Figma Make Can Speed Up Landing Page Tests. Keep the Campaign IDs Close.

Figma Make changes the pace of landing page work

Figma announced Figma Make on May 7, 2025, then moved it to general availability on July 24, 2025. That matters for growth teams because Make is a prompt-to-app tool, not a mood-board generator. It can take an existing Figma frame, keep the structure and metadata, and turn it into an interactive prototype. Figma said the first release used Claude 3.7 Sonnet, with additional models planned later.

That is a real shift.

If you run paid search, paid social, SEO, and founder-led experiments in the same week, the slow part is rarely writing another headline. The drag is getting a believable page in front of the team before the media plan goes stale. A $7,500 LinkedIn test for a B2B demo offer can sit in limbo for 10 days because the design file, copy doc, CMS page, CRM campaign, and analytics notes all live in different places.

Figma Make can compress that first loop. A campaign brief can become a working page prototype in an afternoon. The trap is that speed makes sloppy variants cheap. Three generated pages turn into eight. Someone changes the hero offer. Someone else swaps the form. By Friday, nobody can say which HubSpot campaign owns the page, whether the Webflow page is being measured, or why Google Ads shows 412 clicks while HubSpot shows 119 sessions tied to the campaign.

I have seen that movie. The ending is ugly.

The operator move is to treat campaign measurement as part of the prototype, not a launch chore. If the page is born in Figma Make, the HubSpot campaign ID, UTM campaign value, Webflow Analyze check, form mapping, and naming convention should be born with it.

Start with the campaign object

For a paid landing page, I start in HubSpot before I open Figma. HubSpot’s campaigns tool groups related assets under one marketing campaign, and its 2026 Campaigns API returns a campaignGuid, formatted as a UUID. HubSpot’s own example uses edb9b6c3-d2e2-4ca8-8396-832262aed0d4. In the same API docs, the campaign properties include hs_name, hs_start_date, hs_end_date, hs_campaign_status, and hs_utm.

That last field is the one people skip.

HubSpot’s campaign UTM settings can auto-generate a unique campaign UTM value. As of the February 20, 2026 help article, that setting is on by default for new campaigns unless the account changes it. HubSpot also keeps previous UTM values as secondary values when you update a campaign, which is useful when a team renames an offer mid-quarter and does not want to break old reporting.

For a real paid campaign, I want three fields copied into the Figma Make prompt and the Webflow page brief: the HubSpot campaign name, the campaignGuid, and the default utm_campaign value. If the campaign is called q4_2026_demo_ai_search_ops, the page prototype should carry that exact string in its notes, CTA specs, and handoff checklist.

Small detail. Big consequence.

Here is a plausible prompt I would use inside Figma Make for a November 2026 B2B test:

“Create a responsive paid landing page prototype for a HubSpot campaign named q4_2026_demo_ai_search_ops. Audience is marketing operators at 25 to 250 person SaaS companies. Primary CTA is Book a 20-minute measurement audit. Include two offer variants, one proof block with three customer metric cards, and a hidden implementation note that every final Webflow URL must preserve utm_campaign=q4_2026_demo_ai_search_ops and HubSpot campaign GUID 5c7f0b8a-3f2e-4f7b-9df8-17a70f8276ad.”

Would I invent those customer metrics inside a live launch? No. In a prototype, I will use placeholders like “replace with verified Q3 2026 win rate” rather than fake a 38 percent conversion lift. But I do want the measurement identifiers in the prototype from hour one.

Use Figma Make for interaction, not attribution logic

Figma Make is good at producing the artifact everyone can react to. It can turn a static hero into a page with states, tabs, form behavior, animation, and responsive layouts. In July 2025, Figma said Full seat users could publish or privately share Make files, while Starter users could create unlimited drafts and share up to three Make files with a team. Figma also listed AI credit estimates for Full seats: Professional at 3,000 credits per month, Organization at 3,500, and Enterprise at 4,250.

That pricing detail changes behavior. A founder can ask for five page directions before lunch. A demand gen manager can prompt a proof block, a pricing objection panel, and a mobile rewrite without opening a ticket. The prototype becomes a working conversation.

I still would not let Make decide the analytics taxonomy.

Use it to expose measurement requirements in the page itself. Add a small internal-only section at the bottom of the prototype called “ops notes” or put the notes in the Figma file description. Include the HubSpot campaign URL, campaign GUID, default UTM campaign value, form name, thank-you page path, and Webflow collection or static page path.

For example, a clean handoff might say:

• HubSpot campaign: Q4 2026 Demo - AI Search Ops
• Campaign GUID: 5c7f0b8a-3f2e-4f7b-9df8-17a70f8276ad
• Default UTM: q4_2026_demo_ai_search_ops
• Webflow path: /lp/ai-search-ops-audit
• Form: hs_demo_audit_q4_2026
• Primary paid source: LinkedIn Ads
• Success event: booked meeting after form submit

That is the minimum I would hand to a Webflow builder. If a page cannot carry those details, it is not ready for production.

Webflow Analyze belongs in the preflight check

Webflow Analyze became available on October 31, 2024 as a native analytics add-on for eligible Basic, CMS, Business, and Enterprise Site plans. In June 2025, Webflow added lower Analyze tiers at 2,000 sessions per month for $9 per month billed yearly and 5,000 sessions per month for $19 per month billed yearly. By August 2, 2026, Webflow’s help docs described Analyze as tracking top pages, bounce rates, traffic sources, click behavior, and page-level engagement inside Webflow.

For landing pages, the useful part is proximity. The person editing the page can check behavior without waiting for a Looker Studio dashboard or a GA4 exploration. Webflow’s current Analyze page lists sessions, unique visitors, page views, bounce rates, click events, device type, language, location, referrer source, clickmaps, scrollmaps, and custom goals. Webflow also says Analyze setup takes one click once the add-on is enabled.

That does not replace HubSpot. It catches page-level issues faster.

My preflight for a Webflow paid page is boring, and boring is the point. Before the ad goes live, I want the published page to load with the final query string, the CTA click visible in Webflow Analyze, the HubSpot form submission tied to the campaign, and the thank-you page excluded from the wrong reports. If the page uses a Webflow form that syncs into HubSpot later, I want one test contact with a timestamp and source note. If it embeds a HubSpot form, I want the form asset associated with the campaign.

Run the test like a buyer would. Open a URL such as /lp/ai-search-ops-audit?utm_source=linkedin&utm_medium=paid_social&utm_campaign=q4_2026_demo_ai_search_ops&utm_content=founder_pain_v2. Click the main CTA. Scroll below the proof block. Submit with an internal email address such as mops-test+2026-11-03@company.com. Then check three places: Webflow Analyze for the page session and click behavior, HubSpot contact activity for the form submission and UTMs, and the HubSpot campaign dashboard for attribution.

Five minutes here can save a week of argument later.

The handoff should name the variant

Fast prototyping creates variant sprawl. Figma Make makes that easier because the cost of trying another layout is low. That is useful when the team is still deciding whether the paid offer should lead with a demo, a calculator, or a teardown. It gets messy when the naming scheme is “final”, “final2”, and “new-new”.

I use a variant name that matches the ad and the page. A LinkedIn creative called founder_pain_v2 should point to a Webflow page or query string that also says founder_pain_v2. HubSpot’s tracking URL builder supports UTM content, and the April 15, 2026 help article says utm_content is meant to differentiate ads when several point at the same page. Use it. Do not bury the variant in a Figma comment that nobody will read after launch.

A workable format looks like this:

utm_source=linkedin
utm_medium=paid_social
utm_campaign=q4_2026_demo_ai_search_ops
utm_content=founder_pain_v2

That gives paid media, Webflow, and HubSpot a shared handle. When Webflow Analyze shows shallow scroll depth on mobile for founder_pain_v2, the operator can find the ad, the Figma Make prototype, the Webflow page, and the HubSpot campaign without detective work.

Where this breaks

The breakage usually starts with a harmless sentence: “Let’s just get the page live.”

A founder sees a promising Figma Make prototype and wants traffic on it by Monday. The Webflow builder publishes the page. Paid search gets URLs from a spreadsheet. Organic social uses a cleaner URL because it looks nicer. HubSpot receives contacts, but the campaign UTM value does not match the default campaign UTM. Webflow Analyze shows visits and clicks, but the team cannot reconcile them with HubSpot campaign performance.

Nobody did one catastrophic thing. They skipped six small joins.

The fix is a launch record. One row per page variant. Keep it in Airtable, Notion, a HubSpot campaign note, or a plain Google Sheet. I care less about the system than the fields: Figma file URL, Figma Make prototype URL, Webflow page URL, HubSpot campaign GUID, default UTM campaign, source, medium, content, form ID, thank-you path, publish date, owner, and last QA date.

For a small team, that row is enough. At 20 active landing pages, it becomes the map. At 80 pages across paid search, paid social, partner webinars, and SEO offers, it becomes survival gear.

My operating rule

Figma Make is a speed layer. Webflow is the publishing layer. HubSpot is the campaign record. Webflow Analyze is the page behavior check. Each tool has a job.

When those jobs blur, the team gets motion without memory. You can ship ten landing page variants in a week and still learn nothing clean enough to trust. When the identifiers travel with the prototype, the same speed starts to compound. The paid team can test sharper angles. The founder can react to a working page instead of a copy deck. Marketing ops can trace the lead back to a campaign object with a real ID, a real UTM value, and a real page session.

That is the standard I would set for 2026 landing page work: prompt fast, publish carefully, and make every variant accountable before the first dollar hits the ad account.