That's what I also use. I uploaded the schematic as PDF as well. The hierarchical sheet feature was what I was looking after but I did not know how to enable it.
Ah, that's a bit better.
That's what I intend to do, balancing MTTF and performance against cost but idk how. Where can I read more about it? What are things I can change from the above to achieve a good cost performance ratio?
How many are you making, and how much do you consider your time to be worth? You're pushing pennies here.
Can the max ripple current be assumed to be three times that of the max DC load current? The max current in the above design will be 5A.
I don't see where you're going to get 5A out of it. Third and fourth paragraphs of the overview, thanks to Google Translate:
The IP5328P's synchronous switching boost system provides up to 18W of output capability, maintaining an efficiency of over 90% even when the battery voltage is low. Automatically enters hibernation when unloaded.
The IP5328P's synchronous switch charging system provides up to 5.0A of charging current. Built-in IC temperature, battery temperature and input voltage control loop to intelligently adjust the charging current.
Since there's only one of those boost systems in the chip, 18W is probably a whole-system specification, not a per-port spec. VSYS doesn't participate in charging, only output.
With that cleared up, according to p39 of
a presentation in Avnet's 2012 Power Forum virtual conference, the output capacitor's rms current can be estimated by:

At the maximum boost of 12V at 1.5A with a worst-case 3V battery voltage, that works out to ~2.4A output cap ripple current for the entire bank. So, what's the ESR look like?

Murata has built
SimSurfing, a comprehensive selection and data tool for their capacitors. I chose GRM21BR61E226ME44, the general purpose line's 0805 22µF/25V for the caps on VSYS. At 500kHz, the given ESR is about 2mΩ each.
Let's look at a "low impedance" cap example, Panasonic's FC series, claiming 1/2 the impedance of the regular high-temperature HA series. The impedance of our 220µF 25V cap in a 10x10mm case is given as 0.15Ω, and permissible ripple current at 100kHz at 105°C is 670mA. For the standard high-temperature HA series, let's take their word for it and figure 0.3Ω. Not accounting for copper losses which are on the order of the ceramic caps' ESR, we can figure the ceramic caps are each going to see 0.3/0.002 = 150x the ripple current of the elcap, leaving 2.4/451 = 5mA for the elcap. No problem here.
Having done all that I can say that the use of the 220µF/16V cap is perfectly fine from the ripple current standpoint, but has a slight impact on service life.
So how much ripple voltage will we see? Our boost converter is rated for 18W. I'm gonna handwave and use the common rule-of-thumb inductor ripple current estimate of 40%p-p of the maximum inductor current, which is about (18W/3V)=6A*0.4=2.4A. Now let's shove that up the 66µF ceramics for about (1/300kHz) = 3.3µS * (3/12) = ~0.8µs * 2.4A = ~1.9µC of charge applied. A capacitance of one farad will increase its voltage by one volt when one coulomb of charge is applied to it. Our 2µC/66µF = 1/33V = 30mVp-p. Most devices won't have a problem with that!
Now let's multiply by 1.5A max current = 45mWp-p = ~15mWrms. That's a fair bit of power to have to deal with from an EMC standpoint. Poorly dressed, cheap cables might cause interference and undesired operation. The extra capacitance will reduce the ripple to about 30mVp-p * 66µF/(66µF+220µF) = 6mVp-p = ~3mWrms. So that 220µF is probably there to control ripple on the output, and therefore EMI. The elcap is 6 cents in 5-packs at LCSC. The ceramics are 8 cents in 10-packs. How hard do you
really want to push on that, and do you have a scope to experiment and validate?

Any greyerbeards want to tell me how far off my tree I am?