Digital District Partners

// 01 · Boundaries drawn by capability, not hours

Each tier is bounded by something the tier below structurally cannot do. Not by page count and not by hours. Triage is repair only: one critical fix, a plain-English report on what broke, keys and documentation handed back. Foundation is the first tier where a stranger can buy something without phoning the store. Growth Engine exists because a standard checkout physically cannot take a deposit on custom work, so the boundary is deposits and part-payments. Flagship is the tier that includes owned infrastructure.

What is out is stated as loudly as what is in. Every tier carries an explicit excludes list, and the site publishes a standing "what you will not get from us" block: no team, no paid ad management, no brand identity work, no photography or copywriting, no around-the-clock coverage. Scope control moved out of the negotiation and into the price sheet, where it costs nothing to enforce.

The published comparison against hiring an agency includes the two rows where being one person loses. Publishing your own weaknesses next to your price is a product decision, not a modesty one.

// 02 · The pricing reversal

The original policy published only the two cheapest tiers and scoped everything else on a call. It was wrong, and what made it wrong was watching a jeweler compare us against the vertical incumbent, who publishes a $5,000 to $15,000 ladder. Against that, our two visible numbers sat next to a case study describing a multi-million-dollar catalog, and the gap did the talking.

A hidden price cannot be judged. "We will scope it on the call" is the exact experience this buyer has been burned by before. So the whole ladder went public at real figures.

The lesson generalises past pricing: an unquotable number is not a negotiating position. It is missing information, and the buyer fills the gap with the worst assumption available.

A quieter version of the same lesson sits in the media platform. The recurring fee is for operating the platform and is never a markup on hosting, because Cloudflare bills the client directly at cost. A markup on a twelve-dollar bill invites the question "why am I paying you to pass through twelve dollars." A fee to run the thing is defensible on its own terms. That is a pricing model chosen to survive the client’s most obvious question rather than to maximise the line item.

// 03 · One array behind four promises

The delivery checklist is the spine of the whole business. Twelve binary checks live in a typed data file, each tagged with the tiers that deliver it. The marketing claim, the legal guarantee, the delivery QA gate, and the scheduled follow-up emails all resolve to that one array, so they cannot drift apart. The file carries a comment noting the numbering is load-bearing because the emails cite checks by number.

The guarantee is written against it: if the build does not pass every check marked for its tier, on the client’s own phone, at handoff, the client does not pay the final balance. Objective and invokable instead of vague satisfaction language.

It was deliberately narrowed once the checklist went public. The original promised every check, and publishing the grid made that unfulfillable, because one check is product photography that no tier delivers and another is a paid add-on. Scoping a guarantee against your own published exclusions, before a client finds the gap, is the discipline in one artifact.

Scope creep is handled by four mechanisms rather than by argument: the per-tier excludes list, a 100% upgrade credit that turns "can you also do X" into an upsell with no sunk-cost fight, month-to-month terms with full client ownership of accounts and code, and the guarantee itself.

// 04 · Building the exit ramp on purpose

The media platform is sold as a product with its own price: a build fee, a monthly operating fee, and inclusion in the top tier. The engineering behind it is a separate case study. What belongs here is the tenancy decision.

It is deliberately not multi-tenant in the usual sense. Each client gets their own deployment, on their own subdomain, on their own Cloudflare account, with the storage, the domain, and the billing relationship in their name. Cloudflare invoices them directly at cost, typically under twenty dollars a month, and I never touch that invoice.

The competitors this sells against bill by visitor on their own domain. That means a product doing well costs the jeweler more than one nobody sees, and it means their media is hostage. I cannot press that wound while doing the same thing, so the exit ramp is built in. The reasoning I keep coming back to is that the exit ramp being there is the reason clients never take it.

It works economically because of an architecture choice: frames are pre-baked at fixed sizes and served as files, so the CDN hands back an object instead of generating one. No transform is ever billed, and the bill stops scaling with traffic.