Prospect Segment & Category Attributes — Digital-Footprint Profiles

A digital-footprint profile of every business segment (and category) in the prospect universe: how many are listed vs NOF, how many have a Google place_id / website / were crawled, what tech and CMS they run, how deep their sites are, and how many sit on a dead domain. This is the raw material for product targeting (e.g. which segments are greenfield for BizSite, which already run ecommerce) and for segment-level opportunity sizing.

Built by scripts/crawl-tail/build-segment-attributes.cjs (segment grain) and build-category-attributes.cjs (segment × category grain). Keyed on the canonical b_segment (17-segment scheme) resolved via gp_category_segment + the CATEGORY_MAPPING Google→GP bridge, plus an Uncategorised residual (= 18 rows). Snapshot: 2026-06-03, 228,832 in-universe rows (grown from ~210K via the book / SERP-harvest sweeps — both tables below share this universe). Related: Prospect Enrichment & Opportunity Index, Prospect DB — Composition.

---

Segment profiles (17 segments + Uncategorised, in-universe — 228,832 total)

place_id% / website% are of the segment’s universe; crawled% is of the segment’s websites — crawl coverage of the addressable site pool (has_crawl_data ÷ with_website), not of universe, so a low-website segment doesn’t read as “uncrawled”. ecom / analytics are listing counts with that signal detected. dead = domains confirmed NXDOMAIN.

Segment	Universe	NOF	place_id%	website%	crawled% (of sites)	Main CMS	ecom	analytics	dead
Home Services	37,981	6,451	35.9%	41.0%	90.9%	WordPress	2,213	9,706	800
Retail	36,711	6,748	45.9%	47.5%	85.6%	WordPress	5,079	10,457	541
B2B	32,339	6,158	43.7%	55.2%	89.1%	WordPress	3,349	10,882	683
Other	22,819	6,624	51.6%	54.9%	90.3%	WordPress	2,554	7,965	430
Health & Wellbeing	19,484	4,655	47.8%	53.9%	88.2%	WordPress	1,757	6,512	360
Eat, Drink, Takeout	13,688	4,060	59.8%	43.8%	81.5%	WordPress	968	3,095	113
Finance, Property & Legal	12,481	2,111	51.5%	58.2%	89.6%	WordPress	478	4,598	114
Vehicle	11,729	2,813	49.6%	53.4%	87.6%	WordPress	612	4,090	183
Personal Care	7,929	1,931	56.9%	40.1%	80.3%	WordPress	689	1,749	125
Travel	7,917	1,845	51.4%	50.5%	88.9%	WordPress	389	2,303	91
Leisure	6,280	1,847	53.3%	59.2%	84.5%	WordPress	589	2,189	63
Learning	4,886	1,247	59.3%	63.3%	88.2%	WordPress	460	1,966	57
Gov	4,756	379	30.5%	78.9%	93.9%	Drupal	50	3,018	5
Agri	2,953	485	39.0%	31.1%	90.0%	WordPress	271	572	26
Personal Development	2,830	420	35.5%	63.4%	93.9%	WordPress	231	1,432	26
Uncategorised	2,061	349	21.7%	90.9%	91.7%	WordPress	146	800	216
Communication	1,011	178	43.0%	52.9%	89.0%	WordPress	109	286	34
Insurance Agents, Brokers & Consultants	977	164	52.3%	67.3%	90.6%	WordPress	42	395	5

Reading the profiles:

• WordPress dominates every segment except Gov (Drupal). CMS mix (cms_* columns in the CSV) is a build/rebuild-fit signal — DIY-builder share (Wix/Squarespace) vs custom.
• Low website% = greenfield (Personal Care 40.1%, Agri 31.1%, Home Services 41.0% have a site) — the no-website pool for entry-level web products. (Crawl coverage of the sites that do exist is uniformly high — 80–94% across every segment — so crawled% now reads as completeness, not gap.)
• High ecommerce = storefront segments (Retail 5,073, B2B 3,348, Home Services 2,213) — where a one-page site is not sufficient.
• analytics / with_g_ads / pixel columns flag already-marketing-active businesses (warmer for paid services).

Crawl coverage to date (by segment)

Of each segment’s websites (in-universe, http*), the crawl outcome so far plus the signals detected. Outcome buckets are mutually exclusive (priority dead › social-only › blocked › slow/other-error › fully-crawled › pending) and sum to Websites. Social-only = the “website” is a Facebook/Instagram/etc. page, so there’s no real site — never crawled (a SitePro/BizSite greenfield signal, not a backlog); pending = a real site still queued. Feature & CMS counts are what the crawl detected (a site can carry several; CMS counts are among reachable sites). Ecom* is the shallow-crawl ecommerce signal — pre full-crawl confirmation. Built by scripts/crawl-tail/build-crawl-coverage.cjs → crawl_coverage_by_segment.csv.

Refreshed 2026-06-08 — shallow crawl complete. This table is re-run on the finished shallow sweep, so Pending has drained (11,191 → 4,774) and Fully crawled has risen (81,956 → 87,283, 70% → 75%). The Segment profiles and Contactability tables above/below still carry the 2026-06-03 snapshot — they’ll be refreshed once the deep (full) crawl completes, since deep-crawl results confirm Ecom* and the page-depth signals they depend on. Deep crawl in flight: ~33K done, ~64K reachable sites still queued at ~630/hr.

Segment	Websites	Dead	Social-only	Blocked	Slow/other	Fully crawled	Pending	Booking	Ecom*	GA4	Google Tag	Meta	TikTok	WordPress	Wix	Shopify	Square	Other CMS
B2B	17,854	683	150	1,056	1,931	13,238	796	1,250	3,401	11,577	1,261	1,193	60	6,278	993	478	496	452
Retail	17,452	541	579	1,383	1,552	12,605	792	1,325	5,410	11,280	2,178	1,886	124	3,877	677	2,291	306	251
Home Services	15,526	800	356	834	1,613	11,396	527	1,222	2,176	10,102	999	924	39	5,357	1,444	213	340	350
Other	12,521	430	236	855	1,193	9,400	407	1,369	2,576	8,388	927	1,068	59	4,330	792	355	451	392
Health & Wellbeing	10,505	360	216	472	991	8,032	434	2,868	1,771	6,927	468	890	33	3,702	957	103	525	245
Finance, Property & Legal	7,263	114	11	469	716	5,675	278	323	455	4,913	322	405	11	2,754	320	5	188	124
Vehicle	6,319	183	145	354	434	4,755	448	687	607	4,259	316	1,435	63	1,446	276	78	64	127
Eat, Drink, Takeout	5,996	113	532	355	485	4,269	242	943	987	3,333	231	576	46	1,516	293	180	339	57
Travel	3,997	91	32	230	442	3,039	163	1,590	402	2,484	307	369	9	1,221	153	14	92	70
Gov	3,751	5	7	98	349	3,231	61	80	50	3,163	18	60	0	486	15	0	7	818
Leisure	3,718	63	89	342	266	2,794	164	687	590	2,477	143	471	19	1,215	174	48	139	80
Personal Care	3,179	125	377	135	263	2,164	115	1,067	714	1,848	114	289	11	784	189	118	148	31
Learning	3,094	57	31	150	354	2,360	142	300	461	2,114	161	397	28	1,087	197	27	146	68
Uncategorised	1,873	216	19	112	330	1,122	74	90	166	850	82	110	3	367	51	46	26	42
Personal Development	1,795	26	13	60	99	1,545	52	143	227	1,468	41	88	2	704	235	6	105	42
Agri	919	26	33	44	86	701	29	74	268	595	53	75	5	345	70	55	20	16
Insurance Agents, Brokers & Consultants	658	5	1	25	44	557	26	30	41	420	23	63	3	280	20	1	6	22
Communication	535	34	8	14	55	400	24	31	112	304	46	38	0	164	19	25	10	16
TOTAL	116,955	3,872	2,835	6,988	11,203	87,283	4,774	14,079	20,414	76,502	7,690	10,337	515	35,913	6,875	4,043	3,408	3,203

Reading the crawl (2026-06-08, post-shallow):

• Coverage is now 75% and clean — of 116,955 sites, 87,283 (75%) fully crawled (up from 81,956/70%), ~3.3% dead, ~6.0% blocked, ~9.6% slow/other-error. With the shallow sweep finished, pending has collapsed to 4,774 (~4.1%) — real sites still queued (the small residual tail + newly-added rows), down from 11,191 — and social-only (2,835, ~2.4%) — not a backlog, see below.
• Social-only (2,835) = greenfield, not a crawl gap — these listings’ only web presence is a Facebook/Instagram page, so there’s no site to crawl (deliberately excluded). They’re a BizSite / SitePro prospect list (their social URL is in Superset urls_urls): Retail 579, Eat/Drink/Takeout 532, Personal Care 377, Home Services 356, Other 236, Health 216, B2B 150, Vehicle 145 lead.
• Blocked (6,988) + slow/other (11,203) = the recovery backlog — the targeted deep-crawl and the Ahrefs-on-403 fallback chip away at these rather than leave them dark.
• Ecom* (20,414) is the shallow signal, not confirmed storefronts — the full crawl validates it; Retail (5,410) and B2B (3,401) lead.
• GA4 (76,502) is near-universal; ad pixels are the gap — Google Tag 7,690, Meta 10,337, TikTok just 515. Pixel presence is a paid-readiness signal (warmer for ads).
• WordPress (35,913) is ~4× the next platform; Shopify (4,043) concentrates in Retail, Gov skews to “Other CMS” (818, mostly Drupal/gov platforms).

Contactability by segment

A full-universe consent split — every in-universe listing lands in exactly one of five mutually-exclusive tiers, so the whole prospect base is accounted for in one table (the tiers sum to Universe). Reach needs an email — Email is the only marketing-reachable tier; a listing with only a phone is a call list, split mobile (call/SMS) vs landline (call only). Email % is of the segment’s universe. Built by scripts/crawl-tail/build-contactability-tiers.cjs (re-run any time → contactability_tiers.csv). Snapshot 2026-06-08. The same split is also cut into two children below — On file (a record we hold) and Not on file (net-new).

Segment	Universe	Email	Email %	Mobile-only	Landline-only	Opted-out	Dark
Home Services	37,886	14,509	38%	6,417	7,173	9,554	233
Retail	36,711	12,532	34%	3,013	11,785	8,930	451
B2B	32,339	12,148	38%	3,092	7,438	9,280	381
Other	22,819	8,923	39%	2,643	5,004	5,533	716
Health & Wellbeing	19,484	7,712	40%	2,659	4,447	4,384	282
Eat, Drink, Takeout	13,688	3,253	24%	995	7,166	1,773	501
Finance, Property & Legal	12,481	4,841	39%	483	2,925	4,185	47
Vehicle	11,824	3,738	32%	1,784	2,430	3,721	151
Personal Care	7,929	2,309	29%	1,252	2,909	1,302	157
Travel	7,917	2,784	35%	716	1,841	2,315	261
Leisure	6,280	2,199	35%	753	1,578	1,316	434
Learning	4,886	1,902	39%	490	1,024	1,402	68
Gov	4,756	1,100	23%	519	811	2,315	11
Agri	2,953	728	25%	242	1,298	610	75
Personal Development	2,830	1,220	43%	449	733	386	42
Uncategorised	2,061	730	35%	151	336	745	99
Communication	1,011	318	31%	78	330	281	4
Insurance Agents, Brokers & Consultants	977	416	43%	20	188	347	6
TOTAL	228,832	81,362	36%	25,756	59,416	58,379	3,919

Universe-wide: of 228,832, only 81,362 (36%) are email-reachable — the real marketing-reach number. A further 25,756 (11%) are mobile-only and 59,416 (26%) landline-only (call lists, no email). 58,379 (26%) are opted-out of marketing (correctness + onboarding only), and 3,919 (1.7%) are dark. The five tiers are mutually exclusive and sum to Universe. (Counts drift slightly as the live crawl keeps adding emails; the table is one consistent run.)

The five tiers (priority — opted-out pulled out first, then email > mobile > landline):

• Email (81,362, 36%) — not opted-out, and has a usable email: a consented email (opted_in = TRUE in mc_opted_in_20260326, the Mailchimp consent snapshot), or — for a prospect with no opt-out on record — a crawl_email (our webcrawl of the site) / outreach_email (harvested from the SerpAPI/competitor archive). This is the only marketing-reachable tier.
• Mobile-only (25,756, 11%) — not opted-out, no usable email, but has an Irish mobile (083/085/086/087/089, +3538x). A call/SMS list, not email-reachable.
• Landline-only (59,416, 26%) — not opted-out, no usable email, no mobile, but has a landline. Call-only.
• Opted-out (58,379, 26%) — subscriber in the consent DB with opted_in = FALSE: a recorded marketing “no”. No marketing of any kind (email or call) under the conservative framing — contactable only for listing-correctness and onboarding (service basis), via phone or a non-marketing email. Heaviest in Home Services (9,554), B2B (9,280), Retail (8,930), proportionally Gov (49%). The cohort the marketing-email DB exists to protect — never fold it back into a marketing send.
• Dark (3,919, 1.7%) — not opted-out, but no email of any kind and no phone anywhere. The pl_original net-new harvest tail with no website, no place_id and no phone — the only genuinely unaddressable residual.

Child 1 — On file (199,026: a GP listing or a subscriber)

Businesses we already hold a record for — a GoldenPages listing_id or a subscriber_id (current/past customers + the GP directory). This is the correctness, re-engagement and upsell base: we know who they are. Every opted-out row lives here (opt-out is subscriber-keyed), and dark is near-zero (105) — we almost always have a phone on file. Email reach 71,289 (36%).

Segment	Universe	Email	Email %	Mobile-only	Landline-only	Opted-out	Dark
Home Services	33,606	12,794	38%	4,947	6,289	9,554	22
Retail	33,623	11,719	35%	2,196	10,761	8,930	17
B2B	28,760	10,643	37%	2,337	6,484	9,280	16
Other	17,634	7,011	40%	1,402	3,673	5,533	15
Health & Wellbeing	16,330	6,469	40%	1,694	3,777	4,384	6
Eat, Drink, Takeout	11,090	2,745	25%	555	6,015	1,773	2
Finance, Property & Legal	11,564	4,404	38%	369	2,600	4,185	6
Vehicle	10,031	3,236	32%	1,042	2,025	3,721	7
Personal Care	6,782	2,090	31%	732	2,655	1,302	3
Travel	6,834	2,470	36%	437	1,609	2,315	3
Leisure	5,100	1,856	36%	497	1,428	1,316	3
Learning	4,115	1,603	39%	334	776	1,402	0
Gov	4,675	1,084	23%	513	762	2,315	1
Agri	2,617	636	24%	146	1,224	610	1
Personal Development	2,582	1,130	44%	371	692	386	3
Uncategorised	1,869	714	38%	120	290	745	0
Communication	915	300	33%	34	300	281	0
Insurance Agents, Brokers & Consultants	899	385	43%	11	156	347	0
TOTAL	199,026	71,289	36%	17,737	51,516	58,379	105

Child 2 — Not on file (29,806: net-new, no listing or subscriber)

Truly net-new prospects — no GP listing_id and no subscriber_id (≈ FCR’s operational NOF; the pl_original / SERP-harvest pool). The cold-acquisition base. No opt-out is possible here (opt-out is subscriber-keyed → column is 0), so the consent question is open; but it carries all but ~105 of the dark residual (3,814) — net-new with no channel at all. Email reach 10,073 (34%).

Segment	Universe	Email	Email %	Mobile-only	Landline-only	Opted-out	Dark
Other	5,185	1,912	37%	1,241	1,331	0	701
Home Services	4,280	1,715	40%	1,470	884	0	211
B2B	3,579	1,505	42%	755	954	0	365
Health & Wellbeing	3,154	1,243	39%	965	670	0	276
Retail	3,088	813	26%	817	1,024	0	434
Eat, Drink, Takeout	2,598	508	20%	440	1,151	0	499
Vehicle	1,793	502	28%	742	405	0	144
Leisure	1,180	343	29%	256	150	0	431
Personal Care	1,147	219	19%	520	254	0	154
Travel	1,083	314	29%	279	232	0	258
Finance, Property & Legal	917	437	48%	114	325	0	41
Learning	771	299	39%	156	248	0	68
Agri	336	92	27%	96	74	0	74
Personal Development	248	90	36%	78	41	0	39
Uncategorised	192	16	8%	31	46	0	99
Communication	96	18	19%	44	30	0	4
Gov	81	16	20%	6	49	0	10
Insurance Agents, Brokers & Consultants	78	31	40%	9	32	0	6
TOTAL	29,806	10,073	34%	8,019	7,900	0	3,814

(The two children sum to the parent, segment by segment. On file vs not-on-file = a GP listing_id or subscriber_id present vs neither. Built by the same build-contactability-tiers.cjs run → contactability_tiers_onfile.csv / contactability_tiers_notonfile.csv.)

Consent rules baked in:

• The flag is the consent, not the table. mc_opted_in_20260326 holds both TRUE and FALSE rows — presence is not consent. Email-marketable = opted_in = TRUE, or (for a prospect with no opt-out record) a crawl_email/outreach_email. We never email a recorded opt-out, and the raw Superset listing emails field is not a separate channel — opted_in is the consent layer over it.
• Opt-out is subscriber-keyed. Net-new NOFs / non-subscribers have no subscriber_id, so they can be neither opted-in nor opted-out — they reach the Email tier only via a crawl_email/outreach_email; otherwise they fall to the mobile/landline tiers or dark.
• Phone is drawn from five sources, deduped: spine phone, website crawl_phone, the GoldenPages listing phone (dim_listings_with_subscriber_id.phnumbers by listing_id), the GBP phone (dim_place.phone), and the same listing phone looked up by subscriber_id (the fallback for account-only stubs whose listing_id is null — this recovered ~12.8K stubs that had read as falsely “dark”).

What changed (2026-06-08). This replaces the earlier reachable-waterfall + unreachable-tail tables, and tightens “reachable” to email-only. The old waterfall counted phones (and even opted-out crawl-emails/mobiles) toward “reachable”, inflating it to ~130K/57%. The tiers instead pull opted-out out first, require an email for the marketing-reachable tier, and account for the whole universe in one place. Net: true email reach is 81,362 (36%); phone-only is surfaced as mobile 25,756 + landline 59,416 (call lists, not email-reachable); 58,379 opted-out are an explicit correctness/onboarding cohort; 3,919 dark. The same split is then cut into two children — On file (199,026) vs Not on file (29,806) — below.

Column dictionary (full CSV)

The CSVs carry ~45 columns per row. Beyond the summary above:

• Counts: listed, nof, universe, with_place_id, place_id_synthetic (LST: stubs), with_website, crawled.
• Tech signals: with_booking, with_ecommerce, with_analytics, with_g_ads, with_meta_pixel, with_tiktok_pixel, with_schema.
• CMS distribution: cms_wordpress / cms_wix / cms_shopify / cms_squarespace / cms_drupal / cms_webflow / cms_joomla / cms_duda / cms_mono / cms_other / cms_none, plus the derived main_cms.
• Site depth (page-count buckets): pages_0_index_unwalked, pages_0_empty, pages_0_other, pages_1, pages_2_5, pages_6_10, pages_10plus, plus pages_no_website, pages_site_uncrawled, pages_crawled_no_count and pages_known_pct_of_sites.
• Dead-domain: dead_nxdomain, dead_on_gp, dead_on_gmb, active_client_dead (a paying client on a dead domain — escalate).

Category grain

build-category-attributes.cjs produces the same columns at (segment × category) grain — 4,446 rows, one per primary_category — ordered by universe. It aggregates into a BigQuery table then paginates out past the 500-row API cap. Use it to drill a segment into its categories (e.g. Eat, Drink, Takeout → Restaurants 4,030 / Public Houses 3,881), which is the grain product targeting actually works at — see the BizSite category-prioritisation analysis. Full data: segment_category_attributes.csv / the category-attributes BQ table.

Regenerate both with the build scripts; counts drift as the crawl/enrichment refreshes.