Should an Agency White-Label AI Visibility? Four Paths and 8 Questions Before You Sign (2026)

Table of Contents

1. The four paths in one table
2. 8 questions to ask before signing a white-label deal
3. Isolation is not a copywriting problem
4. How to explain AVS to your clients
5. UltraGrowth direct subscription vs white label
6. Summary

The 30-second answer: When clients start asking about AI visibility, web and marketing agencies have four paths. If you are only asked now and then and have few clients: refer them, or let clients subscribe to an existing service directly. If you have a steady client base and want to keep the relationship: white label. If you have an R&D budget and want to stand out: build it yourself; the scanning engine is MIT-licensed open source and you can take it.

Whichever path you choose, before you sign, pin down client ownership, brand leakage, who owns the assets after termination, and what you can promise clients. Below are 8 questions, plus three sentences you can put straight into a proposal that describe without promising.

Disclosure: This post is written by Ultra Lab. We sell both white-label licensing and the UltraGrowth subscription, so two of the four paths are ours. We describe building it yourself as a real option too.

Disclosure: Ultra Lab, MindThread and Ultra Advisor are products of the same group, Ultra Creation (傲創實業).

#The four paths in one table

	Build	Refer	White label	Client subscribes directly
Upfront cost	Highest: front end, reports and operations are all yours	Lowest	Medium: monthly licence fee, quoted by tier	Low: the client pays
Client ownership	You	Usually moves to the other company	You (end clients stay under your name)	Client deals with the vendor
Who owns delivery quality	You	The company you referred to	Vendor runs the product; you face the client	Vendor
Brand shown	Yours	Theirs	Yours	Vendor's
Assets after termination	All yours	Depends on your agreement	Ask before signing (question 5)	Per vendor terms
Best for	R&D budget, want differentiation	Rarely asked, do not want to handle it	Steady clients, want to keep the relationship	Few clients, or clients who want to manage it themselves

UltraProbe's scanning engine is published under the MIT licence on GitHub, so using it to build your own is fully permitted; the open-source library's checks are not identical to the web version's, so compare before you build. The real work is outside the engine: the client-facing front end, report templates, support, and maintenance after rules change.

There is no public price list for white-label licence fees; the /partner page says we charge a monthly fee by licence tier. For outsourcing price bands in Taiwan, see SEO Pricing in Taiwan 2026.

#8 questions to ask before signing a white-label deal

1. Who owns end-client data and the client list? Will the vendor contact my clients directly? For us, the /partner page states that end clients always stay under your name and that we will not go around you to contact them. Whoever you work with, put this in the contract.
2. How is brand leakage prevented? Ask item by item: how the production domain decides whose brand to show, whether preview and production are separated, and whether report files or outgoing emails contain the vendor's name.
3. Do product updates reach the white-label site? Are there changes I need to review? Syncing is good, but you need to know about updates that change report formats or wording before your clients see them.
4. Is the scoring or reporting method public? Can I show the method when a client questions it? AVS, for example, uses a public formula; the current v2.0 is SEO × 0.35 + AEO × 0.35 + AAO × 0.30. If you cannot show the method, you cannot answer the client.
5. After termination, who owns the content, settings and reports on the client's site? Ask about articles, site settings and past reports separately, and write each into the terms.
6. When service goes down or data is wrong, who answers to the end client, and how fast? White-label clients only know your brand, so you take the call. Agree in advance how quickly the vendor responds and how incidents are reported.
7. Can I set my own prices? Is there a minimum volume? With us, pricing is your call; minimums and similar terms are not public, so confirm them when you talk.
8. What can I promise clients, and what can't I? Neither recommendations nor rankings; the reasons are in "How to explain AVS to your clients" below. Our own UltraGrowth page also states that it does not promise specific rankings.

#Isolation is not a copywriting problem

Question 2 deserves more space, because we hit this ourselves. Adding a preview-environment parameter to our main site URL switched the whole homepage to a partner's brand, and the state was stored in the browser, so opening the clean URL afterwards still showed the partner's version. The fix: the production domain decides by domain only, and the parameter works only in preview. The incident and fix are in Two Agents, Two Deploys.

No amount of careful copy stops this kind of leak; the real white-label seam is in the code. So do not just ask a vendor whether they are careful. Ask: what does production use to decide the brand? Are preview and production separated? How do you verify after a change?

#How to explain AVS to your clients

The easiest way for white-label delivery to go wrong is overselling the score. Get three things straight:

• It is a technical scan, not mention tracking. AVS checks how ready the site itself is for search and AI; it does not tell you whether ChatGPT mentioned the brand this week. The difference is covered in Choosing an AI Visibility Tool.
• It is not Semrush's AI Visibility. The names are similar, but they measure different things, the numbers cannot be converted, and there is no causal link between them.
• A high score does not guarantee a recommendation. Our AVS validation study observed that sites cited by AI mostly sit in higher score bands, but that is correlation, not causation, and it was measured under the v1 formula (SEO × 0.5 + AEO × 0.5).

Three sentences you can put into a proposal (they describe, they do not promise):

1. "This score is the site's technical readiness: whether search engines and AI can reach and read your website."
2. "It is not the same thing as the AI Visibility score in Semrush and similar tools, and the two numbers cannot be converted or compared."
3. "A high score does not guarantee a recommendation from ChatGPT or any AI, nor a ranking; we use it to prioritise fixes and to confirm they worked."

Also, do not name the score an "AI recommendation score" on a white-label report. The name itself is a promise.

#UltraGrowth direct subscription vs white label

They solve different problems.

Client subscribes to UltraGrowth directly: per the public plans on /growth, a one-time setup of NT$19,800 and monthly tiers of NT$6,800, NT$12,800 and NT$24,800, with no long-term contract; prices are as shown on that page. Content is published on the client's own website and stays theirs after they cancel. The client sees the UltraGrowth brand.

White-label licensing: per /partner, two product lines are currently open for white label: AI security and visibility scanning (the UltraProbe licence tier) and social auto-posting (the MindThread licence tier), on your domain, logo and report templates, with you charging the client. UltraGrowth is not one of these lines and is currently offered mainly as a direct subscription.

How to choose: if you want your own brand and your own billing, and the need is scanning or social posting, choose white label. If the client wants an ongoing content production line and is fine with the vendor's brand, a direct subscription is simpler.

For the channel-partner pitch behind this model, see White-Label AI Products for Channel Partners.

#Summary

1. None of the four paths is always right; it depends on client volume, whether you want to keep the relationship, and whether you have an R&D budget.
2. Ask the 8 questions before signing, and write client ownership, post-termination assets and service responsibility into the contract.
3. The score describes technical readiness only. Do not promise AI recommendations, and do not compare it with Semrush's AI Visibility.

To see what a report looks like first, scan a client site for free with UltraProbe. To talk about white-label licensing, leave your details at /partner or email contact@ultralab.tw.

FAQ

What is white-label AI visibility, and how is it different from a referral?

White label means taking a vendor's existing product, putting it on your domain, logo and report templates, and charging your clients directly; the client sees your brand. A referral hands the client to the other company, and the client deals with them from then on. The difference is who keeps the client relationship and who answers to the client: with white label it is you, with a referral it is usually the other company.

If I white-label, will the vendor contact my clients or use my client list?

That depends on the vendor's terms, so ask before signing. For Ultra Lab, the /partner page states that end clients always stay under your name and that we will not go around you to contact them. Whoever you work with, put this in the contract rather than relying on a web page.

Can I call the score on a white-label report an "AI recommendation score"?

No. AVS is a website technical scan score. It shows whether search engines and AI can reach and read a site; it does not mean AI will recommend it. It is also not Semrush's AI Visibility: the two measure different things and cannot be converted. The link between the score and being cited by AI is correlation, not causation, and a high score does not guarantee a recommendation from ChatGPT or any other AI.

After the contract ends, who owns the content on the client's site?

It depends on the plan, so ask item by item before signing: articles, site settings and reports. For a direct UltraGrowth subscription, the /growth page states that content stays on the client's own website and remains theirs after they cancel. For white-label licensing, write the post-termination arrangements into the contract.

Can UltraGrowth be white-labelled?

According to the current public /partner page, white-label licensing covers two product lines: AI security and visibility scanning (the UltraProbe licence tier) and social auto-posting (the MindThread licence tier). UltraGrowth is currently offered mainly as a direct subscription; for other arrangements, email us.

With white label, can I set my own prices?

For Ultra Lab's white-label licensing, yes. The /partner page states that we charge a monthly fee by licence tier and you decide what to charge your clients. Licence fees are quoted by tier and there is no public price list; confirm minimums and other terms when you talk to us.

How much work is it to build AI visibility scanning in-house?

You do not have to write the scanner from scratch: UltraProbe's scanning engine is published under the MIT licence on GitHub (ppcvote/ultralab-scanners). The main costs are everything around the engine: the client-facing front end, report templates, support, and maintenance when rules change. Real effort varies a lot by team and scope, so we do not estimate it for you.