In this post, I will discuss about splitting design and engineering across two vendors.
Most splits are inherited rather than chosen. A design partner was already on retainer, procurement had a preferred development supplier, and the two met for the first time at your kickoff.
That arrangement works more often than its reputation suggests. It fails in predictable places, and every one of those places can be patched without firing anybody.
This piece is about the seams. If you are still choosing, the decision runs differently. A user interface design company that also builds will name the conditions that favour one owner.
The five seams
Work through them in order, because each seam makes the next one worse. Five is the complete list in our experience, and none of them is about talent.
The definition of done, the design system, the estimate, QA ownership and the release calendar. Each produces a distinct symptom on your side, which is how you diagnose which seam is open.
Seam one: the definition of done
Two vendors mean two definitions unless somebody writes one. Design considers a screen finished when it is approved, and engineering considers it finished when it matches the file.
Neither definition covers the states a real user hits. Nobody owns the empty list, the partial permission or the failed save, so those arrive as bugs in the final week.
The patch here is cheap enough. Make the design vendor write acceptance criteria per flow, then make the build vendor quote against those criteria rather than the screens.
Ask who signs those criteria off. Web app development quoted without them leaves the interpretation to whoever is closest to the deadline.
Seam two: the design system and its exceptions
Systems survive a split, and exceptions do not. A component gets reused in a slightly different context, somebody adapts it, and six weeks later two patterns exist.
The symptom is visual drift that nobody can attribute. Your stakeholders see inconsistency, the design vendor blames implementation, and the build vendor points at the file.
Name an exception owner at kickoff. One person decides whether an adaptation becomes a new pattern, and that decision gets recorded in the library rather than in a chat thread.
A web design agency will usually accept that role when asked, since it protects their own work. The role matters more than which side holds it.
According to McKinsey, large IT projects ran 45 percent over budget and delivered 56 percent less value than predicted. (McKinsey, 2012)
That research covered more than 5,400 projects with the University of Oxford. The value gap is the part worth noting. Coordination failures rarely show up as missing features, and nearly always as a product that does less than promised.
| Seam | What you see | Cheapest patch |
| Definition of done | Edge-case bugs in the final sprint | Acceptance criteria written by the designer |
| Design system | Two patterns for one interaction | A named owner for exceptions |
| Estimate | Change requests on work you assumed was included | One shared flow inventory both vendors price |
| QA ownership | Defects bounce between suppliers for days | A triage call with both vendors twice a week |
| Release calendar | Design finishes work that waits a month to ship | One backlog with one priority order |
The third row causes the most expensive arguments. Two separately produced estimates against two readings of the same request guarantee a gap, and the gap becomes your problem at invoice time.
What do the two vendors need from each other?
Each side needs three things from the other, and asking for them early costs nothing.
The design vendor needs the build constraints in writing before drawing. Response times and the current data shape both change the design. Web app development teams know those limits on day one.
The build vendor needs states rather than screens. A web design agency handing over happy-path layouts leaves the implementer to invent the rest, and invented states rarely match the intent.
Both need a decision log they can read. A website development company joining mid-project spends its first week asking questions that were settled months ago.
Ask each vendor what it wishes the other had sent last time. A UX design agency with split experience answers that in one sentence, and the answer tells you what to prepare.
Seam three: the estimate nobody owns
In a split, each vendor prices its own half against its own assumptions. The work between the halves belongs to neither estimate.
That work is real and it is substantial. Somebody has to translate flows into components, answer implementation questions and decide what happens when a design cannot be built as drawn.
Give both vendors the same flow inventory and ask each to list its assumptions. A website development company that prices against your document rather than its own reading gives you a comparable number.
Then budget that translation work explicitly. Teams that leave it out discover it as change requests, and change requests at week eight cost more than the same work at week one.
Pricing the boundary work
The work between the two halves has a price, and somebody pays it whether or not it appears in a quote.
Name it as a line item in both contracts. Design review of built screens, implementation questions and state decisions are real hours, and they belong in a budget rather than in goodwill.
Ask how each vendor has handled it before. A mobile app development company that has worked opposite an external design partner will quote the review time without being prompted.
Buyers of outsource web app development services should expect that line to be larger when the designs come from elsewhere. Reading somebody else’s library takes time, and web development services quoted without it absorb the cost in contingency you cannot see.
Then compare the two numbers carefully. A website development company quoting zero for boundary work has either absorbed it without saying so or plans a change request.
Seam four: QA ownership
Defects in a split engagement have a natural home nowhere. A visual bug looks like implementation to the designer and like a specification gap to the engineer.
Both readings are usually half right, which is why the ticket bounces. Each bounce costs a day, and a week of bounces turns a small defect list into a release risk.
Run a joint triage twice a week with both vendors in the room. The call needs no agenda beyond the open list, and it ends the bouncing because categorisation happens with both parties present.
Decide who fixes visual defects before the first sprint. A mobile app development company and a design vendor can share that work when the boundary is written down.
Seam five: the release calendar
Design and engineering run on different clocks by default. A design vendor works in weekly batches, while a build vendor works in two-week sprints with a frozen scope.
The mismatch produces finished design work that waits, and waiting design work goes stale. Requirements move, and the screens approved a month ago no longer match the product.
Put both vendors on one backlog with one priority order. Whoever owns that order on your side becomes the single point where sequence gets decided.
A mobile app development agency working to its own roadmap will push back on this, and the push-back is informative. Suppliers that cannot work from a shared order are telling you something about how they run projects.
According to McKinsey, every extra year a large IT project runs increases its cost overrun by 15 percent. (McKinsey, 2012)
Duration is the risk multiplier in that data, and coordination overhead is what stretches duration in a split. Every seam left open adds days, and the days compound into the pattern the research describes.
Who arbitrates, and how often?
Splits need an arbitrator on the client side, and most companies appoint one by accident. The role lands on whoever answers messages fastest.
Make that appointment a deliberate one. One person with authority over scope and sequence, available for two short slots a week, removes most of the friction the seams create.
That person does not need to be technical. They need the authority to say which of two reasonable options happens, and the memory of why the last decision went the way it did.
A UX design agency will often offer to run coordination for a fee. That works when the authority comes with it, and it fails when the agency can recommend but not decide.
What a local option changes
Proximity helps with exactly one seam, which is arbitration. Decisions that need two vendors and a client in agreement move faster in a room.
Buyers shortlisting a web development agency Dallas option often have that room in mind. Use it for the halfway review rather than for weekly status, since status travels fine.
The rest of the work is indifferent to geography. A web development agency Dallas shortlist and a distributed one face identical seams, with identical patches.
Judge any web development agency on the same five questions wherever it sits. A web development agency Dallas candidate and a remote one should answer them in the same amount of detail.
Ask any web development agency Dallas candidate how it has worked alongside an external design vendor before. The specifics of that answer predict the next six months better than the location does.
Common mistakes
- Letting each vendor write its own definition of done, then discovering the gap during the final sprint.
- Treating visual drift as a quality problem when no one was ever given authority over exceptions.
- Comparing two estimates built from two different readings of the same request.
- Running separate defect lists, so the same bug gets triaged twice and fixed by neither side.
- Appointing the arbitrator by accident, usually the person with the emptiest calendar.
- Asking a website development agency to absorb coordination without giving it any decision rights.
The last one is the most common. Coordination without authority produces meeting notes rather than decisions, and web design services billed for that time deliver very little.
Contract wording that holds a split together
Three clauses do most of the work, and they belong in both contracts in matching language.
One names the acceptance criteria author and the acceptance approver. Another names the exception owner for the design system. A third sets the triage cadence and who attends it.
Add one sentence about shared artifacts. The repository, the component library and the decision log live in your accounts, so neither vendor can hold the project’s memory.
Then set a joint review at the halfway point. Both vendors in one room, reading the same defect list, produces a more honest picture than two separate status calls.
When the split exists for procurement reasons
Sometimes the split is not a product decision at all. A framework agreement, a preferred supplier list or a group-level contract decides it before anybody asks about delivery.
Say so openly with both vendors. Suppliers behave better when they know the arrangement is structural, because they stop campaigning for the other half of the work.
Then spend the saved energy on the seams. A split with five patched seams outperforms a single owner with none, which is the part buyers rarely hear.
Ask each vendor for one named person as the counterpart. Teams relying on group chats for cross-vendor work lose decisions in the scroll.
Outsource web app development services and a separate design vendor both cope with this arrangement when the counterparts are named. A user interface design company left to guess who owns a question will ask the loudest voice instead.
Outsourcing the build while keeping design close
A common variant keeps design in-house and sends the build outside. The seams are the same, and two of them get easier.
Outsource web app development services make two seams easier here. Your designers write acceptance criteria without a contract change, and the exception owner is already on the payroll. That leaves the estimate and the calendar as the live risks.
Buyers of outsource web app development services should ask how the supplier handles designs it did not produce. The review step we run before pricing somebody else’s screens is set out at https://phenomenonstudio.com/outsource-web-app-development/.
Then check the reverse case too. Outsource web app development services bought alongside an external design vendor need the same triage call, since neither side owns the boundary by default.
A mobile app development company in that position usually asks for the design files two weeks before the sprint. Hearing that request early lets a web design agency plan its own work around it.
How our teams work inside a split
At Phenomenon Studio about half of the engagements that reach us as a product partner already have a second vendor in place. Our designers start by writing acceptance criteria for the next two flows. That single artifact removes most interpretation traffic on a SaaS platform. The team then asks for a joint triage slot twice a week, with both suppliers present and one client-side arbitrator. On a HealthTech product that call cut a defect list from forty items to nine in a fortnight. Correct categorisation on the first pass did most of the work. Our engineers also review the other vendor’s component library before quoting, since pricing somebody else’s patterns without reading them is how estimates drift.
Signals the split is costing more than it saves
Three readings tell you when to stop patching and consider consolidating. Each one is easy to collect.
Count the hours your arbitrator spends on cross-vendor questions each week. Past four or five, the split is consuming a part-time role nobody budgeted.
Track how many defects get reassigned at least once. A rate above a quarter means the boundary is still unclear after the triage call.
Watch the gap between design completion and release. Website design services producing work that waits more than a month are producing work that will need revisiting.
What changes when one side of the split is in-house?
Half of all splits are internal on one side. Your own designers brief an outside build team, or your engineers implement work from an external studio.
The seams stay the same and the politics change. Internal teams rarely get held to a written definition of done, so the discipline has to be deliberate rather than contractual.
Keep the artifacts identical to a two-vendor setup. A mobile app development agency receiving acceptance criteria from your in-house designers behaves exactly like one receiving them from a studio.
Watch the review bottleneck on the internal side. One in-house designer supporting a mobile app development agency and a web team becomes the constraint within a month. Website design services for the marketing site get dropped first.
Decide early who speaks for the internal half. A user interface design company on the other side needs one counterpart. Web design services and mobile app development services bought separately make that list longer.
Merging the two vendors later
Consolidation is a real option and it needs a plan rather than a decision. The transfer costs about a sprint when the artifacts live in your accounts.
Decide which half of the pair survives. A build vendor that inherits design usually needs a designer added, while a design vendor that inherits the build needs engineers it may not have.
Ask the surviving supplier to quote the handover separately. Mobile app development services and web development services both price transfers differently, and the number tells you how often they have done it.
Keep the other vendor for two weeks past the switch. The overlap costs little and prevents the loss of context that makes consolidations look worse than they are.
A short diagnostic you can run this week
Five questions, one per seam, and ten minutes with each vendor. Ask them separately and compare the answers.
Ask who writes acceptance criteria and who approves them. The second question covers who decides when a component exception becomes a pattern. Number three asks which document the estimate priced.
Then ask who owns a defect that could belong to either side. Finish with how work gets prioritised when the two vendors disagree.
Mismatched answers map directly onto the open seams. A user interface design company and a build supplier answering question one differently have shown you where next month’s problems start.
Oleksandr Kostiuchenko, Marketing Manager at Phenomenon Studio, suggests running that diagnostic in week one rather than at the first crisis. His point is that every answer is cheap to change before work starts and expensive afterwards, when each vendor has already priced its own interpretation.
What does a healthy split look like?
A healthy split is quieter than people expect. The arbitrator spends two slots a week on it, defects land in one list, and nobody debates whether a screen is finished.
Visual consistency holds because exceptions have an owner. The estimate holds because both vendors priced the same inventory.
Teams that reach this state stop thinking about vendor structure. A user interface design company and a separate build supplier become an arrangement instead of a daily problem. That is the most any procurement decision promises.
Frequently asked questions
Is a split always worse than one vendor?
No, and a well-run split beats a single owner that lacks either skill. The difference is whether the five seams have owners. Splits fail from unassigned responsibility far more often than from vendor quality.
Who should write acceptance criteria in a split?
The designer who drew the flow, with whoever handles web app development reviewing before pricing. That order keeps interface detail in the definition of done. A user interface design company unwilling to write them is worth questioning closely.
How much client time does a split need each week?
Budget two short slots for arbitration plus one triage call. Below that, decisions queue and both vendors slow down. Above five hours a week, the coordination load is worth comparing against consolidation.
Can we make one vendor manage the other?
It works when the managing vendor gets real decision rights and a fee for the work. Coordination without authority turns into note-taking. Check how disputes get resolved before agreeing to that structure.
What should we keep in our own accounts?
The repository, the design files, the component library and the decision log. Vendors holding any of those hold the project’s memory. UI UX design services and build work both continue smoothly when the artifacts stay with you.
How do we compare two estimates from different vendors?
Give both the same flow inventory and ask for assumptions in writing. Then map each number onto your flows rather than onto their phases. Gaps between the assumption lists are the scope you have not defined yet.
Does branding belong with either vendor?
Usually with neither, since identity work runs on its own clock. Branding companies hand over guidelines that a product team translates into interface patterns. Agree who performs that translation before the first sprint.
