You mostly cannot, and any system that claims to close that gap completely is overstating it. What you can do is measure the size of the gap and stop making decisions that assume it is zero.

On this page

• Why can't LSA see a call that came through my website?
• What partially closes the LSA attribution gap?
• How much attribution loss should I expect?
• What did I get wrong the first time?
• What should I do about unattributable calls?

Why can't LSA see a call that came through my website?

LSA counts the leads that come through its own call and message flow. A person who sees the ad, searches your name, visits your site and dials the number on the page is a direct or organic interaction as far as every tracking system is concerned.

The credit goes somewhere else. Nothing is broken.

What partially closes the LSA attribution gap?

Dynamic number insertion on the website, so paid visitors see a distinct number. Helps for people who arrive via a click, not for people who saw the ad and searched separately.

Ask the caller. Unfashionable and genuinely useful — a CSR question at intake, captured in the booking system, is the only source that sees the whole path.

Geographic and temporal correlation. Blunt, but LSA activity in a specific area followed by direct calls from that area is real signal at volume.

How much attribution loss should I expect?

On a 14-location home-services network, the share of bookings that could be matched back to a click sat in a 0.56 to 0.83 band across settled windows, and the roughly 30% miss was traced to structural causes rather than a fault: customers dialling a static or listing number rather than the tracked one, booking later, and cross-device journeys.

That investigation is worth knowing about because of what it ruled out. Runtime testing on all 13 sites confirmed the tracking identifier was captured, transmitted to the booking system, and the number swapped — identical on the best-matching and worst-matching accounts. Duplicate tag containers did not correlate. The call-extension theory was refuted.

Conclusion: no queryable ad-platform setting explained the variance. The gap was downstream capture, not a Google problem, and most of the apparent spread between accounts was short-window noise rather than real difference.

What did I get wrong the first time?

The account-level match figures ranged 0.46 to 0.91, which looked like some locations had badly broken tracking. Decomposed over the full settled window, the real band was a tight 0.56 to 0.83. The headline spread was noise. Several locations were flagged as broken and would have been managed bluntly on the basis of a short-window snapshot.

Two genuine signals survived: two locations declined together at the same time while their websites stayed healthy, which pointed at a downstream operational change rather than anything in the ad account.

What should I do about unattributable calls?

Stop trying to attribute perfectly and start measuring the total. Total booked jobs against total media spend across all channels tells you whether the money is working. Per-channel attribution tells you where to push, with a known error bar.

If you want the matched share higher, the lever is enhanced conversions for leads plus consistent hashed customer data on every booked lead — which lifts coverage materially. That buys measurement accuracy, not automatic permission to spend.

Related questions

If someone clicks my ad and then phones instead of filling in the form, does Google Ads know that lead came from the ad?

Read the full answer.

How do I dispute a bad lead in Google Local Services Ads now that there's no dispute button?

Google removed the per-lead dispute button and moved to automatic crediting, which means your job shifted from arguing individual leads to shaping what the system sends you in the first place. Read the full answer.

Why am I first on one Local Services Ads search and nowhere on the next?

LSA ranking is recomputed per search and is heavily driven by proximity to the searcher, your responsiveness, review profile, and whether your budget can still serve. Two searches minutes apart from different postcodes are genuinely different auctions. Read the full answer.

Written by Usama Khan, who runs Google Ads for e-commerce brands and home-services companies at AdventurAds. Figures are from accounts under active management and are anonymised. Last reviewed 2026-08-26.