Inconsitencies with qPCR results

[TonyBrooks]

I've begun quantifying all libraries with qPCR (Kapa kit) but I'm getting massive inconsistencies with the number of reads I get from samples in a multiplex pool.
I dilute each final library 1:1000000 (three 1:100 serial dilution) mixing at least 10 times by pipette after each dilution. This then puts me somewhere in the middle of the Kapa standard curve. Everything is run in triplicate and the R2 of the standard curve is always great >0.99.
I then normalise, pool and requantify by both Qubit & qPCR and generally get a pooled concentration of 9-11nM (what I might expect). Cluster density for the overall run is fairly good, but after indexing there's massive variation sample-to-sample (5-fold differences for my last set of TruSeq RNA libraries).
I've also Qubit'd and Bioanalysed each library (I use the average size between 200-700bp to normalise my Kapa qPCR). There are no dimers or evidence of over-amplification. The Qubit concetrations are slightly better than qPCR but still not great (R2 between qubit and cluster# is 0.52, but for qPCR it was 0.41)
Does anyone have a fool-proof method for qPCR? I was wondering if it's worth running mulitple dilutions of the same sample on the PCR plate, but I would quickly run out of space on my 96 well plate (I often have to do 24 libraries at a time).
Has anyone tried a homebrew probe assay?

[microgirl123]

Strange, I pretty much use your approach for our gDNA libraries and don't see such huge inconsistencies in number of reads.

Are your samples PCR-enriched after library prep? I usually make PCR-free libraries and have found that the structure of the adapters can make the samples run strangely in the Bioanalyzer, resulting in incorrect sizing which affects the KAPA math.

Are all your pooled libraries approximately the same size?

Pipetting error?

The other thing that I've been doing (I'm running a MiSeq) is denaturing the samples with NaOH and then denaturing them again at 96 degrees. Somewhere there's an Illumina bulletin that explains how different adapters/indices denature less effectively than others.

[nucacidhunter]

We use concentration from TapeStation or bioanalyser and normalise to 6 nM. Then dilute it 10000 times in two steps for QPCR using KAPA kit. Our pooled samples gives very close figures specially when they are the same type of library with similar trace.

[TonyBrooks]

We use concentration from TapeStation or bioanalyser and normalise to 6 nM. Then dilute it 10000 times in two steps for QPCR using KAPA kit. Our pooled samples gives very close figures specially when they are the same type of library with similar trace.

I was thinking about trying a double normalisation too. First using the Bioanalyser and next by qPCR. Might try that for the next run.
We'll also work on automating qPCR set up to avoid any pipettting error.