Skip to content
Client case study · Foundation / basement services · New Hampshire

Google Search clicks up 248% while paid media spend fell 46%.

Caldrin began SEO/AEO work with 603 Basement Solutions on June 1, 2026. Across matched June and August Search windows, Google Search clicks rose from 113 to 393. Over the three-month campaign period, monthly paid media spend fell from $34.7K in June to $18.7K through August 28.

Search growth and paid-spend reduction are separate measurements. This case study does not claim that SEO caused every budget change.

113 to 393Search clicksGoogle Search Console · manual export
June 1-26, 2026 vs August 1-26, 2026
$34,673.41 to $18,670.49Paid spendGoogle Ads, Microsoft Ads and Meta platform totals supplied by the operator
June 2026 vs August 1-28, 2026
101 to 149Form submissionsPostHog · lead_submitted events
June 1-26, 2026 vs August 1-26, 2026
The business, and where it started

A basement and foundation contractor working three states.

603 Basement Solutions sells basement waterproofing, crawl-space work and structural foundation repair to homeowners across New Hampshire, Maine and Massachusetts. It is a local-service business: the work is sold by inspection, the demand is seasonal, and the buying decision starts with a problem a homeowner can see.

Caldrin started on June 1, 2026. June is therefore the baseline, and every figure on this page compares against it. There is no earlier measurement, so this page does not describe what the site did before the engagement began.

Baseline windows match the comparison windows exactly. Where a window is partial, it says so.

Observable at the startJune 2026
Google Search clicks113
Google Search impressions16,881
Average Search position18.24
Local URLs surfacing69
Foundation parent average position31.83
Q&A Search impressionsNo rows surfaced
The 90-day scoreboardthree systems, kept separate
Search clicks+248%

113 to 393

Google Search Console · manual export
June 1-26, 2026 vs August 1-26, 2026
Paid spend-46%

$34,673.41 to $18,670.49

Google Ads, Microsoft Ads and Meta platform totals supplied by the operator
June 2026 vs August 1-28, 2026
Form submissions+48%

101 to 149

PostHog · lead_submitted events
June 1-26, 2026 vs August 1-26, 2026
Supporting evidenceBeforeAfterChangeSource / window
Google Search impressions16,88155,133+226.6%Google Search Console · manual export · June 1-26, 2026 vs August 1-26, 2026
Average Search position18.2413.504.74 positions betterGoogle Search Console · impression-weighted average · June 1-26, 2026 vs August 1-26, 2026
Q&A Search impressionsNo rows surfaced4,662New visibilityGoogle Search Console · Search Appearance export · June 2026 vs August 1-26, 2026
Local URLs surfacing69120+51Protected local-inventory Search Console analysis · June 2026 vs August 1-26, 2026
Weighted local position25.9718.257.72 positions betterProtected local-inventory Search Console analysis · June 2026 vs August 1-26, 2026

Eligible inventory also grew, from 178 URLs to 228. A percentage across a changed denominator would compare two different things, so the absolute move is shown instead.

Search acquisition

Google Search clicks grew 248%

113 → 393
113Jun 1-26247Jul 1-26393Aug 1-26
PeriodClicksImpressionsAverage position
Jun 1-2611316,88118.24
Jul 1-2624740,41815.21
Aug 1-2639355,13313.50

Full-month clicks before and at campaign start were 167 in March, 137 in April, 143 in May and 134 in June. The matched-period rise was not the continuation of an established monthly growth trend.

Google Search Console · manual export · matched 26-day periods
Acquisition mix
$34.7K → $18.7K
June$34,673.41
July$29,119.04
Aug 1-28$18,670.49
Google AdsMicrosoft AdsMeta
PeriodGoogle AdsMicrosoft AdsMetaTotal
June$22,600$6,310$5,763.41$34,673.41
July$15,800$4,978$8,341.04$29,119.04
Aug 1-28$11,500$3,564$3,606.49$18,670.49

The reduction happened across Google, Microsoft and Meta. August is measured through August 28, not a completed month.

Google Ads + Microsoft Ads + Meta · platform spend totals
Context, not an outcome

Paid spend fell during the engagement. We are showing it because it is true and dated, not because we can show it followed from the Search work. August is a partial month. This is context, not an outcome: the evidence does not show that SEO work caused the budget to fall, and the source analysis lists paid substitution as unknown. The source analysis lists paid substitution among the things it cannot answer, and this page does not answer it either.

Website demand

Website form submissions rose 48%

101 → 149
101Jun 1-26
116Jul 1-26
149Aug 1-26

These are PostHog `lead_submitted` events. They are not CRM-qualified leads, booked jobs or revenue.

PostHog · lead_submitted events · matched 26-day periods
What changed in Search

Growth came from a wider Search system, not one page.

Visibility expanded across commercial decision content, Answer-style results, local pages and structural/Foundation topics. The evidence shows the combined movement. It does not assign sitewide growth to one release or tactic.

Expanded commercial problem, method and cost content around real homeowner decisions.
Strengthened service-area coverage and checked local inventory URL by URL.
Improved Foundation and structural topic ownership through architecture and internal links.
Corrected technical and entity inconsistencies that made the business harder for search systems to interpret.
Kept Search, paid spend, forms, calls and AI observations in separate measurement systems.
Answer / AI Search expansion

Google Answer visibility and AI recommendations are different proof types.

Search Console records Answer and Q&A presentation inside Google Search. AI recommendation captures document individual model responses. We do not merge the two.

Google Answer / Q&A Search

Answer-style impressions reached 4,662

0 → 4,662
June
No rows surfaced
July
846
Aug 1-26
4,662

This is Google Search presentation evidence. It does not establish ChatGPT visibility.

Google Search Console · Search Appearance · June through August 26, 2026
Mentions
not published
No repeated, comparable mention rate has been published for this account. Historical runs exist but are not like-for-like.
Citations
not published
Caul is the only source permitted for a citation figure, and no comparable June baseline exists for the current prompt set.
Recommendations
dated examples only
Individual model responses that named 603 can be shown as dated captures, each with its exact prompt and capture date. One response is not a rank and does not generalise.

Only 17 of the August 55-prompt set overlap June, and provider modes and grounded denominators differ. The data supports dated snapshots, not a June-to-August AI trend, so none is shown.

Protected local inventory

More local pages began surfacing

69 → 120
June69of 178 eligible URLsweighted position 25.97
July78of 178 eligible URLsweighted position 19.69
Aug 1-26120of 228 eligible URLsweighted position 18.25

Eligible inventory expanded from 178 pages in June and July to 228 in August. The larger footprint did not produce proportional click growth.

Protected local-inventory analysis · Google Search Console · June through August 26, 2026
Measurement discipline

We don’t manufacture a blended “lead” number.

Search Console measures Search acquisition. Paid platforms measure spend. PostHog measures website form-submission events. CallRail measures calls. 603 does not have reliable CRM pipeline logging, so Caldrin does not invent booked-job or revenue attribution.

Search
Google Search Console clicks, impressions and Search presentation.
Paid
Google Ads, Microsoft Ads and Meta platform spend.
Forms
PostHog `lead_submitted` events, not qualified opportunities.
Calls
CallRail calls, reported separately from forms.
Unknown
Booked jobs, close rate, revenue and gross profit.

What this page does not show.

  • Search growth and the paid-spend reduction are separate measurements from separate systems. The evidence does not show that one caused the other.
  • This does not show that total lead flow held steady. Website form events rose; phone handling moved separately and is not presented here as a demand signal.
  • Form submissions are website events, not qualified leads, booked jobs or revenue. 603 has no reliable CRM pipeline logging, so none of those are claimed.
  • August is a partial month in every window that says so. It is not presented as a full-month comparison.
  • No single release or tactic is assigned sitewide causation. Several changes shipped in the same period, and one, the August 24 Service Areas redesign, is a named confound.
What shipped, and when

The work behind the numbers, dated.

Several changes ran in the same window. That is why no single item below is credited with sitewide growth, and why one of them is marked as a confound rather than a win.

June 1, 2026
SEO/AEO engagement began.
June 23-24
20 local service pages published.
July
No precise intervention can be reconstructed.Repository history begins July 31.
August 4
Seven waterproofing town pages published.
August 14
43-city and service-page cohort published.
August 18
Protected helical-cost page wording and FAQ updated.
August 19
20 pages across four towns published.
August 20
95 URLs submitted for discovery and recrawl.
August 23
LocalBusiness schema enrichment.
August 24
Service Areas redesign.Named confound: it changed the local surface mid-measurement.
August 26
Four unsupported services retired with 410.
August 28
Five Foundation Repair town pages completed as drafts only.
Next phase

Foundation Repair is the next deliberate growth lane.

The broad Foundation parent moved from 234 to 971 impressions and average position 31.83 to 18.90. It is becoming more eligible, while specific structural problem, method and cost content still carries much of the click demand.

Foundation impressions234 → 971Google Search Console · June vs Aug 1-26, 2026
PeriodClicksImpressionsAverage position
June023431.83
July155127.90
Aug 1-26297118.90

The next sprint

  • Broad Foundation Repair city pages
  • City pages for bowing walls, settlement and cracks
  • Local project proof and stronger internal topic ownership
  • Source and citation expansion
  • Repeated AI recommendation testing
See how Caldrin approaches Foundation Repair
Reading this for your own business

What transfers, and what does not.

This is one contractor, in one region, over three months. The numbers are theirs. What generalises is the shape of the problem and the way it was measured, not the size of the movement, and any firm telling you otherwise is selling you an average.

Local-service demand starts at a visible problem. Content built around the problem, the method and the cost is what earned the impressions here, not brand pages.
Service-area coverage is an inventory problem before it is a content problem. Eligible URLs and surfacing URLs are two different counts, and only one of them is a result.
Answer-style Search presence and assistant recommendations are separate systems. Google reporting one does not tell you anything about the other.
Keep the instruments apart. A blended lead number would have hidden that phone handling moved differently from form submissions here, which is exactly the thing worth knowing.
Growth this fast is normal only from a low base. Position 18 to 13 is real movement and is still not page one for most queries.
Your market, measured the same way

Want this mapped against your market?

The free report measures your market against named competitors and keeps each source, period and limitation attached to the finding.

Want to know who is behind the work? Meet Caldrin’s founder.