Dealing with power outages involves a tangle of conflicting
requirements. I suggest an iterative approach:
First, guess how much you'd like to spend, then decide what you'd 
like to power, then figure out how much it'll cost.  If the 
numbers are consistent, decision made. If the numbers are not 
consistent, adjust something and start over. 

Electrical power stands on a three legged tripod: 
The generator chosen, determined by what you want to power 
The power distribution system, set by home layout and what's to be powered 
The fuel supply arrangements, set by fuel consumption and storage constraints. 

For the price of a serviceable used car, a totally automatic whole-house
backup system can be installed. For a few hundred dollars, if you're
skilled and patient, a setup to keep the fridge cold can be assembled. 

Back in 2016 I found a Miller AEAD200LE welder-generator for $700. 
200 amp welder at 3000 rpm, 4.8 kW generator at 1800 rpm, fairly 
quiet.  I'd been wanting one for years, being a mechanical hobbyist. 
A recent tree fall in the neighborhood had cut my power for a couple 
of days, so the idea of a backup generator, for which service these 
machines are superb, seemed like an added bonus.  I didn't see PG&E's 
PSPS decision coming, but when it did the purchase seemed prescient. 
It provided enough power to run the fridge, coffeemaker, microwave and 
even the toaster oven. After fixing up the generator it looked like 
I was all set, but...... 

It quickly became apparent that the extension cords I had on hand were 
quite inadequate. It also developed that extension cords of gauge and 
reach to deliver power to the kitchen appliances were surprisingly 
expensive. 

My electrical panel was too old to be modified at all, but worked fine 
otherwise. Replacing it for what amounted to an experiment seemed a bit 
extravagant: "If it ain't broke, don't touch it".  I set up a 100 foot 
10/4 cable and 2-phase outlet box. The layout of my house made it possible 
to route the cable into the kitchen without causing any new air leaks.

In the clarity of hindsight, two errors became evident:

First, an extension cord setup, even when it's well done, is a nuisance.
It takes about an hour to deploy and another to put away, just to keep
the kitchen appliances and computers running. And, it cost ~$300 in parts.  

Second, for a lengthy outage, fuel use was excessive. The most essential 
appliance is the fridge. It needs only 700 nameplate watts and drops to
about 100 watts steady-state, but it's designed to run about 30 minutes
per hour. The AEAD-200LE burns .4 GPH at no load, maybe .5 GPH with fridge 
and computers connected. At 24 hr/day that's about 10 gallons per day, 
far too much, at 12 hr/day still too much. And, largely unnecessry.

Being able to run fridge, toaster, microwave and coffemaker at once is nice, 
but needless. My car has a 33 gallon fuel tank. Being old it's relatively 
easy to siphon gas out, so that's my tanker for now. Not a great setup, 
tolerable only if I limit generator runs to a timed schedule, no more than 
a few hours per day. Even so, I'll need to get gasoline about once a week 
and it has to come from a fifty-mile radius to leave a useful amount in the 
tank. 
 
For a project that started on a whim, the outcome isn't half bad.
I'd have bought the welder anyway. Equipping it as an emergency
generator with a week's endurance brought the total cost to ~$1100.
Good value for the money, but luck played a big role.  
 
Here's a more systematic approach to getting ready for outages.

Buying an emergency generator is at least partly an emotional decision.
How  much practical good it will do is impossible to predict, but its 
presence reduces anxiety whether it's used or not. That's worth something. 
By making a guess how much to spend, you're effectively measuring your 
level of anxiety. So, jot down how much you're comfortable spending and 
file the number away for future reference. If it's over the price of a 
nice used car, stop fretting and phone an installer. If less than around 
$1k, you'll probably be limited to what turns up second-hand. In either
case, it does limited good to simply trade power-failure anxiety for 
fuel-supply anxiety. To some extent that's what I've done.  

Next, decide what to power during an outage. Be careful to take into 
account motor startup surges for pumps, fridges and A/C.  Unless you 
can ensure that automatic starting loads can be kept from starting at 
the same time add the starting loads, not the running loads. Beware, 
peak capacity bites twice; once at adoption, again at feeding time. 

To a good approximation, fuel burn is

gph= .1 x (rated kW + load kW) 

In other words, capacity costs dearly in fuel consumption even
when it isn't being used. 

Next, decide how to distribute the power. A generator interlock, if it's
mechanically and legally possible, is probably the best method. A single 
cable runs from generator to breaker panel using an inlet receptacle and
cord. In the best case it might cost only a couple hundred dollars for 
the interlock, cord and receptacle.  

If your panel is incompatible, or your local codes forbid an interlock, 
the next-best choice is add a transfer switch or replace the panel. A 
new panel will be at least $2k, a small transfer sub-panel might be $500. 
That's what sent me down the extension cord route. 

Finally, decide how you're going to supply the required fuel. That's 
potentially the most expensive leg of the tripod. A day is easy, a week
takes some luck or planning. Much more means either local bulk storage
or a natural gas connection unless you're willing to live with very
limited electrical service. For my worry level, a week is enough, but
more is better if it comes readily to hand.


Now, compare the total cost to what you wanted to spend and make adjustments.  
Don't be surprised if the conclusion is "skip it". That's true for most folks.
Outages simply aren't that big a problem.

Here's what I meant by saying peak capacity bites twice:

Miller AEAD200LE
1.0 gph at 4.8kW
0.4 gph at no load 

Honda eu2200i
.3 gph at 1.8kw  .17 gph/kw
.12 gph at .45 kw .27 gph/kw

Honda eu1000i
.19 gph at .9kW
.084 gph at .36 kW

Inverter machines have relatively low no-load fuel consumption, which
minimizes the fuel supply problem. More generally, it's fair to 
estimate any generator will use about .1 gph per kW of rated 
capacity and will burn about another .1 gph per kW of actual
load. Overhead valve engines with inverter generators like Hondas 
tend to be a little better, L-head synchronous machines like mine 
a little worse.  The difference won't be enough to justify a 
replacement outright, but it's worth taking into account when 
shopping.

Likewise, I wouldn't go out and replace a working fridge to lower
the starting surge requirement, but if you're shopping anyway it's
a helpful thing to minimize. Being able to use an eu1000i instead
of an eu2000i saves a couple hundred dollars up front and at least
that much (possibly much more) in fuel supply arrangements.

Using fuel in a car's gas tank has several advantages: It's 
already on-hand, safe, bigger than most portable tanks and fresh. 
On the downside, you will have to deal with gas transfer to the 
generator, a small hazard and large nuisance. Old cars like mine
are fairly easy to siphon from. New cars that have anti-spill
features in the filler neck can be close to impossible.  
 
Barbecue propane tanks are probably too small to be very useful.
The trouble is that a 5 gallon tank is equivalent to only about 3.5 
gallons of gasoline and the tanks are heavy to boot. Not useless 
if already on hand, but hardly worth buying extra. Adding a gaseous 
fuel adapter to a gasoline generator is a prudent hedging of bets 
if you have gaseous fuel available.  Still, it's not cheap; a few 
hundred dollars for the demand regulator and associated fittings in kit 
form. Plus installation, which can be tricky on enclosed generators.

Portable fuels get scarce in emergencies. A much better option 
is bulk propane or a natural gas connection. The setup cost is
substantial, easily more than the generator, unless you already
have a gas tap of sufficient capacity close to an acceptable 
generator location. A barbecue connection might be sufficient 
for a small generator, typically no more than 50k btu/hr for a
portable grill, enough for roughly a 2 kW generator.

A whole house generator  will need its own line. For example, a 
6.5 kW Onan is said to need 150k BTU/Hr at full load, over 1.5 
gallons of propane per hour. The zero load burn will be near 1 gph. 
A 500 gallon tank will last only about ten days running continuously. 
And, you'll have to persuade somebody to come to your place to fill 
it again. Neither is a concern with piped natural gas.

Limiting run time is the obvious economy, but there are limits. 
Fridges cool down slowly, with the best run time concentrated 
right after mealtimes and before bed. Running at night is fraught: 
The noise will irritate neighbors. Problems with the generator 
can easily lead to unfixable harm to equipment and inhabitants. 
Running generators out of oil is a leading cause of destruction. 
The "low oil" protections usually act too late to prevent serious 
damage. A wind shift is enough to drift exhaust into buildings. 
Hard wired whole house backups are designed to run unattended. 
Portable generators and welding machines are definitely not.

The economics of home power generation are surprisingly bad. 
For example, my generator is rated at 4.8 kW burning 1 gallon 
per hour. That's $1.25 worth of electricity for $3.30 worth 
of gasoline. Even worse, it burns about .5 gph at zero load.
A modern generator is slightly better, but not enough to repay 
the added cost under plausible usage scenarios. Natural gas is 
much cheaper, but the piping hookup is unlikely to pay for itself. 
It will, however, pay dividends on the worry ledger.

Generator type selection

Once the overall size of the generator is set one still has the
choice of new or used and synchronous or inverter.  Used equipment 
can be a very good buy, with some caveats: Fuel use is apt to be 
higher, in  particular. On the upside, old RV generators that run 
at 1800 RPM are relatively quiet and widely available, often nearly 
unused. They're also designed for starting air conditioners, a very
useful ability. Those with "blow out" cooling can be sheltered for 
weather and noise protection. 

Realistically, a portable inverter generator bought new is likely 
the most expedient  option. Noise and fuel consumption at light 
load is usually better with inverters, though they can be screamy 
at full power. Small synchronous generators are obnoxiously loud 
all the time. Inverter units under 2kW should be tested before 
purchase for refrigerator use, most won't work. 

The biggest drawback to inverter generators is fragility. It's
difficult to design protection circuits which can prevent damage 
to the inverter without nuisance tripping. A damaged generator 
is typically non-repairable, regardless of type. Synchronous 
generators are somewhat  harder to kill. 
 
One of the more surprising ideas to emerge is the use of two
generators. One large conventional unit, run only to support
hard-starting high current loads like  pumps, air conditioners
and maybe electric cooking appliances. A second, much smaller
inverter unit could supply long duration loads like refrigerators,
electronics and lights, none of which draw much power. This would 
make sense only where outages are frequent and prolonged, and
pipeline or bulk fuel is impractical.  

Perhaps a better setup is that used on RVs and yachts: An inverter
charger passes AC when available to loads and charges a battery 
bank. When grid power goes away, the inverter picks up the small loads.
The genny can  fire up at the resident's leisure, with continuous power 
to essential loads like fridge and communications gear. When high power
loads are needed like cooking or airconditioning the generator can run. 
Absolute fuel efficiency won't be great, but reduced runtime relieves
some of the pressure on fuel stock.  Continuity of power means the fridge 
stays cold, the Net is up and you can figure out what to do with some 
time to spare.

Costwise it's roughly a wash; an inverter/charger and battery bank
will touch $1k if lead-acid, around $2k if lithium. Not much different
than a second generator. But, far more convenient. Still, a lot of 
money for a "what if" scenario. 
  
My emergency power setup bloomed out of a whim and a Craigslist
advertisement. Until  more practical experience can be gained,
it's good enough. I've seen only four bothersome outages from 
2005-2020. The system was tested in earnest once, in 2019. It worked
well, but the power came back on very shortly after the setup was
deployed, so no fuel-handling experience was gained.

If the house service panel needs to be replaced I'll add a transfer switch 
or interlock. If outages become routine a natural gas hookup is next. 
In the meantime, the extenion-cord-and-siphon will suffice. A fully
automatic system won't become of interest unless I'm unable to handle
the rigors of portable generator setup/stowage and outages become 
common. If I need to replace the fridge, I'll pay more attention to 
starting power needs. 

I've since added two inverter-charger-battery units, one for the 
fridge and a second for communications equipment.  They have only 
overnight endurance, but that's enough for now. They've let me
simply ignore a couple of brief outages that didn't interfere 
with mealtimes.  

20240122