How to Measure PPC When Customers Research in AI and Never Click

To measure PPC when the click is not the whole story, stop judging campaigns by clicks alone. Track real outcomes like calls, booked estimates and closed jobs, feed them back to the ad platform, watch impressions and branded searches as directional signals, and check platform numbers against your own job records.

That is the short version. The rest of this article is how to do it without a marketing degree.

The idea comes from Navah Hopkins at Search Engine Journal, who answered a question a lot of advertisers are asking right now. I'm adding the home-services angle, because a roofer or HVAC owner has different problems than an online store.

Why don't clicks tell the whole story anymore?

Because people now research inside AI answers before they ever visit a website. They compare options, ask follow-up questions, and only click a citation if it helps. Your ad or your name can influence the decision without producing a click, so click-based reports undercount what paid search actually did.

Hopkins says it plainly. In her words, it is "harder to measure and attribute in the blended SERP and AI experience world."

She also points out that AI experiences are not built to send traffic to a website unless visiting the site actually helps the person. That means fewer signals for you to track.

Here is what that looks like in a trade business. A homeowner notices water stains on the ceiling. They ask an AI assistant what to do and who to call. They read an answer, see a few company names, and close the app. Later, they search your company name or call the number on your Google profile. Your ad may have been part of that story. Your report shows a direct visit or a phone call from nowhere.

Last-click attribution gives credit to the final touch and ignores the rest. That worked better when the path was shorter.

What should a home-services owner track instead of clicks?

Track the outcomes that pay you: phone calls, booked estimates, and jobs sold. A click is a step along the way, not the result. Set up call tracking and form tracking, mark which leads turned into real work, and send that back to the ad platform so it learns what a good lead looks like.

Hopkins calls conversion tracking your measurement superpower. She says conversions remain the clearest signal of value for any PPC campaign, whether or not AI is part of the journey.

For lead generation, she lists qualified form submissions, calls, appointments, and opportunities that advance in your CRM. For a contractor, that translates to something simple. A call from a tire-kicker is not the same as a call that became a signed job.

Here is the process I'd follow:

List the actions that matter: calls over a minimum length, form fills, booked estimates, sold jobs.
Make sure each one is tracked in your ad account, not just in your head.
Separate real leads from junk. Spam calls and wrong numbers should not count.
Record which leads became paying jobs.
Send that information back to the ad platform where you can.

Step five is the one most owners skip. Hopkins notes that automated bidding can only make good decisions when it receives reliable conversion data. If the platform only hears about every call, it will chase more calls. If it hears about the calls that became jobs, it learns to chase those.

She also warns that volume requirements vary by platform, campaign type and bidding strategy. Check the current rules before you switch bidding strategies. A small account may not have enough data to support every option.

How do I use impressions without fooling myself?

Treat impressions as evidence that people saw you, not proof they bought. Pull impression trends, then compare them with branded searches, direct traffic and calls over the same stretch. If they move together, your visibility is probably helping. If they don't, don't credit the ads.

Every time an ad serves, an impression happens. Hopkins notes that is true on a regular results page, an AI surface, or a display placement. You cannot map each impression to one person's AI conversation, but you can see that your ad was shown.

Her advice is to look for correlation across channels. The signals she lists are:

An increase in direct or other non-paid traffic.
Growth in branded searches across paid and organic.
More brand discussion picked up through social listening.
Changes in downstream conversions during periods of greater exposure.

For a local business, I'd translate that into things you already see. Are more people searching your company name? Are more calls coming in on the main number? Did your Facebook page get more mentions after a heavy ad month?

None of this proves a specific impression caused a specific job. Hopkins is careful about that, and so should you. It is a directional signal. It tells you whether more visibility lines up with more demand.

She also describes impression-based remarketing, where you target or adjust bids based on whether someone saw your ad. She discloses that she works for Microsoft Advertising and that the tool she describes is a Microsoft one. Whether your platform offers something similar is worth asking. The principle holds either way: exposure can change behavior without an immediate click.

How do I check platform numbers against my own records?

Keep a simple record of every lead and what happened to it: source, date, booked or not, job value. Then compare that to what Google or Microsoft reports each month. The platform shows what it saw. Your books show what you got paid for. When they disagree, trust the books.

Hopkins says to validate platform data against your own sources of truth. For a home-services company, that source is your CRM, your scheduling software, or even a spreadsheet.

Ask your office staff to note where each caller says they found you. It is not perfect. People forget, and some say "Google" for everything. But a few months of notes will show patterns that no dashboard will.

Look for gaps. If the ad platform reports a certain number of leads and your records show far fewer booked jobs, you have a quality problem, not a volume problem. If your records show more work than the platform credits, some of your paid search influence is hiding in other channels.

She also says to make consent part of your measurement strategy so you can capture the click-based data that is available. In plain terms, if your website has a cookie banner, make sure it is set up properly. A badly configured one can quietly block the tracking you are paying for.

Does local search matter when I'm measuring paid search?

Yes. If a homeowner reads an AI answer, then searches your name or calls from your Google Business Profile, paid search never gets the credit. Your profile, reviews and local rankings feed the same decision. Measure them together, not as separate silos.

This is where a lot of contractors get stuck. They look at ads in one report and their Google profile in another. The customer does not make that distinction. They see your ad, your star rating, your photos and your answer in an AI summary as one impression of your company.

If your profile is thin or your reviews are old, you may be paying for clicks that lose to a competitor with a better listing. Start by checking where you stand. The free Local Visibility Score shows how you appear in local search. If you want to see how AI assistants describe your business, the AI visibility report covers that.

If you want help with the profile side, our local search service is built for it.

What is a simple monthly routine for measuring PPC now?

Pick one hour a month and follow the same steps each time. Compare ad reports with your own job records, check branded search and call volume, and note what changed. The routine matters more than the tool. Consistency lets you spot trends before they cost you money.

Here is the routine:

Pull impressions, calls and form leads by campaign for the month.
Count how many of those leads became booked estimates and sold jobs.
Check branded search volume and total inbound calls against the previous month.
Compare what the ad platform reported against your own records.
Note any gaps, fix tracking if something looks broken, and send updated conversion data back.
Decide where the next month's budget goes based on jobs, not clicks.

Do not overreact to one month. Seasonal swings are real in roofing, HVAC and plumbing. Compare against the same month last year where you can.

What about last-click attribution?

Last-click is still useful, but it is incomplete. It tells you what happened right before a lead came in, not what shaped the decision. Use it as one input beside impressions, branded search trends, and your own job records.

Do not throw it out. Click data is still valuable. Hopkins is clear that clicks and conversions remain essential performance signals. The challenge is the extra influence that may happen before the click.

Think of it like a referral. The last person who mentioned your name gets the credit in your head, but five other things may have led the customer to trust you. You would not stop thanking referrers. You would just stop assuming they did all the work.

Common questions

Is PPC still worth it if AI answers are taking clicks?

Yes, if you track the right results. Ads can still generate calls and booked jobs, and impressions can influence later action. The question is whether your own records show profitable work from the spend. If they do, keep running it. If they don't, fix targeting or tracking first.

What is the single most important thing to track?

Sold jobs tied back to a lead source. Calls and forms are useful, but a lead that never becomes revenue is not a win. Even a simple spreadsheet that links callers to closed work will tell you more than click counts ever will.

Can I measure AI influence directly?

Not perfectly. According to Navah Hopkins, there are fewer attributable signals through AI experiences. You can use impressions, branded search growth, direct traffic and downstream conversions as directional evidence. Treat them as correlation, not proof, and back them up with your own customer notes.

Do small contractors need advanced attribution tools?

No. Start with accurate call and form tracking, a record of which leads became jobs, and a monthly comparison against your ad reports. Advanced tools help larger accounts, but most home-services owners get more value from clean tracking and honest bookkeeping.