13 Weeks to Black Friday: The Real Timeline for Launching a Configurator Before Q4

Black Friday 2026 falls on 27 November. Most furniture e-commerce teams freeze their production code somewhere in the first week of November, which means the actual deadline for anything touching the product page is closer to 1 November than to Black Friday itself. Counted from the first Monday of August, that is 13 weeks.

Thirteen weeks is enough. It is not comfortably enough, and it stops being enough somewhere around the middle of September. That is the entire argument for why autumn planning decides whether a configurator contributes to this year's peak season or next year's - and it is worth spelling out with real numbers rather than the usual "get ahead of Q4" framing.

We have shipped 40+ configurators across 9 markets, and the pattern in the delivery data is consistent: the projects that miss their target launch date rarely miss it because the build took longer than estimated. They miss it because the inputs arrived late. This article works backwards from the November deadline, breaks down what the standard 8-to-12-week estimate actually contains, and identifies the three variables that decide where inside that range a project lands. It also makes the case for not launching before Q4 at all, if your conditions look a certain way.

The Calendar Nobody Draws

Vendor timelines are usually quoted as a duration - "live in 8 to 12 weeks" - which is a genuinely useful number and also the wrong unit for the decision you are making in August. The useful unit is a date, counted backwards from the one that is fixed.

Working back from a 1 November code freeze:

• Start in early August: 13 weeks. A standard scope lands with two to three weeks of buffer for the things that always take longer than planned.
• Start in early September: 9 weeks. A standard scope lands only if 3D assets are already in hand and the integration is a well-trodden path. No buffer.
• Start in early October: 4 weeks. Not a launch window. This is a window for scoping the project properly and launching in January against a clean quarter.

The asymmetry matters more than the arithmetic. Starting a month earlier does not make the project a month better - it makes the difference between a launch that absorbs one bad surprise and a launch that absorbs none. In practice, every configurator project has at least one bad surprise. Product data turns out to be inconsistent across variants. A material that looked straightforward in the brief needs a scan. Someone discovers two weeks in that "seat height" means floor-to-top in the spec sheet and seat-to-top in the CAD files, and the measurement convention has to be agreed retroactively across an entire catalog.

None of these are exotic. They are the normal texture of a configurator project, and the buffer is what converts them from a missed launch into a Tuesday.

What "8 to 12 Weeks" Actually Contains

The estimate is not a single block of development. It is five workstreams that overlap, and understanding the overlap is what lets you see where your own decisions sit on the critical path.

Our internal delivery framework runs roughly like this for a standard first-family scope:

• Week 1-2, project setup and design direction. Brand guidelines, UI direction, environment. Mostly our work, but it stalls immediately if brand assets are not available.
• Week 2-6, 3D models. Review, correction, and optimisation of the geometry. This is the workstream with the widest variance, for reasons covered below.
• Week 3-6, product data and integration. SKU structure, variant mapping, pricing logic, cart handoff. Runs in parallel with 3D, and is the most common place for a project to quietly stall.
• Week 4-6, platform features. Save and share, AR, PDF export, analytics. These ship as platform capabilities rather than custom builds, which is why they compress into a two-week band instead of expanding to fill the project.
• Week 5-8, testing and launch. QA across devices, edge-case configurations, load handling, and the embed on the live product page.

The parallelism is the point. Weeks 2 through 6 have three workstreams running simultaneously, which is what makes an 8-week delivery possible at all. It is also what makes the timeline fragile: a delay in any one of the three parallel tracks does not compress the others, it extends the whole project. A two-week wait on product data does not get absorbed by the 3D team working faster.

The Three Variables That Decide Where You Land

Between 8 weeks and 20 weeks sits a set of conditions that are almost entirely knowable in advance. If you want a realistic read on whether a Q4 launch is available to you, these three questions will tell you more than any vendor estimate.

1. 3D Asset Readiness

This is the single largest driver of timeline variance, and it is measurable before the project starts. Across our deployments, the review-and-preparation effort per product family splits roughly three ways: models supplied by an external 3D asset provider need review only, in the range of 5 hours. Brand-supplied GLB files typically need review and adjustment in the 15-20 hour range. Models produced from scratch run from 15 to 50 hours per family depending on complexity.

That is a tenfold spread on the same deliverable, and it is not a quality judgement on the files. It is a statement about what condition they arrive in. The technical target is specific: geometry under 100K vertices, ideally under 10K, textures capped at 512x512 pixels, and optimised files landing under 5MB with 1MB as the goal. Files built for a rendering pipeline or a product catalog are typically an order of magnitude heavier than this, because they were never built to load on a phone over a mobile connection.

The useful diagnostic in August is not "do we have 3D files" but "do we have 3D files that meet those numbers, and if not, who is doing the optimisation and when did we ask them." Upholstered products deserve particular attention, because foam adds volume that CAD geometry frequently does not carry - the resulting models look thin in the seat or gapped at the joins, and correcting that requires reference photography or physical measurements from the brand.

2. Integration Depth

A configurator that writes to a standard e-commerce cart is a well-trodden path. Integration work in the range of 4-5 hours is typical, rising to around 25 hours for catalogs with 500 or more variants. A configurator that has to reconcile against a PIM, an ERP, and a separate order management system is a different project, and our own estimates for that shape have moved from 8-10 weeks to 16-20 weeks on the strength of experience rather than optimism.

If your product data lives in more than two systems and none of them is the agreed source of truth, that question needs answering before the timeline means anything. It is also the cheapest question to answer in August, because the answer costs a meeting rather than a sprint.

3. Feedback Discipline

This is the variable teams underestimate most consistently, and the one they control most directly. Across our projects, feedback rounds per product family average three to four. The cost profile is steeply non-linear: a single clean review round costs roughly an hour of correction work, while a project that runs to eight or more rounds accumulates around 30 hours on the same family.

The mechanism is not that the extra rounds find more problems. It is that scattered feedback arriving from three stakeholders across two weeks produces corrections that contradict each other, and the resulting rework is pure loss. The pattern that compresses timelines is unglamorous: one named decision-maker, consolidated feedback, and a review cadence agreed before the first model is delivered.

In one deployment, a data-setup question resolved through 26 back-and-forth comments over three months, which pushed a straightforward table-and-chair scope past its intended launch by a full quarter. Nothing about the build was hard. The decision loop was.

When Not to Launch Before Q4

The honest counter-case deserves more than a footnote, because for a meaningful share of brands it is the right answer.

Do not push for a November launch if your peak-season traffic is the only traffic you have that is worth learning from. A configurator launched three weeks before Black Friday goes live into your least forgiving traffic, with no baseline to compare against and no window to iterate. You will have a conversion number, and you will not know whether it reflects the tool or the season. Launching in January into ordinary traffic gives you a clean read and eleven months to act on it.

Do not push for it if the launch requires shortcuts in the product data. Configurations that produce orders your fulfilment team cannot actually build are worse than no configurator at all, and peak season is the worst possible time to discover a variant-mapping error at volume.

Do not push for it if your e-commerce team is already carrying a Q4 roadmap. A configurator launch needs their attention for the embed, the tracking, and the inevitable first-week adjustments. Competing with a payment provider migration for the same two engineers is a good way to get a launch that technically happened.

The version of the autumn window that consistently works is narrower and more useful than "launch before Black Friday": start the project in August, get the first product family live in October, and use November as a real-traffic stress test with the catalog-wide rollout scheduled for Q1. That sequencing gets you peak-season data without betting peak-season revenue on a launch week.

What August Buys You That September Cannot

If the decision is genuinely open, the work that pays off most in August is not signing anything. It is removing the unknowns that turn a 9-week project into a 16-week one.

Four things, none of which require a vendor:

• Audit the 3D files you actually have against the vertex, texture, and file-size targets above. The answer determines several weeks of the timeline and takes an afternoon to establish.
• Name the source of truth for product data and confirm that variant structure is consistent across the family you would launch with. Inconsistency here is normal and fixable, but only if it is found in August.
• Agree the measurement conventions in writing. Seat height, depth, overall width. This is a fifteen-minute conversation that prevents a catalog-wide correction.
• Name one decision-maker for design feedback. Not a committee, not a shared inbox. The difference between three rounds and eight is usually this single choice.

Every one of these is free, none of them commits you to a project, and all four are the difference between the 8-week end of the estimate and the 20-week end. A brand that walks into September with those four answers in hand has a genuine Q4 option. A brand that walks into September planning to find out has a Q1 project it has not admitted to yet.

The seasonal argument for autumn is real, but it is not really about autumn. It is that configurator timelines are dominated by input readiness rather than build effort, and August is simply the last month in which the inputs can be fixed without the calendar doing it for you.