When should a company run growth-driven design in-house versus hiring an agency?

The deciding factors are capacity, site traffic, and decision structure, not the headline budget number. Growth-driven design is not a one-time redesign; it is an ongoing cycle of planning, building, measuring, and transferring learnings, run in short sprints that repeat indefinitely [3]. That shifts the question from who can build a website to who can hold a testing cadence quarter after quarter. Three inputs settle it. The first is dedicated capacity: someone who can protect a real block of hours every month rather than squeezing optimization in around other work. Traffic is the second, because a page needs enough visitors for a test to reach a readable result inside a sprint or two. The third input is a lean decision team, since GDD moves only as fast as its slowest approval.

Neither path is inherently better; each is a fit for a different situation. Running GDD in-house fits organizations with a steady pipeline of website work and a team that already knows the brand and audience [1]. Hiring an agency fits companies that need launch speed, want specialist depth across strategy, UX, and development, or have no internal team to staff the work [1]. A sequenced or hybrid model fits the many companies that want an agency's speed early and internal ownership later. Underneath all three sits a capacity floor: practitioners put credible GDD at roughly 15 hours or $1,500 to $2,500 of real work per month, and the same floor that makes a token agency retainer pointless also sinks an under-resourced internal program [2]. So the useful comparison is not salary versus retainer in the abstract. It is which structure lets your organization protect that capacity, feed it enough traffic, and decide quickly on what to test next.

In-house fits when the website is steady, ongoing work rather than a periodic project, and you can commit three things: dedicated capacity, enough traffic to read tests, and a lean decision team. Internal teams already carry the brand's voice, audience knowledge, and product context, collaborate faster day to day, and keep output consistent across channels, which suits a continuous pipeline of content and product work [1]. The methodology's own requirement is that someone actually holds the plan, build, learn, transfer cycle on a repeating cadence [3].

Dedicated capacity

Someone has to own the sprint work as a real part of their job, not an after-hours extra. Practitioners put the floor near 15 hours or $1,500 to $2,500 of equivalent effort per month, and an internal program that cannot protect that block stalls the same way an underfunded retainer does [2]. Before choosing in-house, confirm who owns the hours and what comes off their plate to make room.

Enough traffic to read tests

Optimizations in GDD are meant to be driven and proven by data, not opinion [3]. That only works if a page draws enough visitors to reach a readable result inside a sprint or two; on thin-traffic pages, tests never resolve and the team ends up guessing anyway. If your key pages get low traffic, plan to test only the highest-impact changes and pick the clearer variant when a result will not reach significance.

A lean decision team

GDD moves at the speed of its slowest approval, so a small decision group with real authority is a prerequisite rather than a nicety. The workable shape is one decider plus one or two subject matter experts who can approve what ships each sprint. Approval by committee is the common stall; it slows internal sprints as reliably as it slows an agency, so it is worth naming a single decider before the first sprint rather than during it.

Training is no longer the barrier. HubSpot Academy publishes a free growth-driven design course an internal team can certify against, covering user research, agile web design, UI, and conversion optimization [5]. The remaining constraint is coverage, since a single designer rarely spans strategy, UX research, development, and experimentation at once, so an honest in-house plan names who covers each of those [1].

An agency fits when speed and specialist depth matter more than building internal muscle. Agencies bring objective, cross-industry perspective, staff specialists across UX, development, and strategy, scale capacity up or down by project, and run current testing and analytics tooling an internal team would otherwise have to buy and learn [1]. For a major redesign or a hard launch deadline, that assembled capability is the product you are buying.

Speed from a running process

The process is already running on day one. GDD agencies sell defined monthly capacity in standard packages, so the sprint discipline is contractual from the start rather than something the team has to build, and there is no hiring or ramp before work begins [2]. When a launch date is fixed, an established team compresses a calendar an internal build cannot match while it staffs up.

Senior specialists across disciplines

One retainer covers strategy, UX research, visual design, development, and experimentation at once, which no single internal hire spans [1]. That breadth matters most on a first GDD build, where the strategy and research decisions set the ceiling on everything downstream. Buying the full skill set for a defined window usually costs less than hiring four specialists to keep permanently.

Proven page flows

Agencies that have run the process many times arrive with page patterns that already convert, informed by cross-industry experience rather than invented per client [1]. That shortens the learning curve a fresh internal team would otherwise climb by trial and error. Inheriting a tested flow is often the difference between a launchpad that performs in its first month and one that spends two quarters finding its footing.

The trade-offs are real, and the next section weighs them in full: chiefly retainer cost that reads high against salary math, and the shared control that comes with structured review cycles, plus the ramp time even a strong agency needs to learn your brand and internal process [1].

The hybrid path is most common because it matches how the methodology is built to end: with a handoff. Continuous improvement in GDD includes an explicit transfer step that moves learnings and capability into the client's marketing, sales, and service functions, so an agency engagement can be structured from the start as a capability handoff rather than an open-ended dependency [3]. In practice that means an agency runs the strategy phase and builds the launchpad at full speed, then continuous improvement either transfers to a trained internal team or continues as a smaller retainer.

Agency builds, internal team sustains

The agency does the heavy, front-loaded work of strategy and the launchpad, where specialist depth and speed pay off most, then hands the repeating optimization cycle to internal staff once patterns are established. This concentrates the expensive specialist hours in the phase that needs them and puts the cheaper, steady cadence work where an internal team can carry it. The handoff window is also when brand and process knowledge transfers, which closes the gap that made the agency slower to ramp in the first place.

Agency builds, then a lighter retainer

The alternative keeps the agency on a reduced retainer for continuous improvement instead of transferring fully. An entry-level GDD package, near $3,300 per month in practitioner data, buys ongoing sprints without the fixed cost of internal specialists, and it fits teams that lack the traffic or capacity to run tests well on their own [2]. Companies often start here and shift toward internal ownership as traffic and in-house skill grow.

Either way, the hybrid answer resolves the capacity question by placing each kind of work where it is easiest to sustain. It buys speed when speed is scarce and moves to ownership once the internal team can hold cadence [4]. That is why, across real engagements, the sequenced model tends to beat a pure in-house or pure agency commitment made too early.

Control and cost move in opposite directions across the three paths: more direct control comes with more fixed cost and staffing risk, while lower fixed cost comes with shared control and ramp time. Each path carries its own version of that trade.

In-house trade-offs

The main in-house cost is fixed and compounding, since salaries are paid whether or not the sprint program produces learning, and a stalled internal team is the most expensive outcome of all, carrying full payroll with no test results to show [2]. The upside is direct control, because an internal team owns priorities immediately with no external review cycle. The risk sits in coverage and continuity: one departure can remove the only person who spans strategy and experimentation, and there is no bench behind them [1].

Agency trade-offs

An agency's cost is flexible operating spend that can be paused or scaled, though it reads high next to salary math and can exceed a single hire's fully loaded cost at the larger package tiers, roughly $3,300 to $14,600 per month [2]. Control is shared rather than direct, since work moves through the agency's structured review cycles and priorities are negotiated rather than simply assigned [1]. The other cost is ramp: even a strong agency needs time to learn your brand, product, and internal process before its output reflects them [1].

Hybrid trade-offs

The hybrid path trades a clean single-vendor relationship for a handoff that has to be managed. The transfer step only works if the agency actually documents patterns and trains the internal team, and if the internal team has capacity to receive the work, so a rushed handoff can leave a launchpad without anyone owning its continuous improvement [3]. Its cost shape is the most forgiving, front-loaded during the build and lighter afterward, but it asks the buyer to plan the transition deliberately rather than assume it happens on its own.

Across all three, naming which side of the trade you can most afford, direct control at fixed cost or flexible cost with shared control, is most of the decision.

Four questions settle it, and answering them honestly usually names the path for you.

Can you protect the hours?

Count the real capacity first. If the people who would run GDD internally cannot protect 15-plus focused hours a month after their existing jobs, the in-house plan fails the same capacity floor that makes token agency retainers fail [2]. If no one can own those hours, an agency or hybrid is the realistic path.

Do your key pages get enough traffic?

A site without enough visitors to read a test inside a sprint or two frustrates either model, but it punishes a paid retainer faster because you pay for cycles that cannot resolve [3]. Low-traffic teams either concentrate testing on the few highest-impact pages or lean on an agency's cross-client pattern library instead of their own thin data.

How fast do you need to launch, and who decides?

A fixed, near-term launch date favors an agency's running process, while a longer horizon leaves room to hire and train [1]. Pair that with decision structure: if approval runs through a committee, settle that before choosing either path, because slow approval stalls sprints regardless of who runs them.

What cost shape can you carry?

Retainers are flexible spend that can stop; salaries are fixed commitments that compound whether or not the program produces learning [1][2]. Choose in-house only if the internal pod will actually hold cadence, because a stalled internal program is the most expensive outcome of all, paying salaries with nothing to show. When speed is needed now and ownership matters later, the sequenced hybrid answers all four questions at once [3].