Geo keyword intelligence
Predict the queries a domain could rank for, then measure demand and commercial value in the exact geography.
This stage runs after pendingDelete is confirmed and the domain passes safety and history checks, but before final scoring.
Keyword-generation process
Generate candidates from five sources:
- Parsed domain entities — service, city, county, state, and brand terms found in the name.
- Service taxonomy — primary services, subservices, materials, problems, and job types.
- Commercial modifiers — contractor, company, installer, repair, replacement, cost, quote, near me, emergency, and best.
- Local modifiers — city, neighborhood, county, metro, ZIP, state abbreviation, and nearby cities inside the realistic service radius.
- SERP expansion — related searches, People Also Ask topics, competitor keywords, and common keyword combinations.
Example for a South Jordan garage-floor domain:
garage floor coating south jordan
garage floor epoxy south jordan
polyaspartic garage floor south jordan
garage floor coating near me
garage floor coating cost south jordan
epoxy garage floor salt lake county
garage floor coating draper
garage floor coating riverton
Prevent keyword explosion
Do not submit every possible combination.
Apply rules before the SEO API:
- service must match the approved taxonomy
- geography must be inside the intended service area
- phrase must have clear local or commercial intent
- remove unnatural word order
- deduplicate close variants
- group singular, plural, and spelling variants
- cap initial keyword candidates at 200 per domain
- expand only the strongest clusters
Geo-specific SEO API lookup
Use DataForSEO Keywords Data or Google Ads search-volume endpoints with:
- keyword
- location code or location name
- language code
- search network
- device when required
Store:
keyword
normalized_keyword
service_cluster
intent
location_code
location_name
language_code
monthly_search_volume
monthly_search_history
cpc
competition
competition_index
low_top_of_page_bid
high_top_of_page_bid
api_timestamp
api_source
Geographic fallback ladder
City-level volume is often sparse. Query demand at several levels:
- city
- county
- metro or DMA
- state
Never treat missing city volume as zero demand. Estimate the city share from broader data and label it as modeled.
estimated_city_volume =
metro_volume
× city_population_share
× service_demand_adjustment
× local_SERP_adjustment
Keep the measured and modeled values separate.
Commercial value model
CPC is a useful commercial-intent signal. It is not the value of the keyword.
expected_monthly_clicks =
geo_monthly_volume × expected_organic_CTR_at_target_position
expected_monthly_leads =
expected_monthly_clicks × site_conversion_rate
expected_qualified_leads =
expected_monthly_leads × qualified_lead_rate
expected_monthly_lead_value =
expected_qualified_leads × value_per_qualified_lead
For an owned or joint-venture service company, replace lead value with expected gross profit per closed job:
expected_monthly_gross_profit =
expected_monthly_leads
× appointment_rate
× close_rate
× gross_profit_per_job
Keyword opportunity score
keyword_opportunity =
normalized_demand
× commercial_intent
× expected_lead_value
× local_relevance
× rankability
× evidence_confidence
Calculate domain-level keyword value from the strongest non-overlapping clusters. Do not add hundreds of closely related keyword volumes because this double-counts the same demand.
Confidence rules
| Evidence | Confidence |
|---|---|
| Direct city-level API volume with monthly history | High |
| Metro volume allocated with reliable city share | Medium |
| State volume allocated to a small city | Low |
| CPC but no usable volume | Low |
| AI-predicted keyword with no SERP or API validation | Very low |
The scoring system must penalize low-confidence demand estimates.