How to Migrate From ADF XML to ACP: A Practical Guide
Migrating from ADF XML to ACP (Automotive Commerce Protocol) is a four-phase project that takes most dealerships 60 to 90 days end to end. The phases are audit your current ADF flows, run one pilot lead source on ACP in parallel, measure the fidelity and conversion difference, then sequence the rollout to your remaining lead sources by volume. The work is mostly coordination, not engineering. The biggest risk is not technical. It is vendor cooperation. This guide walks through every phase, what to do, what to measure, and where the migration typically gets stuck.
I have walked dealers through this migration. I have also done it inside my own products. The pattern is consistent. Here is the playbook.
Why Migrate From ADF to ACP at All?
Three reasons that compound.
AI metadata loss. Your AI vendor produces 30 to 50 structured fields per lead. ADF carries roughly 20. The rest die at the email boundary. I cover this in What Happens When AI Captures a Lead but the CRM Cannot Hold It.
GLBA compliance gap. ADF defaults to plaintext SMTP. GLBA 2022 requires encryption in transit. The gap is enforceable now. I go deep on this in Is ADF XML Still GLBA Compliant in 2026?.
Vendor lock-in. Every legacy ADF integration is bespoke. Every bespoke integration is maintenance debt. ACP collapses that debt because every ACP-compliant system speaks the same shape.
Any one of these justifies the migration. All three together make it inevitable.
Phase 1. Audit Your ADF Flows (Days 1 to 14)
Before you migrate anything, map the existing state. Two outputs from this phase.
Output one. The ADF flow inventory. List every lead source, every CRM, every system that touches ADF in your stack. For each one, capture: lead provider name, monthly lead volume, contract term, current transport (SMTP, HTTPS-wrapped, TLS-enforced), and which CRM consumes the lead. This is one spreadsheet. It does not need to be fancy. It needs to be complete.
Output two. The fidelity baseline. Pick three to five recent leads from each lead source. For each lead, count the number of AI-generated fields the vendor sent. Count the number of fields that landed in your CRM. The gap is your baseline. Most dealers find 30 to 60 percent of fields are being lost. You cannot improve what you have not measured.
Phase 1 takes about two weeks because the data lives in different places: vendor portals, CRM logs, IT inventories, and contract files. The work is mostly coordination across teams that do not usually talk to each other.
Phase 2. Pick One Pilot Lead Source (Days 15 to 30)
Do not migrate everything at once. Pick one lead source and run it in parallel on ACP. Three criteria for picking the right pilot.
Criterion one. The vendor already supports ACP, or is willing to. Some vendors are early adopters. Some will implement it on request. Some will refuse. Start with the vendor that says yes the fastest. The first migration is about proving the pattern, not about picking your most important source.
Criterion two. Moderate lead volume. A few hundred leads per month is the right size. Enough volume to produce meaningful comparison data. Not so much volume that a problem creates a business emergency.
Criterion three. A defined output consumer. The leads have to flow to a specific CRM and reach a specific BDC team. Pilots that flow to "the system" or "the team" have no measurable consumer and produce no measurable result.
Once the pilot is selected, run ADF and ACP in parallel for 30 days. Every lead lands twice: once through the legacy ADF flow, once through the new ACP flow. The CRM team works the ADF version. The fidelity tracking happens on the ACP side. Nobody is forced to change behavior during the pilot.
Phase 3. Measure the Difference (Days 30 to 60)
Track five things during the pilot.
Field count delivered. ADF: 20. ACP: 30 to 50. The first metric is just "how much more arrived."
Field count consumed. Even when ACP delivers 50 fields, your CRM may only know how to store 25. The consumed count tells you where the data layer downstream of ACP also needs work.
Time to first contact. Did the BDC reach the lead faster because the AI-generated routing fields actually arrived? Most dealers see a 30 to 60 percent reduction in time-to-first-contact on ACP pilots.
Conversion rate. Did the lead close better because the BDC arrived with full context instead of name and phone? Compare the same 30 days against the same lead source's ADF historical rate.
Compliance posture. Document that all ACP traffic is HTTPS-encrypted. Capture the audit trail logs. This becomes evidence for the broader compliance case.
If the pilot data supports the migration case, you have what you need to take it to leadership for the broader rollout. If the data does not support the case, the pilot saved you from a multi-source migration that would have produced the same disappointing numbers.
Phase 4. Sequence the Rollout (Days 60 to 90)
Now you migrate the rest. Sequence by three factors.
Factor one. Vendor cooperation. The vendors that say yes go first. The ones that say maybe go second. The ones that say no get scrutiny. Some of them you may decide to replace as part of the migration.
Factor two. Lead volume. Bigger sources have bigger impact. Migrate them once the pattern is proven, not before.
Factor three. Compliance exposure. Sources sending consumer financial data over plaintext should be prioritized regardless of volume because the GLBA exposure does not scale with volume.
Most dealerships have 3 to 8 lead sources. Migrating them one per week through this phase is realistic. Some sources will require contract renegotiation, which extends the timeline. Some will be replaceable, which shortens it.
What About Backward Compatibility?
ACP can wrap ADF. A vendor that has not migrated yet can keep sending ADF, and an ACP-aware gateway in front of your CRM can wrap that ADF payload in a CORD envelope before passing it on. You do not lose existing data. You also do not gain the AI metadata, because ADF does not carry it. But the compliance and audit trail benefits arrive immediately.
This is the bridge that makes the migration practical. You do not have to coordinate every vendor change before you can move. You can move first and let vendors catch up.
Where Migrations Get Stuck
Four common failure modes.
Failure 1. CRM does not have fields for the new data. ACP delivers 50 fields. The CRM has space for 25. The other 25 either get dropped or get stuffed into a notes field. This is not an ACP problem. It is a CRM problem. Fix it by working with your CRM vendor to add custom fields or by switching to an AI-native CRM that was designed for this shape.
Failure 2. Vendor refuses to implement ACP. Some vendors are not willing to do the engineering work. Three responses are reasonable. Replace the vendor with one that does. Keep the vendor and use the wrap-ADF bridge until they come around. Or escalate through procurement leverage during contract renewal.
Failure 3. BDC training did not happen. ACP delivers richer leads. The BDC was trained on the old field set. They keep working leads the old way and the new fields go unused. Schedule training before the migration completes, not after.
Failure 4. No defined success metric. The migration completes. Nobody measures whether it produced value. Leadership asks what the project achieved. The team produces process metrics. Define the business metric in Phase 1 and report it in Phase 4.
What Does the Vendor Conversation Sound Like?
Three scripts that work.
For lead providers. "We are moving our lead intake to ACP as part of a 2026 data layer modernization. Can you confirm your timeline for ACP support? If you do not have one, we need to discuss how we maintain our relationship during the transition."
For CRMs. "We are receiving lead data in ACP format with 50+ structured fields. We need our CRM to natively store and surface these fields. What does that implementation look like on your roadmap?"
For OEMs. "We are seeing significant compliance and AI-readiness benefits from moving our data layer to ACP. We would like to discuss what OEM portal support for ACP would look like."
These conversations move faster than expected once dealers actually start having them. Vendors are not blind to the trend. They are waiting to be asked.
What Does the Migration Cost?
Less than most dealers expect. Three cost categories.
Engineering. If your CRM and lead provider both support ACP, the integration is a configuration change. Maybe a week of vendor implementation time. If they do not, the engineering scales with the gap. Plan for 2 to 8 weeks per vendor depending on cooperation.
Vendor renegotiation. Some vendors will use the migration as a moment to discuss contract changes. Plan for procurement involvement. The leverage is usually on the dealer side because ACP support is becoming a market expectation.
Internal coordination. One person owning the migration. Not a CTO. An AI Operator or operations lead. The role is half project management, half technical translation. Without a dedicated owner, the migration drifts and the timeline doubles.
Total cost for a typical mid-market dealership: $15K to $60K all-in, depending on vendor cooperation and CRM gaps. Payback is usually under 6 months from the conversion lift alone, before counting the compliance and AI-metadata benefits.
The Strategic Window
The dealers who migrate to ACP in 2026 have a window that closes once the standard becomes the assumed default. Early adopters get three years of compounding advantage: cleaner data, better AI integration, and compliance posture their competitors cannot match.
The dealers who wait pay the migration cost on someone else's timeline, usually during a vendor-forced switch or a post-breach compliance scramble. Same destination. Worse terms.
What Comes Next
Read the broader case in The Modern Automotive Lead Format: What Replaces ADF in 2026. Read the head-to-head in ADF XML vs ACP: A Side-by-Side Comparison. The full ACP specification is at ACPspec.org. The universal envelope it rides on is at CORDspec.org.
Final Thought
The migration is not the hard part. The hard part is deciding to start. Once you start, the work is structured, the pattern is repeatable, and the wins compound. Audit, pilot, measure, sequence. Four phases. Sixty to ninety days. The standard exists. The reference implementation is open. The only thing left is the decision to move.
Structure before tools. Strategy before speed. Systems before scale.
The AI Operator Framework is the methodology for designing, governing, and scaling AI operations inside a business. Explore the full framework at JamilAshkar.com/framework. Start with the free course at JamilAshkar.com/course.
