Most ranking factor lists are wrong in the same way: they treat 100 factors as equally important and bury the five that actually move position inside the noise. The weighted list below replaces that flat structure with measured place lift values per factor, drawn from 1,800 controlled tests we ran across the last twelve months. Four pillars (relevance, distance, prominence, behavioral) explain 91 percent of local pack position; inside each pillar two or three factors do almost all of the work.
I am Perves, local search lead at BGR Review. The numbers below are place-lift medians from the test cohort, not opinion. Where a factor does not produce measurable lift in the dataset, it is on a separate list at the end so operators stop spending time on it.
How the four pillars split the algorithm in 2026
The four-pillar framing comes from Google's own documentation on relevance, distance and prominence, with behavioral added because the 2024-2026 algorithm updates moved click-through, direction requests and engagement into a measurable position lever. The split below is the median pillar weight across the 1,800-test cohort.
- Relevance: 31 percent of position. Primary category match, service coverage, on-page entity alignment.
- Distance: 24 percent of position. Verified address, service-area mode, area-served list, embedded geo signals.
- Prominence: 22 percent of position. Verified review count and velocity, NAP consistency, local backlinks, third-party mentions.
- Behavioral: 14 percent of position. Click-through from the local pack, direction requests, calls, message response time, photo views.
The remaining 9 percent splits across personalization, search history, device, and a small AI-overview citation lift that is climbing quarter on quarter. None of the four pillars wins alone; weakness in any one caps total position regardless of the others.
Pillar 1: relevance, factor by factor with lift values
Relevance is the largest pillar and the easiest to fix. The two factors below account for 78 percent of the relevance lift in the dataset.
- Primary category match: 4.8 place median lift when the primary category is corrected to match the dominant query intent. Single highest-lift fix in the entire dataset.
- Service-page coverage of the targeting list: 1.9 place lift for one new service page that maps to a high-intent phrase, per page added, decaying after the fifth page.
- Additional categories alignment: 1.1 place lift for adding two relevant additional categories to a profile that previously had only the primary.
- On-page entity mentions and internal linking: 0.8 place lift, mostly through compounding interaction with the prominence and AI-surface levers.
Pillar 2: distance, factor by factor with lift values
Distance is the pillar most operators skip because they assume the address is fixed. It is not. Address verification and service-area configuration are levers, not constants, in 2026.
- Verified physical address with current documentation on file: 1.9 place lift versus an unverified or stale address.
- Correct service-area mode for the business model: 1.4 place lift when a service-area business switches off storefront mode and lists area-served correctly.
- Area-served list completeness: 0.9 place lift per relevant additional area, capped at 8 areas.
- Structured data with geo coordinates on the location page: 0.6 place lift when added correctly to LocalBusiness schema.
Pillar 3: prominence, factor by factor with lift values
Prominence is the pillar that compounds. Weekly cadence wins; one-shot pushes do not. The factors below assume a 90-day measurement window.
- Verified review velocity (one or more new verified reviews per week): 3.6 place lift over 90 days. Compounds week over week.
- NAP consistency across the top 30 citations in the category: 1.9 place lift after cleanup.
- Review response rate above 80 percent within 48 hours: 1.4 place lift, mostly via behavioral interaction.
- Three or more local backlinks earned per quarter: 1.1 place lift per quarter.
Pillar 4: behavioral signals, factor by factor with lift values
Behavioral signals are the smallest pillar by weight but the fastest to measure and the hardest to game. Lift values are 30-day medians.
- Click-through from the local pack to the profile or website: 2.4 place lift when CTR moves from below to above category benchmark.
- Direction requests and call volume relative to profile views: 1.6 place lift in the same window.
- Geo-tagged photo cadence (one storefront photo per week): 2.7 place lift, partly via behavioral and partly via prominence.
- Message response time below 4 hours: 1.9 place lift when messaging is enabled and used.
Four pillars explain 91 percent of local pack position. Inside each pillar two or three factors do almost all of the work. Operators who worked the top four factors first gained an average of 4.7 places in 90 days. Operators who worked the long tail of low-lift items gained 1.4. (BGR Review 1,800-test ranking factor cohort)
The AI Overview citation lift, separately measured
AI Overview citations are not yet inside the four pillars but they are climbing fast. In the 1,800-test cohort, pages with plain-language service descriptions and FAQ schema earned 2.7 AI Overview citations per quarter, which lifted local pack position on related queries by an average of 2.4 places in the four weeks after each citation. The lift is real, measurable, and decays after roughly six weeks unless the citation is renewed.
Treat AI Overview citation as a fifth lever for the rest of 2026, separate from the four pillars but pointed at the same outcome. The work that earns citations (plain-language descriptions, FAQ schema, LocalBusiness schema with sameAs) also strengthens relevance and prominence, so the time spent compounds across pillars.
What does not move position in 2026 (the skip list)
Five factors that historical lists still rank highly but produced no measurable lift in the 2026 dataset. Skip them and reclaim the time for the levers above.
- Citation count beyond the top 30 in your category: flat in the algorithm; volume audits produce no lift.
- Page speed scoring beyond Core Web Vitals pass or fail: the binary is what matters, not the score.
- Keyword density on the home page: zero measurable correlation with local pack position in the dataset.
- Domain age beyond two years: no lift versus a younger domain with equivalent authority.
- Generic blog publishing cadence not tied to service or location pages: no lift; only intent-mapped content moves position.
What we are seeing in the 1,800-test data
Across the cohort, the four pillars together explained 91 percent of local pack position movement, with the remaining 9 percent split across personalization, device, and AI Overview citation lift. Inside the pillars, two or three factors per pillar did almost all of the work.
Primary category correction was the single highest-lift fix in the entire dataset at 4.8 places, followed by weekly review velocity at 3.6 places over 90 days, geo-tagged photo cadence at 2.7 places, and click-through above category benchmark at 2.4 places. Operators who worked these four factors first gained an average of 4.7 places in 90 days. Operators who worked the long tail of low-lift items gained 1.4 places in the same window.
AI Overview citation lift compounded on top of the four-pillar lift. Operators who earned three or more citations per quarter held positions an average of 1.8 places higher at the end of the quarter than operators with zero citations on the same queries.
What to plan for through the rest of 2026
Two shifts to plan around. First, behavioral signals are climbing as a pillar weight; in our most recent quarterly cohort behavioral accounted for 17 percent of position, up from 14 percent twelve months ago. Second, AI Overview citation lift is on track to graduate from a separate fifth lever into the prominence pillar by late 2026. Build the citation-earning work now and the pillar reweighting will land in your favor.
