
Findings according to this new methodology:
- San Francisco talent pool is in fact stronger than average (but not as strong as LA),
- San Diego lost the most ground going from #1 to #3 (expected given the lower weighting of the city fees - San Diego has the lowest fees by far),
- Los Angeles business climate is least business friendly due to a combination of highest rents (class A, desirable neighborhoods) and 2nd highest fees (yet still half San Francisco),
- Due to a severely weak talent pool, Sacramento swapped places with San Francisco to be the least desirable location,
- Interestingly, when pure rankings are the only factor taken into account (SUM RANK), Sacramento is ranked #2 overall due to its cost of living, cheap rents, and relatively cheap city fees (and San Jose falls from #2 to #4 due to its average performance in all three categories),
- High scores are best (previous version had low scores as best)
- Weighted raw data as Cost of Living (25%), Talent Pool (50%), Business Climate (25%) to reflect every business' need to find and retain the best talent (indexed against San Francisco)
- I updated the overall methodology to include weightings for population size and a more refined education scoring. This is expressed in the methodology in two ways, 1. nearby 4-year colleges (overall, large, & w/ MBA programs), and 2. total population with 4-year and graduate degrees.
- I also changed the weightings on the business climate to calculate the potential rent and city fees as percent of Gross Revenue. The reason is that expressing city fees and rents seperately over-weighted the city fees portion significantly given the huge discrepancy between San Francisco and other cities. For example, using the earlier example of $15MM revenue, the assumed (in San Francisco) rent would be $700,000 per year and city fees would be $135,000 - a 5:1 ratio. However, if seperated and indexed against other cities, city fees were outweighing rent as a factor by ratios of anywhere from 5:1 to 20:1 -- in other words, city fees were swinging the results by up to 100 times (5:1 x 20:1) leading to disproportionate results.
- Lastly, I dropped Phoenix from the evaluation. While peaking interesting personally, I did not have the time to continue working on non-relevant data.





