
Exactly as your 5.1 table specifies. Super Admin sees all churches and assigns them to Coaches. Fixed vocabulary from here forward, in the code and in every document.
Simpler to build and probably fine at your size today. Becomes awkward the first time you add a coach who should not see a sensitive engagement.
The two instruments of record
Your section 7 asks for both. Two picks in here.
Wireframe 2.2 · 3.4 · 7 item 3What is actually true
Whatever we wire becomes the thing every future benchmark compares against. After Develop starts, changing a question means migrating everyone's historical answers. This is the cheap moment.
Part one, the lead generator
Your own 2.2 flags the conflict: the Drive version has three per category, an earlier spec had four. Twelve is what is live today, and shorter converts better at the top of a funnel where volume is the whole point.
A better diagnostic and a stronger readout. Costs completion rate.
Part two, the team health assessment
Twenty dimensions across the four Essentials areas, each scored twice, Performance against Importance, gap revealing where attention matters. That is where your real benchmark data already lives.
One flag from my side, not from your doc. Going through the sheets in your assessment history I found an earlier, shorter version still in circulation, and a few questions whose wording drifts between sheets. Worth ten careful minutes so the benchmark is wired to one thing. I will map the older responses onto the current set so churches with history keep their year-over-year comparison.
If anything about the instrument has been bothering you, now is the only cheap moment. Costs us a week of Discover and it is worth it if the answer is yes.
All four are on your section 7 list. I have ordered them by what blocks code first, not by importance. Email works for any of them if the clock runs short.
The marketing engine, and one honest correction
Your 4.3 settled that the engine is ours. This is how it gets built, not whether.
Wireframe 4.3 · numbers in the appendix belowWhy we are doing this
You own the surface. One login, assessment answers and email behavior in one queryable place, your brand on every screen, and no per-contact tax as the list grows. That is the reason, and it is a good one.
I owe you a correction on something I said. When I showed you my Monday brief and told you it was all native, I meant what you would mean: no Mailchimp, no third-party dashboard, the whole surface mine. That is exactly what you get here.
What no small sender runs themselves is the last hop, the actual SMTP handoff. Cloudflare blocks outbound port 25 from the platform we are building on, and even outside it you would face 30 to 60 days of IP warm-up before your first campaign could go out. That is a fact about the plumbing, not a reduction in what you asked for.
And one thing you should hear from me rather than work out later: the money case for leaving Mailchimp is weaker than it looks. You save $480 to $2,940 a year depending on list size. If cost were the reason, the numbers would say stay. Owning the surface is the reason. I would make that argument and not the other one.
Part one, who carries the last hop
Everything in your 4.3 lives in your platform and carries your brand. Behind it, delivery rides on Resend, which you never see. It has the cleanest Cloudflare integration of the ones I tested and already handles unsubscribes, bounces, and open and click webhooks. Deliverability stops being ours to get wrong at the infrastructure layer. Content and list hygiene are still on us and I will hold those.
Postmark, Amazon SES and Mailgun are all real answers with different tradeoffs on price, support and compliance handling. If you would rather see the comparison than take my pick, I will bring it before Design.
Part two, what version one does
You still see every open and click, exactly as your 4.3 asks. What version one will not do is branch a sequence based on an open, because Apple Mail Privacy Protection has made that signal unreliable enough that branching on it costs more than it tells you. Sequences run on fixed intervals and exit on the things that are actually true: unsubscribed, booked a call, became a partner. Branching lands later if you still want it.
Everything in 4.3 with nothing held back. It is the single most expensive thing on the list at roughly double the trimmed build, on a signal Apple has already degraded. I will build it if you want it, and I would rather spend those days on two more toolkit tools.
Email authoring surface
Builder in the platform, or import the HTML Drü and Renée already make.
Wireframe 4.3, marked CONFIRMWhat is actually true
A real drag-and-drop email builder is one of the most expensive UI surfaces in software. Drü and Renée have already produced a mock-up email in HTML. Importing their work costs a fraction and produces better-looking email, because a designer made it.
Drü and Renée own how it looks. The platform handles personalization, the book-a-call block, sending and reporting. Days instead of weeks, and it puts the design where the designers are.
Heading, text, image, button, divider. Anyone on the team can build an email without touching code. Materially more build, and it will never look as good as what a designer hands you.
Review before release
Which artifacts pass through your hands before a church sees them.
Wireframe 3.4 · 5.3 · 7 item 6What is actually true
This is the best idea in your wireframe and I had missed it entirely. It is also real architecture: every artifact needs a draft, review, release state, and the church side has to honor it everywhere. Cheap if it goes into the spine now, expensive to retrofit onto nine tools later. The only question is how wide it goes.
Assessment results, Weekend Experience Evaluation, coach reports, anything we add later. Built into the data model on day one so every future tool inherits it free. The church only ever sees the released version.
The two your 3.4 names. Narrower, slightly less to build, and the pattern still exists to extend later.
Coach-configurable intake, and the baseline set
What toggles per engagement tier, and exactly what we capture at the start.
Wireframe 3.3 · 5.4 · 7 items 7 and 8What is actually true
Your section 7 lists these separately but they are one conversation, so I have merged them.
The cost of getting it wrong is worth knowing: baseline data not captured at intake cannot be recovered later. If a metric is not collected in month one, the before-and-after for that church does not exist, ever. Over-capture here. Storage is free and regret is not.
I have your phase workbooks, your session assignments and your scorecards now. I will propose the intake toggle matrix by engagement tier and the exact baseline metric and document set, built out of what your own process already asks churches for. You spend fifteen minutes correcting it rather than an evening writing it. That is the version that actually gets done before Design.
You know which intake pieces an On-Demand church gets versus an Onsite one better than any document does, and you know which numbers make a case study land. More accurate. Costs you real time.
Brand palette
Your 1.5 drops two colors your 2021 guide carries, and adds a typeface it does not.
Wireframe 1.5What is actually true
We are following your lead and this page runs on your gold and charcoal. The warm paper tone is the one liberty I took, and I will drop it if you want it pure white.
Flagging two things only because I would rather ask than quietly correct you. Your 2021 brand guide also sanctions navy #003b5c and green #6b9560 as secondaries, which a platform full of dashboards, locked and unlocked states and progress indicators genuinely wants. And it names Roboto as the Gotham substitute, with Montserrat not appearing in the guide at all.
Black, white, gold, charcoal. I work within it and find other ways to signal state.
Marketing surfaces stay on the four. Dashboards get the two secondaries from your own guide for success, warning and progress. Nothing off-brand, since both are already yours.
The assets, and which versions are current
I already have a lot of this. The real question is whether it is the version you would want me building from.
Wireframe 8What is actually true
Between your shared folder and things you have sent me over the years, I am not starting from nothing. The phase workbooks, the session assignments, the scorecards, the sample leader expectations and the metrics snapshots are all here, and they are the methodology I needed to design against.
What I cannot tell from the outside is which of it is still current. Some of what I have is from earlier engagements, and a tool built on a superseded version of your material is worse than no tool. So the useful thing on Thursday is not you sending files, it is me showing you what I already have and you saying current or superseded.
Your ninth section 8 item, walkthrough access to the churchops.ai scheduler, is on my side. I will run you through it in Discover.
I have a version. Tick if it is current.
I do not have these at all
Ten minutes at the end. I show you each thing I have, you say current or superseded, and you only have to go find the ones that are genuinely missing. Much faster than either of us describing files in an email.
Cleanest guarantee that I am building on current material, and no ambiguity at all. Costs you an afternoon of gathering.
Backing for decision 6. Vendor pricing as published on August 11, 2026.
What Mailchimp costs, and what you would save
The saving is real but smaller than it feels.
Mailchimp bills on contacts, not sends, and unsubscribed contacts still count against the total. Sends were never your constraint. You also need Standard rather than Essentials, because Essentials caps automations at four steps with no branching, and your 4.3 asks for a four to five email journey, which sits right on that ceiling.
Resend prices the other way, on emails sent. The comparison below assumes roughly five emails per contact per month.
| Your list | Mailchimp Standard | Resend, at ~5 sends each | Saved per year |
|---|---|---|---|
| 2,500 contacts | $60/mo | $20/mo | $480 |
| 10,000 contacts | $135/mo | $20/mo | $1,380 |
| 25,000 contacts | $310/mo | $65/mo | $2,940 |
Add Thinkific at $125 a month and Calendly on top, and the total subscription picture improves a lot. But on email alone, the honest read is that this is a decision about owning your own surface, not a cost decision.
Why nobody at your size runs the last hop themselves
A platform limit and a calendar problem, not a preference.
The part worth saying out loud
If focus412's sending reputation ever gets burned, it does not only break marketing. Your proposals, your coaching emails and your calendar invites stop landing too. Recovery takes months. A vendor sending from a separate subdomain keeps a marketing mistake away from the domain your business runs on, and that separation is most of what the $20 buys.
What we build either way
A vendor covers exactly one row of this. One.
This is the part that is easy to get wrong, so it is worth being precise. Choosing a vendor is not buying a marketing platform. It is buying the delivery pipe. Everything you actually see and touch is still ours to build:
| Capability | Who owns it |
|---|---|
| Contact store, segmenting on assessment answers | Us. That data is yours and lives in your platform. |
| Sequence engine, exit on booked call | Mostly us. Only your database knows a call got booked. |
| The composer and the campaign screens | Us. No vendor gives your team a usable authoring surface. |
| Campaign reporting the way you want to see it | Us. |
| Unsubscribe enforcement and compliance | Us, with vendor support. |
| SMTP, IP reputation, bounce and complaint plumbing | Them. This one row. |
So the build in front of us is nearly identical either way. The only question is whether we also run mail infrastructure, and there is no version of this project where that is the best use of a day.
Updates live as you tap. One button sends it.
Nothing is saved until you send it. If you close this page, the marks go with it.