The questions performance and creative agencies ask before they put AI UGC in front of a client — how to separate workspaces, who gets a seat, how to price the service, what to hand over, and which promises you should never make.
01
Client Workspaces
1.Can I use Klip Kanvas for client work?
Yes, and agency use is one of the most common patterns on paid plans. You can generate creative for any number of client brands from a single account, provided you hold the rights to use each client's product imagery, footage and claims. The practical setup is one workspace per client so brand kits, avatars, scripts and render history stay separated. Most agencies we work with start with two or three pilot clients, prove a monthly creative volume they can actually hit, then roll the workflow across the roster. Billing stays with the agency — you invoice the client for the service, not for credits.
#Client Workspaces#Pricing Your Service
2.How do I keep client assets separated?
Use one workspace per client rather than folders inside a shared workspace. A workspace carries its own brand kit — logo, fonts, colours, tone rules — plus its own asset library and generation history, so nothing bleeds between accounts. The failure we see most often is an agency running six brands in one space and shipping a video with the wrong lower-third logo. Naming conventions cover the rest: prefix every render with client code, month and test number, for example ACME-2604-H03. That prefix survives export and makes ad-manager reconciliation trivial three months later.
#Client Workspaces
3.Can one workspace hold multiple brands?
Technically yes, and for a two-brand holding company it is fine. Past roughly three brands it becomes a liability: brand kits collide, the asset library turns into a scroll, and the odds of a cross-client mistake rise with every new SKU. The rule of thumb we give agencies is one workspace per billing relationship. If a client pays you one invoice covering four sub-brands, one workspace with disciplined naming works. If they are four separate contracts, use four workspaces — you will eventually need to hand one of them over cleanly.
#Client Workspaces#Limits
4.What happens to the workspace when a client leaves?
Plan the exit before you plan the kick-off. Decide in the contract whether the client receives finished exports only, or exports plus source assets and scripts. Finished exports are the norm and are simplest: you deliver the MP4s, they keep running them, you archive the workspace. If the relationship ends badly you still want a clean boundary, which is why per-client workspaces matter. Note that workspace ownership is tied to your agency account — there is no one-click transfer of a workspace to a client's own account today, so budget an hour for manual handoff.
#Client Workspaces#Deliverables & Handoff
5.What does a first-month onboarding look like?
Week one: build the workspace, load the brand kit, pull product data, and write an angle matrix of 8–12 hypotheses. Week two: script the first six concepts and get them approved. Week three: generate, QA and launch the first batch of six creatives as three hooks by two avatars. Week four: read the results after 48–72 hours or 1,000 impressions per variant, kill the losers and brief the iteration. Do not promise a winner in month one. What you are buying in month one is a baseline and a working approval loop, and saying so up front sets the relationship correctly.
#Client Workspaces#Approvals
02
Seats & Access
6.How many seats does a five-person agency need?
Fewer than you think. Count only the people who actually generate or edit: usually one or two creative strategists and one media buyer. Account managers and clients rarely need a generation seat — they need to see finished files, which a shared link or a cloud folder handles. A typical five-person shop runs three seats. Buy seats to match the bottleneck, not the headcount: if one strategist produces 40 ads a month and nobody else touches the tool, a second seat buys you redundancy, not throughput. Check /pricing for current seat allowances per tier.
#Seats & Access
7.Should clients get their own login?
Usually no, at least not in month one. Client logins create two problems: they see raw generations that were never meant for review, and they start generating creative themselves without the strategic frame that makes it work. Give clients a review link or a shared drive of approved cuts instead. The exception is a mature, in-housing client who has explicitly asked to take generation over — in that case give them a seat in their own workspace and reposition your role as strategy and testing rather than production.
#Seats & Access#Client Workspaces
03
Pricing Your Service
8.How should I price AI UGC as a service?
Price the outcome, not the render. Clients do not buy 30 videos; they buy a tested creative pipeline. The three models that hold up are a monthly creative retainer (a fixed number of net-new concepts and variants per month), a performance-linked retainer with a smaller base plus a bonus on ROAS or CPA thresholds, and a project fee for a launch batch. Whatever you choose, quote in concepts and variants — for example eight new concepts and 24 variants a month — so scope is countable. Tool cost should land well under 15% of what you invoice for creative.
#Pricing Your Service
9.Per video or monthly retainer?
Retainer, in almost every case. Per-video pricing anchors the client on unit cost, which is the one number AI makes look cheap, and it punishes you for iterating — the exact behaviour that produces winners. A retainer prices the testing loop: research, concepting, generation, QA, launch support and a monthly readout. Keep per-video pricing for one-off overflow work and price it high enough that it is never the cheaper path. If a client insists on unit pricing, bundle a minimum of ten variants so iteration is baked into the deal.
#Pricing Your Service
10.Is it acceptable to mark up the tool cost?
Yes — you are selling creative strategy and production management, not software resale, and every agency in every discipline marks up production inputs. What you should not do is misrepresent it. If a client asks what tooling you use, tell them. Problems only start when an agency bills a line item as if it were a pass-through cost at a price it is not. Fold tooling into a single creative fee, or list it honestly as an agency cost. Clients rarely object to margin; they object to discovering a markup they were told did not exist.
#Pricing Your Service#Setting Expectations
11.Can I run this as a done-for-you creative subscription?
Yes, and it is the cleanest model for a small agency. A fixed monthly fee for a fixed creative output, no media management, no strategy calls beyond a monthly readout. It scales because production is the part AI compresses hardest, and it de-risks the client because they can cancel without unwinding a media relationship. The constraint is churn: subscriptions without a performance narrative churn at 4–6 months once the novelty fades. Attach a monthly one-page readout of hook rate, hold rate and CTR by concept and the retention problem largely solves itself.
#Pricing Your Service
04
Approvals
12.How do I build an approval workflow that does not eat the month?
Approve at the concept stage, not the render stage. Send the client the hook lines and the body script in a single document, get sign-off there, then generate. Reviewing finished videos invites cosmetic notes that cost re-renders; reviewing scripts catches the expensive problems — wrong claim, wrong audience, wrong offer — while a fix costs nothing. A workable cadence is scripts approved by the 25th, renders delivered by the 3rd, launch by the 5th. Cap video-stage feedback to legal and factual corrections only, and put that in writing.
#Approvals
13.How many revision rounds should I offer?
One script round and one video round, stated in the contract. Unlimited revisions sound generous and are the single most common way an AI UGC retainer stops being profitable, because generation is cheap enough that clients assume changes are free — and the cost is your team's attention, not credits. Define a revision as a batched set of notes delivered inside 48 hours. Notes arriving in three separate emails count as three rounds. Anything beyond the included rounds is billable at your hourly rate.
#Approvals#Setting Expectations
14.What should I actually send a client for review?
Send a numbered contact sheet, not a folder of files. For each creative, give the ID, the hook line in text, the avatar, the angle it tests and the placement it is built for. Attach the videos beneath. Clients review far faster when they can see the test logic in one scroll, and you get structured feedback instead of 'the third one feels off'. For a batch of 24, that document takes fifteen minutes to build and saves a week of thread. Keep the same IDs in the ad account so the monthly readout maps back to the review doc line by line.
#Approvals#Deliverables & Handoff
15.What do I do when the client hates the avatar?
Swap it and move on — this is a cheap objection to satisfy and an expensive one to argue about. Regenerating the same script with a different avatar costs one render cycle of 3–5 minutes. What matters is separating taste from evidence: if the client dislikes an avatar that is producing a 35% hook rate, show the number and offer to run their preferred avatar as a parallel variant rather than a replacement. Founders in particular have strong opinions about who represents their brand, and that is legitimate. Budget two avatar swaps into every onboarding.
#Approvals#Setting Expectations
16.How do I QA 40 ads before a client sees them?
Run a fixed five-point pass on every file and never skip it because a batch is late: audio present and level, lip-sync holding through the final third, captions inside the safe zone, claims matching the approved script word for word, and the correct logo and CTA end card. Watch at full speed with sound on — muted scrubbing hides the two defects clients notice most, audio drift and a mispronounced brand name. Forty files takes roughly 90 minutes. Log defects against the concept ID so recurring problems get fixed in the script stage next cycle rather than at QA.
#Approvals#Deliverables & Handoff
05
Deliverables & Handoff
17.What formats should I deliver to a client?
Deliver 9:16 at 1080p as the default, plus 1:1 for feed placements and 16:9 only if they run YouTube in-stream. Export each creative with captions burned in and, where the client's editor may want to restyle, a second clean version without captions. Include an SRT if they localise later. File naming is part of the deliverable: CLIENT-YYMM-CONCEPT-HOOK-RATIO, so ACME-2604-C02-H03-9x16. 4K export is available on premium plans but is rarely worth the file size for paid social — platforms recompress aggressively and 1080p is the practical ceiling.
#Deliverables & Handoff
18.Should I hand over source projects and scripts?
Only if the contract says so and you have priced for it. Finished exports are the standard deliverable; scripts and project files are your working method and often your competitive advantage. That said, be realistic — a client who wants to leave will leave, and withholding a script rarely retains anyone. A middle path that works: hand over scripts and exports, keep your research, angle matrix and testing history as agency IP. Be explicit in the contract, because 'all creative assets' means very different things to you and to their legal team.
#Deliverables & Handoff#Limits
19.Who owns the videos — the agency or the client?
Between you and Klip Kanvas, the account that generates a video holds the commercial rights to use it. Between you and your client, ownership is whatever your contract says, and the default assumption in most jurisdictions favours whoever the contract names — so name someone. Standard practice is that the client owns the finished creative on full payment, and the agency retains the right to show the work in a portfolio. Write the portfolio clause in explicitly; brands in sensitive verticals often want it removed, and that is easier to negotiate at signing than after launch.
#Deliverables & Handoff#Limits
20.How do I report creative results to a client?
One page, four blocks, monthly. Block one: creatives launched and spend behind them. Block two: the diagnostic ladder per concept — hook rate, hold rate, CTR, CPA — so the client sees which layer is failing. Block three: what you learned, written as angle-level conclusions rather than file-level trivia. Block four: next month's hypotheses. Report at concept level, not variant level; a client does not need 24 rows. Use the same IDs as the review document so anyone can trace a number back to a specific video without asking you.
#Deliverables & Handoff#Setting Expectations
06
Setting Expectations
21.Do I have to tell the client the creative is AI-generated?
Yes. Tell them before the first invoice, not after a rejection. Most brands are fine with it — many are actively asking for it — but discovering it from a Meta disclosure prompt or a customer comment destroys trust instantly. Frame it accurately: AI avatars performing scripted UGC, not real customer testimonials. That framing also protects the client, because testimonial and endorsement rules are where AI creative gets brands into genuine trouble. Put a one-line description of the production method in the statement of work.
#Setting Expectations
22.What performance should I promise a client?
Promise process and volume, never results. Commit to a countable output — for example eight concepts and 24 variants a month, launched and read within 72 hours — and to the benchmarks you will report against: hook rate above 30%, hold rate 12–20%, cold-traffic CTR above 1%. Do not commit to a CPA or ROAS number. Creative is one input among offer, price, landing page, audience and seasonality, and an agency that guarantees ROAS is one bad quarter from a dispute. Show the client the diagnostic ladder in the first meeting so they judge you on the right layer.
#Setting Expectations#Pricing Your Service
23.How many creatives a month is a realistic retainer?
For a single-brand client, 20–40 finished variants a month is the band where the work stays high quality and the media actually consumes it. Below 20 you cannot test structurally — you need at least six creatives per test cycle, built as three hooks by two avatars on one body script. Above 40 you are usually producing more than the account can spend against, and creative sits unlaunched. Scale the number to media spend, not to what generation makes possible: roughly one new concept per $3,000–5,000 of monthly spend is a sane starting ratio.
#Setting Expectations#Approvals
07
Limits
24.Can I white-label the platform for clients?
Partially, and you should know the boundary before you promise anything. Exported videos carry no Klip Kanvas branding or watermark on paid plans, so the deliverable itself is fully yours to present as agency work. What is not available today is a fully branded client-facing portal under your own domain with your logo in the product UI. If your pitch depends on the client never seeing a third-party tool, keep them out of the platform and deliver finished files. There is a fuller answer in the white-label and reseller FAQ.
#Limits#Deliverables & Handoff
25.What can't I do with client accounts today?
Four things worth knowing before you sell them. There is no one-click transfer of a workspace from your account to a client's own account — handoff is manual. There is no client-facing approval portal with comment threads; you will use your own review tool. Per-client credit budgeting is not enforced by the platform, so a heavy month on one client draws from the same pool as the rest. And billing is per account, not per client, so you cannot generate a client-specific invoice from the tool. None of these block agency work; all of them will surprise you if you promise otherwise.
#Limits
26.How do I handle a client whose product AI UGC handles badly?
Say so before you take the retainer. Apparel is the clearest example — garment drape, fit on a real body and fabric movement are exactly what current avatar video reproduces least convincingly, and a fashion client judging you on creative realism will not be happy. The same applies to anything whose selling moment is a physical demonstration you cannot fake. The honest pitch is a hybrid: AI avatars for hooks, talking-head framing and volume testing, real footage for the demo beat. Agencies that scope this correctly keep clients; agencies that oversell realism lose them in month two.
#Limits#Setting Expectations
Ready to put this into practice?
Create your first AI UGC video ad in minutes — no filming, no actors, no editing.
