How Did You Hear About Us: Options That Actually Track

Every page ranking for this question is a form vendor showing you the same five-item dropdown: Search Engine, Social Media, Word of Mouth, Advertisement, Other. That list is a survey artifact. It is not an attribution instrument, and for a service business spending real money on Local Services Ads and a Google Business Profile, it is close to useless.

The field only earns its place on your form if two things are true. The options map one-to-one onto lines in your marketing budget, and you actually reconcile the answers against GA4 instead of filing them. Here is how to build both.

What Should the 'How Did You Hear About Us?' Options Actually Be?

Use 8 to 12 named options, and write the list by opening your bank statement, not a template. Every option should be a channel you can turn off tomorrow and feel the difference. If an option cannot be cut, funded, or scaled, it does not belong on the list.

For a US service business, that usually lands close to this:

• Google search (organic result or the map pack — say 'Google' plainly, not 'search engine')
• Google Business Profile / Google Maps — the profile is a distinct channel from your website and deserves its own line
• Google Local Services Ads (the 'Google Guaranteed' box) — a separate paid product, not 'an ad'
• A Google or Meta ad
• ChatGPT, Gemini, or another AI assistant — new, growing, and invisible to your analytics
• A friend, neighbor, or coworker referred me
• I am a past customer / you did work for me before
• Another business referred me (agent, contractor, GC, adjuster)
• Nextdoor, Facebook group, or a neighborhood app
• I saw your truck, sign, or job-site placard
• Mailer, door hanger, or local print
• Other — tell us in a few words (free text)

Notice what is missing. There is no 'Social Media' — that single word collapses a $3,000 Meta ad budget and an unpaid Nextdoor thread into one meaningless bucket. There is no 'Internet.' And there is no 'Advertisement,' because you buy at least three different kinds.

Why Does the Generic Vendor Dropdown Tell You Nothing?

Because it was built for e-commerce checkouts and job applications, and it groups channels by technology rather than by spend. HubSpot's own guide to this survey suggests sample answers grouped as Advertisement, The Internet, Word of Mouth, Events, and Other, with 'The Internet' subdivided into items like Search Engines, Social Media, and Customer Review Sites (HubSpot).

That is a fine taxonomy for a survey. It is a terrible one for a plumbing company. 'Search Engines' silently merges your organic rankings, your map pack listing, your Local Services Ads, and your Google Ads — four separate invoices, four separate teams, one useless checkbox.

The test is simple: if a Monday-morning answer of 'Search Engines' would not change a single dollar of next month's budget, delete the option and replace it with the four things it was hiding.

Where Do You Put the Field So It Doesn't Kill Your Form Conversion?

Put it last — below the fields that are required to book the job — or move it to the thank-you page entirely. Field count, not field cleverness, is what costs you submissions: Baymard Institute's 2024 checkout research found that the number of form fields affects usability far more than the number of steps, and that the average checkout carries 11.3 fields when most sites need only 8 (Baymard).

That research is e-commerce checkout, not a quote request, so apply it as a direction, not a law. The direction still holds: every field you add is a tax, and this one is a tax you are paying for your own benefit, not the customer's.

For scale on the downside, Zuko's benchmarking database of over 93 million tracked form sessions puts average form completion at 51.71%, with mobile at 47.53% versus desktop at 54.48% (Zuko). Half your traffic is already leaving. Do not hand it a twelve-item dropdown before you have its phone number. We cover the field-count math in more depth in how many form fields is too many.

The safest placement of all: ask on the thank-you page, after the lead is captured. Response rates drop, but a lower response rate on a lead you already own beats a higher one on a lead you lost.

Should the Field Be Required, Optional, or Open Text?

Optional, closed dropdown, with one 'Other' option that opens a free-text box. Required is the single most common mistake — you are forcing a stranger to complete your marketing homework before they can ask you to fix their roof, and the ones who bail were leads.

Pure open text is the other failure mode. HubSpot's guidance is right on this point: rich text lets a customer write anything, which makes the answers far harder to analyze later, while a fixed list gives you something you can count. You will get 'google', 'Google', 'GOOGLE', 'internet', 'yr website' and 'I forget' — six rows, one channel, zero decisions.

The dropdown plus an 'Other' escape hatch gets you both: clean counts for the channels you already know about, and a free-text stream that tells you when a channel you never listed starts producing. Read the Other box every month. It is the only place a channel you never thought to list — an AI assistant, a new neighborhood app — can announce itself before you knew to ask about it.

What Do You Do When Self-Reported and GA4 Disagree?

Expect them to disagree, treat it as signal, and reconcile monthly on a single sheet — because the disagreement is usually GA4 losing information, not the customer lying. Google's own documentation defines (direct) / (none) as traffic that 'doesn't have a clear referral source,' and lists the causes: links without UTM parameters, redirects that strip UTMs, URL shorteners like bit.ly, people typing your URL straight into the browser, links inside offline documents such as PDFs, and ad blockers interfering with tracking (Google Analytics Help).

Read that list again. Almost every offline channel a service business runs — the truck, the yard sign, the neighbor, the mailer — arrives at your site through one of those doors and lands in the same undifferentiated bucket. Self-reported attribution is the only instrument that can see inside it.

Here is how the common conflicts resolve:

The verdict line: when the conflict is about which button was clicked, GA4 wins. When it is about why the person went looking for you in the first place, the customer wins. Those are two different questions and most attribution arguments are just people answering different ones.

Which One Should You Believe — the Customer or the Analytics?

Believe GA4 on the last click and the customer on the first touch. GA4 timestamps a click in milliseconds; a customer reconstructs a memory three days later, and memory is biased toward whatever felt human — the neighbor who recommended you, not the fourth Google search that actually got them to your booking page.

That bias runs one direction and you should plan for it. Word of mouth is systematically over-reported and search is systematically under-reported, because nobody remembers a search box. The correct reading is not 'SEO is doing nothing.' It is 'the referral created the intent and search closed it,' which means cutting either one breaks the chain. Our take on why last-click reporting misleads service businesses is in why most service businesses can't tell which ad dollar made them money.

One honest caveat: self-reported data is noisy at low volume. Under roughly 30 answers a month, you are reading tea leaves. Let it accumulate for a quarter before you move budget on it.

How Do You Get the Answer Into Your CRM So It's Usable?

One picklist field, the same option names everywhere, plus five or six hidden fields captured on the same form submission. The self-reported answer is only half the record — you want it sitting next to the machine-readable truth so a human can compare them without a data project.

Capture these as hidden fields on every form:

• utm_source / utm_medium / utm_campaign — first-touch values, stored in a cookie on the first visit, not overwritten later
• gclid — if it is absent on a lead who said 'a Google ad,' your ad tagging is broken
• Landing page URL — the page that actually earned the lead
• Referrer — the raw document referrer, before GA4 buckets it
• How did you hear about us — the picklist value
• Other (free text) — the escape-hatch answer, verbatim

Then there are the leads that never touch a form at all. Google's Local Services Ads documentation is explicit: those leads 'come in as phone calls and messages sent through your Local Services ad' (Google). Your website never sees them, so GA4 never sees them, so your form field never asks them. The only fix is a human one — the person answering the phone asks the question and types the answer into the same picklist. Same field, same names, one report.

If your booking flow, your CRM and your call intake use three different vocabularies for the same channel, you do not have attribution — you have three spreadsheets. Standardizing that is one of the first things we look at in a conversion-focused website build.

Does Asking This Question Actually Change Your Budget Decisions?

Only if you write the decision rule before you read the data. Otherwise the field becomes a monthly ritual that produces a pie chart and changes nothing, which is the state most of these forms are in.

Write it down in advance, in one sentence: 'If a channel produces no qualified leads in 90 days — by either self-reported answer or GA4 attribution — it gets cut.' We run a 90-day kill switch on every channel for exactly this reason. A channel that cannot show a qualified lead in a quarter is not underperforming, it is a subscription.

The AI assistant option matters most here, and it is the one nearly every dropdown on the internet is still missing. When someone finds you through ChatGPT or an AI Overview and then types your name into Google, GA4 logs a branded organic click or a direct visit — Google's documentation is clear that a typed URL is (direct) / (none). Without the field, that channel is invisible, and you will keep crediting SEO for demand that AI search created. That gap is the whole argument for treating GEO as part of the same program as SEO, not an upsell.

If your lead reporting currently boils down to 'Google, probably,' the fastest fix is to see where the leaks are before you buy anything else. We will map your form, your channel tagging, and your GA4 setup and tell you which numbers you can trust — Get my free audit.