How to Write Adaptation Guidelines for Your Adlea Form — With 10 Real Examples
The Adaptation Guidelines field is the most powerful part of an Adlea form. It is also the one that most users leave blank.
Not because they do not want to use it — but because writing natural language instructions for an AI system is a different skill from filling in form fields or selecting dropdown options. There is no syntax to enforce, no error message if you phrase something incorrectly, and no immediate feedback confirming that the AI understood your instruction the way you intended.
This post gives you the framework, the signal vocabulary, and 10 real examples to work from. By the end, you will be able to open your Adaptation Guidelines field and write your first instruction without second-guessing the format.
What Makes an Adaptation Guideline Work
Every effective Adaptation Guideline contains three components. All three need to be present for the instruction to produce reliable, consistent results.
1. A trigger condition — the signal the AI should observe before doing anything. This is the "if" part. It tells the AI what to look for: a page the visitor viewed, a UTM parameter on the session, an answer already entered in the form, or a session trait the AI has assigned.
2. An action — what the AI should do when that condition is met. This is the "then" part. The available actions are: hide a field, show a field, pre-fill a value, reorder a field's position in the form, or rephrase a field's label. Every guideline must specify one of these explicitly.
3. A field reference — which specific field the action applies to. Use the exact label text from your form editor, in quotes. Without a precise field reference, the AI has to guess, and it will occasionally guess wrong.
A guideline that includes all three is unambiguous. Compare these two versions of the same intent:
Weak: "Adjust the form for enterprise visitors."
Strong: "If the visitor has the 'Enterprise Lead' session trait, show the 'Annual contract preference' field and move it directly after the 'Company size' field."
The weak version gives the AI interpretive latitude it should not have. The strong version tells the AI exactly what to observe, exactly what to do, and exactly where to do it.
The Signals You Can Reference
Before writing guidelines, you need to know what information the Adlea AI has access to when your form loads. There are four categories of signal available to you:
Behavioral signals — everything the visitor did on your site before opening the form. Pages visited by name or URL path, time spent on the site, scroll depth on specific pages, number of pages viewed in the current session, and whether this is a return visit.
Reference examples: "if the visitor has viewed the pricing page", "if the visitor has spent more than five minutes on the site", "if the visitor viewed more than four pages in this session", "if this is the visitor's second or later visit"
Traffic source signals — how the visitor arrived. UTM parameters (source, medium, campaign, content), referrer domain, and campaign tags are all available at the session level and require no additional configuration beyond standard UTM tagging on your campaigns.
Reference examples: "if the visitor arrived from utm_source=linkedin", "if the UTM campaign contains the word 'international'", "if the referrer domain is a competitor site"
Form answer signals — what the visitor has already entered in the current form session. You can reference values from earlier fields to trigger logic on downstream fields. This is how you build conditional paths inside a single form without creating multiple separate forms.
Reference examples: "if the visitor selected 'More than 500 employees' for the 'Company size' field", "if the visitor entered a non-generic email domain", "if the visitor selected 'Enterprise' in the 'Plan interest' field"
Session trait signals — AI-assigned behavioral labels generated automatically based on the visitor's site activity. The three primary traits you can reference are High Intent, Price-Sensitive, and Enterprise Lead.
Reference examples: "if the visitor has the 'High Intent' session trait", "if the visitor has the 'Price-Sensitive' trait", "if the visitor has the 'Enterprise Lead' trait"
You can combine signals within a single guideline. "If the visitor arrived from a paid campaign AND has the 'High Intent' session trait, pre-fill the 'Inquiry type' field with 'Urgent demo request'" is a valid and specific instruction.
Not using Adaptation Guidelines yet? This is the feature that turns a standard Adlea form into a fully personalized qualification engine. See it in action before reading the examples.
10 Adaptation Guideline Examples
The following examples are grouped by the type of signal they reference. Each one includes the scenario, the guideline as you would write it, and a note on why it works.
Traffic Source and Campaign Logic
Example 1: International campaign pre-fill
Scenario: You run separate paid campaigns for domestic and international applicants. Visitors from your international campaign should not have to select their applicant type manually — it slows them down and introduces selection error.
Guideline: "If the visitor's UTM campaign contains 'international', pre-fill the 'Applicant type' field with 'International' and mark it as read-only so the value cannot be overwritten."
Why it works: The read-only instruction is the part most people miss. Without it, the pre-filled value is just a suggestion. With it, the field is answered and locked, and the visitor moves directly to the next question.
Example 2: Competitor referral
Scenario: Visitors arriving from a competitor comparison page are in active evaluation mode. Surfacing a question about their current tool is highly relevant for them — and irrelevant noise for everyone else.
Guideline: "If the referrer URL contains a competitor domain name or a known comparison keyword such as 'alternative' or 'vs', show the 'What tool are you currently using?' field after the 'Company size' field."
Why it works: The field only appears for the visitors most likely to answer it accurately. Every other visitor never sees it.
Behavioral Session Logic
Example 3: Pricing page visitor
Scenario: A visitor who has already viewed the pricing page does not need to answer awareness-stage questions about what problem they are trying to solve. They know the product. Treat the form accordingly.
Guideline: "If the visitor has viewed the pricing page during this session, hide the 'How did you hear about us?' field and the 'What problem are you trying to solve?' field. Move the 'Preferred plan' field to the second position in the form."
Why it works: You are doing two things at once — removing friction and reordering the conversation to match where this visitor actually is in their decision. Both changes are triggered by a single behavioral signal.
Example 4: High-engagement session
Scenario: A visitor who has viewed more than four pages in a single session has already demonstrated strong product interest. Asking them to describe their business in an open text field is redundant — their session behavior already told you what you needed to know.
Guideline: "If the visitor has viewed more than four pages in this session, hide the 'Tell us about your business' open text field. If a 'Readiness to buy' field exists, pre-fill it with 'High'."
Why it works: You are not guessing at intent — you are acting on a behavioral signal that already confirms it. Removing the open text field for these visitors shortens the form at exactly the point where friction is least justified.
Example 5: Return visitor
Scenario: A visitor on their second or third session has already made the effort to come back. The fastest path to conversion is the most valuable thing you can give them.
Guideline: "If this is the visitor's second or subsequent visit, move the primary contact field to the top of the form and collapse all optional qualification fields so they are hidden by default on first view."
Why it works: For return visitors, the form's job shifts from qualification to capture. This guideline reflects that shift without requiring a separate form.
Form Answer Logic
Example 6: Company size to contract type
Scenario: Enterprise prospects and SMB prospects need to answer different downstream questions. Rather than building two forms, you encode the branching logic directly into the guidelines.
Guideline: "If the visitor selects 'More than 500 employees' in the 'Company size' field, show the 'Annual or monthly contract preference' field and the 'Procurement timeline' field. Hide the 'Self-serve or guided onboarding' field."
Why it works: One adaptive form handles both segments cleanly. Neither audience sees questions that do not apply to them, and you collect segment-specific qualification data without the operational overhead of managing separate forms.
Example 7: Program of interest
Scenario: You offer multiple programs, each with one or two qualification questions that only apply to applicants for that specific program.
Guideline: "If the visitor selects 'Executive MBA' in the 'Program of interest' field, show the 'Years of management experience' field directly below it. If the visitor selects any undergraduate program, hide the 'Years of management experience' field entirely."
Why it works: Qualification logic that previously required a separate form per program now lives in a single adaptive form. The visitor experience is cleaner, and your submission data is more structured.
Session Trait Logic
Example 8: High Intent — compress the form
Scenario: A visitor the AI has tagged as High Intent is close to a decision. At this stage, adding qualification friction is counterproductive. Your goal is capture, not qualification.
Guideline: "If the visitor has the 'High Intent' session trait, hide all fields marked as optional, move the primary contact information field to the first position, and rephrase the form submit button label to 'Book my demo'."
Why it works: The form's behavior changes based on the visitor's readiness state, not their demographic profile. For High Intent visitors, fewer fields and a more direct submit label remove the last remaining friction before conversion.
Example 9: Price-Sensitive — surface budget context early
Scenario: A visitor the AI has identified as Price-Sensitive — based on multiple pricing page views, time spent on plan comparison sections, or scroll depth on pricing — will respond better to budget-framed questions earlier in the form.
Guideline: "If the visitor has the 'Price-Sensitive' session trait, show the 'Monthly budget range' field after the email field and rephrase its label to 'What monthly budget are you working with?'"
Why it works: The label change matters as much as the positioning change. "Budget range" reads as a qualification screen. "What monthly budget are you working with?" reads as the opening of a commercial conversation. Same data, different relationship with the visitor.
Device and Context Logic
Example 10: Mobile — reduce field count without losing data
Scenario: On mobile, long forms carry a significantly higher abandonment rate than on desktop. Where possible, you want to reduce the visible field count without sacrificing the data that matters.
Guideline: "If the visitor is on a mobile device, merge the 'First name' and 'Last name' fields into a single 'Full name' field. Hide the 'Job title' field and the 'Company LinkedIn URL' field."
Why it works: You are not collecting less information from mobile visitors in any material sense. You are removing two fields that add meaningful form friction on mobile but contribute low-value data points relative to the cost of asking for them.
Three Things That Make Guidelines Fail
1. Writing a condition without an action
"If the visitor came from a paid campaign" is an incomplete guideline. The AI knows what to observe but has no instruction for what to do. Every guideline must contain an explicit action: hide, show, pre-fill, reorder, or rephrase, followed by the target field.
Fix: Read every guideline you write and check that it contains all three components — condition, action, and field reference. If any one of them is missing, the instruction is incomplete.
2. Vague or paraphrased field references
"If the visitor has a high budget, adjust the pricing question" will not produce reliable behavior if your form has multiple fields related to pricing or budget. The AI does its best with ambiguous references, but the results will not be consistent.
Fix: Copy the field label exactly as it appears in your form editor and paste it into the guideline inside quotation marks. "Hide the 'Annual budget estimate' field" is unambiguous. "Hide the budget field" is not.
3. Guidelines that contradict each other
If two guidelines both apply to the same field in the same visitor scenario but instruct different actions — one says show, one says hide — the AI will resolve the conflict, but not necessarily the way you intended.
Fix: Before publishing, read your full guideline set and identify any field that appears in more than one instruction. For conflicting guidelines, add an explicit priority note: "Apply this guideline only if the visitor does not also have the 'Enterprise Lead' trait." Explicit override instructions always take precedence over the AI's default resolution behavior.
Ready to put your first guideline into production?
Already have an Adlea form? Open your Adaptation Guidelines field and write your first one using the patterns above. If you want a second pair of eyes on your setup, book a live session with the Adlea team.
