The product is configured. Is the business problem solved?

G2G judgment. Configurable commerce has become sophisticated at turning difficult customer decisions into technically valid orders. But the sale is only one checkpoint in a longer economic process: unmet demand may be lost before it, production errors and channel costs may accumulate around it, and support, installation or returns costs can appear afterward. These costs fall differently on customers, retailers and manufacturers. Existing competitors already address parts of the problem, so the opening is not an unoccupied “after-sale” category. The investigation is which meaningful economic disconnects remain, for which buyer, and who could solve them with a repeatable advantage.

1 · Start with the economic problem, not the software category

A customer wants a fitted blind, cabinetry, staircase renovation, a deck, or a complete kitchen. Their immediate problem is not choosing from a large number of variants. It is committing money to a decision that must work in the physical world. Correct size, compatible components, finish, price, delivery, and installation all matter. They may need advice and assurance, not simply a shorter path to checkout.

A manufacturer needs a technically valid, producible order and wants partners to represent its product correctly. A retailer wants conversion, sales density, customer trust and positive contribution after returns and support. A designer or installer needs specifications and revisions that survive the handoff. The software provider needs the same rules and integrations to be reusable rather than rebuilt for each client.

These jobs overlap but are not interchangeable. The economic consequences of getting one wrong are visible in waste, wrong orders, rework, missed purchases and slower sales. Houzz's 2026 homeowner survey reports that 37% of renovators who set budgets exceeded them and 91% hired at least one professional. Some overruns reflected deliberate upgrades or expanded scope. These findings indicate project complexity; they do not measure the addressable demand for a configurator or prove that software prevents overruns. ¹²

Decision owner	Problem that creates demand	What they can pay for	Outcome that actually matters
Shopper / homeowner	“Can I choose something that will fit and work?”	Indirectly through purchases / paid planning	Fewer wrong choices; completed, workable outcome
Retailer / merchant	“Can we sell complex goods profitably without unlimited specialist labor?”	Commerce / category / digital budget	Contribution per served customer and project; rework and assistance cost
Manufacturer / brand	“How do we sell correctly through many retail partners?”	Channel / sales / commercial operations budget	Incremental distribution, sell-through, mix, order quality and time-to-market
Dealer / installer	“Can the design and specification be trusted on site?”	Software / service fee	Accurate handoff, fewer callbacks, faster quoting
Configurable-commerce vendor	“Can we replicate deployments?”	Recurring platform fees and services	Incremental recurring contribution per connection, product and customer

Category definition for this investigation: Systems that turn customer requirements into an accurate, visualised, priced and fulfillable purchase of complex or configurable goods. They may connect a merchant, a manufacturer, a dealer and an installer. The scope includes product configuration, guided selling, design-to-order and relevant project-planning systems when they materially affect what a buyer can confidently purchase. Adjacent tools matter where they solve the same commercial problem; this is not a claim that every quoting, catalogue or construction application belongs to one competitive market.

2 · Different commercial jobs created different approaches

The capabilities developed in overlapping waves in response to distinct commercial jobs. An industrial manufacturer needed production feasibility; an online retailer wanted customer confidence; a kitchen showroom needed a room plan; a distributor wanted consistent rules across dealers. It would be misleading to say that one category universally went from product selection to project completion.

Commercial pressure	Response that developed	Inherited success measure	Persistent limitation
Bespoke and engineer-to-order products created quoting and specification errors	Rules-driven configuration; configure, price, quote (CPQ); links to computer-aided design (CAD) and bills of materials (BOM)	Quote speed, error avoidance, sales productivity	Correct product does not necessarily mean correct real-world project
Ecommerce made complex choices harder without a salesperson	Consumer-facing configurators, guided selling, dynamic prices	Product-page conversion, completion, cart value	The journey can be polished but still leave unresolved fit or installation doubts
Furniture and kitchens required spatial context	3D visualisation, AR, room planners, professional design tools	Confidence, visual engagement, room sales	Attractive design may diverge from usable supplier, installation or delivery data
Brands sold through many merchants, dealers and channels	Central rules, dealer portals, headless configuration, system connectors	Faster rollout and consistent orders	Every channel may still require bespoke integration and ongoing maintenance
Project spending was scattered across categories and stages	Project lists, takeoffs, saved plans and service networks	Share of wallet, quote speed, project value	Project coordination, responsibility and execution cost may dwarf interface gains
AI made intent capture and ongoing optimisation more feasible	Conversational advice, recommendations, automated experiments	Lift, abandonment, fewer forms	Hallucinated feasibility and opaque decisions can increase downstream risk

Forrester's April 2026 CPQ landscape commentary describes 31 vendors in a mature but heterogeneous market, with specialization, interoperability, governance and complex go-to-market execution increasingly central. This supports convergence around capability requirements, not the emergence of a single winning commercial model. ⁰¹

3 · What the leading approaches actually promise

These are competitive approaches, not a rankings table. Vendor websites show product scope and declared positioning; they rarely establish independent comparative effectiveness. The “executive conversation” column is our interpretation of the promise a buyer could take to a budget owner, not an assertion that the vendor consistently leads with that language.

Player / approach	Positioning summary · G2G paraphrase	Documented product scope	Executive business conversation / limit
Epicor CPQ · manufacturing-led CPQ	Simplify complex ETO selling and engineering	Rules, visual configuration, CAD and BOM automation, dealer and sales channels 02	Sell complex manufactured goods with fewer handoff errors and less engineering rework. Strong alternative where manufacturing execution is decisive.
Threekit · AI visual commerce	Simplify selling with visual and AI-guided configuration	3D/AR, guided selling, quoting, order outputs and manufacturing instructions 16	Make complex choice intuitive. Broad configurator and production capabilities make it difficult for any vendor to claim that connecting the pieces is unique.
Expivi · headless visual CPQ	One central source across channels from configuration to production	Visual CPQ, ERP/PIM integration, BOM and production data 17	The same offer, accurate across channels. Direct counterexample to “one configuration layer is unique.”
Roomle · manufacturer/dealer 3D commerce	Configure and visualise products consistently in every channel	Cross-channel 3D configuration, retailer support, ERP/CAD/CAM links, manufacturer examples 05	Strong manufacturer–dealer deployment story; directly competes on reusable product knowledge.
3D Cloud · room-led enterprise commerce	Make buying complex products as easy as everyday online commerce	Room, kitchen and deck planners; AI; 3D CPQ announced September 2026; shared models and rules 03, 04, 19	From inspiration to actionable design and full-room purchase. A credible competitor in project planning and shared product logic.
Cyncly / Winner Flex · specialist design-to-install	Connect kitchen design, quotation, order and installation	Room measures, manufacturer catalogs, project records, installation planning, REST integration 06	Fewer wrong kitchens and remedial costs. Deep vertical specificity is the competitive strength.
VividWorks · unified product/room data	Reuse the same product catalogue and configuration logic in room planning	Unified Product Catalog announced as an alpha pilot in 2025; now listed in current features and pricing 15	Reusable product logic is an explicit market objective. Public listings do not establish deployment scale or economic results.
Zakeke · personalisation storefront	Make customisation easy and automate order outputs	Ecommerce plug-ins, visual personalisation, production-ready files 18	Automate personalization and production outputs. Suitability for a particular made-to-measure workflow needs direct validation.
IKEA · vertically integrated retail/service	Help customers design a complete room with professional advice	Planning appointments, 3D layouts, estimates, lists and services 14	Relevant competitor is not always software; it may be a staffed operating model.
Home Depot / Lowe’s · retailer-owned project tools	Turn project intent or lists into purchasable materials	AI list generation, takeoffs, quote workflows, project shopping, customer/pro services 08, 09, 20	These companies already pursue project share. They may build, partner or simply combine tools with human service.

A telling 2026 convergence: 3D Cloud launched visual CPQ across its room planners, kitchen designers and configurators in September. Cyncly and Omega announced a manufacturer-backed quote/check service in September to reduce errors between design, order and installation. Both are more concrete evidence than treating “AI advice” or “full project” as white space. ^{04, 07}

4 · What the market's recent decisions say

A. Retailers already want a larger share of the job. At Home Depot's December 2025 investor meeting, Pro executive Mike Rowe described a roofing order for 20 homes that exposed demand for windows, doors and moldings. A manual introduction resulted in a larger sale and repeat business. In May 2026 CEO Ted Decker still described cross-selling among Home Depot businesses as largely manual; in July the company formally coordinated CRM, catalog and fulfillment through an Office of Pro Acceleration. This is a management-backed strategy with nontrivial operating dependencies—not proof a storefront configurator could have captured that order. ^{10, 21, 11}

B. Project lists are becoming digital infrastructure. In January 2026 Home Depot released an AI material-list builder for professional projects; Lowe’s launched an AI tool in May that turns notes, photos and spreadsheets into quote-ready lists and complements blueprint takeoffs. That makes “we understand the whole project” an increasingly occupied starting claim. Neither press release proves completed-project economics. ^{08, 09}

C. Design-to-order validity is a source of margin. Cyncly’s Winner Flex ties room designs to quotation, manufacturing catalogues and installation. Its September Omega announcement explicitly links order checking to avoiding quote and order errors. The business value is not only higher conversion; it is preventing the cost of a bad commitment. ^{06, 07}

D. Visual engagement shows real use, not necessarily causality. 3D Cloud reported nearly 20-minute average room-planner sessions and a 29.7% project-save rate in Q2 2026, from its own enterprise customer network. This shows that customers use extended planning journeys; it does not alone demonstrate incremental revenue, causal lift or retained contribution. ¹³

E. Manufacturers can be the repeat deployment unit. Roomle markets shared product configuration across retail and manufacturer channels; 3D Cloud describes a common product and asset layer, and Expivi describes one source of product data across channels. Reuse across outlets is an established competitive axis, not an unclaimed opportunity. ^{05, 19, 17}

Evidence check · A sale can succeed while costs remain

Independent consumer evidence, with limits. Which? surveyed UK kitchen customers in February–March 2025. Among 797 respondents whose kitchens had been fitted within the previous three years using an in-house installation service, the five named kitchen brands had reported installation-issue rates ranging from 42% to 49%. These are self-reported installation issues in a specific sample—not attributable to configurators, not a category-wide remake rate, and not evidence of avoidable software spend. ²²

Documented improvement, with limits. Deloitte reports a manufacturer's new dealer-to-delivery system reducing manual errors by 98% and dealer configuration time from more than 30 minutes to under two. Configure One's older Scranton Products case attributes a 45% decrease in customer credits in 2008 to improved bill-of-materials accuracy, following CPQ implementation. Both are provider/client case accounts, not controlled studies; importantly, the second outcome is historical, not a new 2026 benchmark. ^{23, 24}

Evidence that competitors are already attacking the problem. The September 2026 Cyncly/Omega service routes Winner Flex designs to Omega experts for order and quote review before final commitment. Cyncly describes it as a way to reduce avoidable errors and protect retailer margins; public claims do not quantify net margin improvement. This is evidence of an active competitive response, including human expertise—not an unoccupied software opportunity. ⁰⁷

A broader analytical lens. A 2026 Journal of Business Logistics paper examines digital product fitting as a way to connect lost sales, unobserved demand, returns and inventory obsolescence. Its engaged research concerns three footwear-related businesses; it does not demonstrate that the same system or economics transfers directly to configurable home improvement. It supports examining the process as a whole, not assuming a specific solution. ²⁵

5 · The assumptions behind familiar answers

Familiar answer	Underlying assumption	Strongest alternative explanation
“Conversion is the proof of success.”	Once a valid order is paid for, the outcome has been created.	The customer, retailer and manufacturer can record different economic results once service, corrections, delivery and actual use are counted. Equally, some categories already have effective controls and low downstream loss.
“Improve configurator completion.”	A completed valid product choice solves the customer's important doubt.	The customer may still not know whether it works in the space, the project or the installation sequence.
“Use AI advice to raise conversion.”	Persuasive answers imply technically reliable decisions.	AI may lower interface friction while adding verification and exception costs.
“Win the whole project.”	More related items sold mean more profit.	Labor, errors, substitutions, refunds, delivery complexity and cannibalization can erase the gain.
“Integrate every manufacturer and retailer.”	Each additional connection increases platform scale.	Each could add more mapping, support, custom rules and retailer-specific work than recurring income.
“Our data makes us smarter.”	Access to events equals the right and ability to learn from them.	Retailers may own the customer relationship, withhold outcomes or forbid cross-account learning.

6 · Five strategic seams worth investigating

The seams below are discrepancies between commercial decisions and underlying outcomes. Each has observed evidence, a rival explanation and a way it could fail. They are not preselected product roadmaps.

Seam 1 — A completed order is not necessarily an economically successful purchase

Observed. Accurate configuration rules and ordering workflows reduce preventable errors, but installation remains a distinct source of friction. Which?'s kitchen survey reports frequent installation issues in its sample, without attributing them to configuration. Deloitte's manufacturer case reports fewer manual order errors after a connected dealer-to-delivery program; Configure One's older Scranton case reports fewer returns and credits following bill-of-material accuracy improvements. Cyncly/Omega already offers expert verification before a kitchen order is committed. Together these demonstrate the problem's economic relevance and the fact that competitors already work on it. ^{22, 23, 24, 07}

Possible divergence. Sales dashboards count completed configurations and order value. The customer measures fitness for purpose and time to completion; the retailer counts net contribution after service and returns; the manufacturer counts correct, profitable fulfillment. A sale can improve the first metric without improving every other one.

Alternative. Many orders already work correctly. Installation failures may come from site conditions, contractors or logistics unrelated to any configuration decision. Human review, narrower catalog rules or improved operations may outperform a broader software platform. The surveys and cases cannot distinguish these explanations.

Economic test. Follow matched orders from specification through delivery and, where applicable, installation. Attribute failure to its originating stage. Calculate incremental contribution after measurements, support, cancellations, returns, corrections, remakes, installation callbacks and added assurance cost; compare a software intervention with expert review or process changes.

Provocation. Who declares the sale successful—and whose P&L tells a different story six weeks later?

Seam 2 — The transaction is smaller than the job, but the job may be unprofitable to own

Observed. Home Depot's roofing cross-sell, Home Depot and Lowe’s project-list tools, IKEA design assistance and Cyncly's project workflows all illustrate active attempts to help buyers with multi-product jobs. These efforts acknowledge that a single basket may not capture the customer's whole need, but none proves that the entire project is profitable to own. ^{10, 08, 09, 14, 06, 21}

Possible divergence. Teams judge e-commerce wins by a configured item, while the customer is funding a staged job and purchasing the rest elsewhere.

Alternative. Project ownership may require installers, credit, measurement, scheduling and fulfillment assets beyond software; a retailer might rationally serve selected profitable components instead of owning the entire renovation.

Economic test. Incremental gross profit from previously uncaptured project items, minus service, delivery, coordination and exception costs, net of purchases shifted from other retailer channels. Compare software-only, assisted and full-service variants.

Provocation. Which parts of the project should a retailer want to win—and which should it deliberately leave to someone else?

Seam 3 — The storefront gets the sale; the manufacturer repeats the difficult work

Observed. Product selection occurs on a retailer's storefront, while technical rules, manufacturability and product updates commonly originate at the manufacturer. Roomle, Expivi and 3D Cloud explicitly describe product logic or assets reused across dealer or merchant channels. Their positioning confirms that manufacturers' distribution burden is a recognised software problem; it does not independently establish how much implementation work is saved. ^{05, 17, 19}

Possible divergence. The project is commissioned as a retailer “better shopping experience,” while the more persistent economic pain may sit with the manufacturer: maintaining correct product decisions, entering another retail account and reducing per-channel support.

Alternative. Retailers own traffic, product ranking, the cart and commercial data. Manufacturers may not get to control the purchase journey, and a retailer may insist on its own experience layer. A network could be a collection of bespoke endpoints, not a scalable economic advantage.

Economic test. Measure one manufacturer's cost and calendar time to activate each new retailer, accuracy of incoming orders, conversion/retail coverage, and the ratio of reused rules/integrations to bespoke work. Establish who actually signs and pays.

Provocation. If a manufacturer must teach every reseller how to sell the same complex product, is its problem really conversion—or the cost of being understood everywhere?

Seam 4 — More intelligent recommendations can increase the cost of being wrong

Observed. AI guidance increasingly sits beside configuration rules and visualisation. The Cyncly/Omega order-checking workflow demonstrates an explicit human validation gate. This is not a gap competitors have ignored. The open issue is whether automation can improve decisions without increasing the volume or cost of plausible-but-wrong commitments. ^{16, 07}

Possible divergence. Teams score AI on dialogue or recommendation lift, while buyers and suppliers care about a correct promise: something that can be produced, delivered and installed as represented.

Alternative. Human approval, manufacturer specifications and narrow catalog boundaries may already manage this risk at lower cost.

Economic test. Track recommendation accuracy by constraint class, human overrides, cost of failed commitments, supplier claim/return rates and responsibility for errors. Explicitly distinguish creative visual suggestions from executable orders.

Provocation. Does AI create a better answer—or a more convincing mistake? Who pays when it does?

Seam 5 — Clickstream learning is not automatically cross-network intelligence

Observed. Visual-commerce platforms advertise shared product models, recommendation engines and analytics. The 2026 Journal of Business Logistics study offers a useful, narrower precedent: in footwear, fitting interactions can help retailers understand returns, lost sales and even unobserved demand together. It does not prove a generalizable, rights-cleared cross-retailer learning platform for configurable home products. Nor does public vendor material establish widespread access to installation and remake outcomes. ^{03, 13, 19, 25}

Possible divergence. Product analytics optimise individual flows and SKUs while the most valuable learning could be why valid customer intent repeatedly fails: unavailable options, technical constraints, retailer assortment gaps or mismatched installation requirements across the network.

Alternative. Privacy, supplier competition and retailer negotiating power can make cross-channel pooling impractical. Statistical value might be too weak in low-volume configurations; standard reporting could be enough.

Economic test. Establish data rights, event-level consistency, outcomes visible after checkout, sample sizes and whether discoveries from one channel improve another without exposing proprietary retailer information.

Provocation. If everyone improves their own configurator, who learns which products customers repeatedly try—but fail—to buy?

What the five seams have in common: No individual metric represents the full economic outcome, and no actor automatically has the data, incentives or authority to optimize it end-to-end. That does not make a universal outcome platform the logical winner. The winning unit of competition could be a narrower, provable reduction in errors, a better selected project offer, cheaper manufacturer distribution or more accurate detection of unmet demand.

7 · Where a defensible advantage could really compound

A 2026 platform does not win merely because it contains visualisation, AI and APIs. Those are broadly offered. A possible compounding advantage needs at least one of the following mechanisms, with demonstrated rights and economics:

1. Reusable product decision logic: each new product range or retailer can be launched with less incremental work. This has strong conceptual value but many providers pursue it.
2. Distribution reach: a supplier can reach additional retail endpoints already integrated, reducing implementation and acquisition friction. Network counts alone do not prove marginal deployment savings.
3. Validated outcome knowledge: manufacturer- and retailer-approved learning from rejected choices, corrections and returns improves future guidance. Hardest to evidence, potentially more durable than generic AI.
4. Trust and responsibility: auditability, versioned specifications, clear human approval and supplier guarantees reduce the cost of errors. Vertical specialists may have a structural lead.

The commercial distinction is crucial: a proprietary rules library that must be rebuilt per client may be sticky services work; a rights-cleared and reusable product/decision layer that lowers activation cost may become software leverage. Neither can be inferred from a vendor's architecture slide.

Economic scorecard to demand before declaring an opening

Stakeholder	Outcome / denominator	How a misleading improvement could arise
Retailer	Net contribution per completed and eligible purchase or project, after support and corrections	Conversion and AOV rise while later service, credits, returns and cannibalization erase margin
Manufacturer	Profit per fulfilled order and active channel; order quality; channel activation cost	Order volume grows while rework, production exceptions and reseller support absorb gains
Buyer	Fit-for-purpose outcome, total cost and time to usable completion	Checkout speed improves while measurement, installation or project completion still fails
Software company	Recurring gross contribution / deployment; marginal effort to add a retailer	Enterprise logo count grows while delivery remains bespoke
Data-learning program	Verified causal lift from reused insights, with rights	Models see more clicks but no trusted fulfilled or failed outcomes

8 · Four commercial battlegrounds — without a predetermined winner

These are four distinct ways vendors and operating businesses might change the basis of competition. They overlap, but they have different buyers, proof requirements and economics. The evidence does not justify declaring any of them a new category, an uncontested territory or the inevitable winner.

Commercial battleground	Potential buyer's reason to act	Evidence of market movement	What could undermine the thesis
Dependable, usable purchases	Reduce costly incorrect specifications, remakes and loss of trust	Deloitte dealer-to-delivery case; Cyncly/Omega verification; historical Configure One/Scranton returns case	Installation problems may be unrelated; human validation and process fixes may be cheaper; no cross-category loss benchmark
Profitable share of the project	Capture adjacent purchases the buyer needs for a job	Home Depot and Lowe's project tools; IKEA and Cyncly planning workflows	Coordination, services and errors may cost more than the additional margin
Distribution without repeated rebuilding	Make technical products consistently sellable through more channels	Roomle, Expivi and 3D Cloud shared product or channel layers	Retailer-specific integrations, pricing and contracts may remain expensive and bespoke
Learning from purchase intent that fails	Use recurring customer difficulties to improve offers and decisions	AI-assisted selling and analytics capabilities across platforms	Data rights, low sample sizes and missing post-purchase outcomes can prevent useful learning

The category question is not which vision has the best tagline. It is which buyer has a painful problem, which organisation is willing to pay, and which provider can improve the economic outcome repeatedly at falling marginal cost. A category-level investigation should keep all four paths open until operational and financial evidence distinguishes them.

9 · The field investigation that would change our conclusion

Start with a manufacturer selling complex goods through multiple retailers and at least one cooperating retailer or dealer. Reconstruct four observed journeys: a qualified need that never became an order; a completed order with a successful handoff and use; an order corrected before production; and a purchase requiring a later return, remake or installation intervention. Follow the same decision from initial request through fulfillment and actual use rather than equating checkout with success.

For every journey, identify who received the sale, who paid for advice and exceptions, why corrections were required, and whether any software could reasonably have prevented them. Compare net contribution and customer outcomes, not just conversion. Ask manufacturers how much work and cost recur when enabling a new merchant; ask retailers who sets prices and controls the cart; ask both who owns return, complaint and installation data. Include the supplier's quality systems and human experts in the alternatives.

Kill criteria: post-sale losses are immaterial or originate outside the configurable-purchase decision; improvements require more human and integration cost than they save; manufacturers lack budget or channel rights; useful outcome data cannot be shared lawfully; observed lift is conversion-only; or existing competitors and simpler process controls already solve the problem at better economics.

Questions the category now forces on leadership

If the dashboard calls the purchase a success, whose costs and outcomes are missing from the calculation?

• When a made-to-measure product sells through many retailers, who pays to keep its specifications correct—and which work could be reused rather than repeated?
• Can manufacturers see which customer intentions repeatedly fail at the point of specification—and distinguish missing demand from ordinary browsing?
• When an AI-generated project looks convincing, which company is prepared to guarantee that the specification will work?
• Should the retailer own the entire project, or should software help it identify the specific portions it can fulfill more profitably than anyone else?
• What would a manufacturer pay to reduce the marginal cost of winning another retail channel, rather than simply improving one shop's conversion rate?

Final category judgment. The evidence supports a more demanding unit of analysis than conversion or configured order count: the economic result of helping a person choose, purchase and successfully use a complex product, including costs before and after the transaction. But the market already includes order verification, project planning and channel-specific product intelligence; these are not vacant spaces. The unresolved questions are which losses are material and preventable, which stakeholder would pay to reduce them, and which competitor can repeat the improvement more profitably than alternatives. A sale may be the moment revenue is recorded. It is not always where the economic story ends.