The Lead Source Problem

When lead sources go uncorrected, the next budget decision rests on a number nobody checked.

The CRM source field looks authoritative because something is written in it. That something may have been picked by a form integration, inherited from a vendor mapping nobody has touched in three years, typed over by a salesperson clearing a required field, or flattened into a generic bucket on import.

In one five-rooftop Harley-Davidson group, one store ran unattributed leads under 1%. Another store — same brand, same vendors, same season — ran above 13%. The marketing at the second store was fine. The two rooftops were simply writing down the same customer journey two different ways.

One customer can produce several competing sources

A rider sees a Meta ad on Tuesday, searches the dealership by name on Thursday, opens a unit page, calls the tracking number and lands in the CRM as a phone lead. Meta credits its click. Analytics credits the last campaign it saw. The call platform credits the landing page. The salesperson picks "phone" from a dropdown, or "walk-in" if he is in a hurry. Every one of those records is telling the truth about a different piece of the same customer.

So the store needs a written source policy: what each field means, which system owns it, and what is allowed to overwrite it. Without one, moving campaign budget on the strength of a CRM source report is a guess wearing a spreadsheet.

Start with a source dictionary

Pull every source and sub-source out of the CRM and map each one to where the lead actually came from. Then stop asking a single field to carry four separate facts.

• Channel: paid search, paid social, organic, OEM
• Platform or partner: Google, Meta, H-D.com or a lead provider
• Contact method: form, phone, chat, walk-in
• Campaign: the specific launch, model or offer

Google Analytics recommends consistent UTM values for source, medium and campaign. Tag every link you control that way, then go find out whether the website, the call platform and the CRM still hold those values by the time the lead lands. Usually one of the three drops them.

Validate routing separately from attribution

A perfectly sourced lead can still land at the wrong rooftop, in the wrong department, on the desk of somebody who left in March. Submit real test leads through every form that matters and call every tracking number, once mid-morning and once after close. Write down where each one landed, how long it took a human being to answer, and whether that person could see which unit the customer was asking about.

Pied Piper's 2026 study found only 47% of powersports dealers answered a website shopper's question by email or text. Attribution that is accurate to the decimal point still cannot rescue a lead nobody works.

Correct the record before the monthly meeting

Source correction is something the store does every month, the way it counts parts. Work the unknown, direct and catch-all buckets against form payloads, call records and campaign logs. Log every correction and keep the raw value sitting next to the corrected one, so anyone can see what was changed and who changed it.

Do not quietly shove every ambiguous lead into paid media because paid media is the line item under review. Some of that demand is organic, some is a repeat customer, some is a referral from a guy in the service lane. A report worth trusting keeps an unresolved bucket and shows how big it is.

Match back to sold units where the data allows

Which source produced the lead is the easy half. The half that pays for itself is which lead became an appointment, which appointment walked through the door, and which of those matched a name on a bill of sale. That takes the CRM and the DMS talking to each other, a customer identifier both systems agree on, and a date window everybody signed off on before the numbers were run.

Keep the denominator on the page. Total leads, unique leads, matched records, unresolved records and sold units each get their own column. A platform conversion has never once been delivered to a customer.

Train the people who touch the record

This cannot live with the agency alone. The BDC, the sales managers and the CRM administrator all touch the record, and any one of them can undo a month of cleanup with a dropdown. We build the source dictionary, the routing map and the monthly review, then train each role on the two or three calls it actually owns.

We stay on as the QA and escalation layer, or we run the whole thing. Either way the store should be able to explain its own source logic without calling us first.

Leadership checklist

• Is there one written definition for every source and sub-source?
• Do website, call, OEM and paid-media leads preserve campaign context?
• Has every critical form and phone number been tested at the destination?
• Are unknown and catch-all sources reviewed before reporting?
• Can the report separate channel, platform, campaign and contact method?
• Are platform conversions kept separate from CRM leads and sold units?
• Does one person own the correction log and monthly QA?