Working on a solution for automating An Posts website

witnessmenow

Hello,

I'm not fully sure if this belongs here, but I felt like people on this form are most likely to find it useful.

TL;DR:
I'm working on a solution for automating order entry onto An Posts website from a CSV file. It's super early on in the project at the moment but I have successfully input my last 20 orders through it.

I thought it might be useful for other people in a similar boat as me:
https://github.com/witnessmenow/an-post-pupeteer
(This is super early in development and comes without any instructions at the moment, so unless you are familiar with NodeJS it's probably too early to use at the moment)


I've been selling small scale (20-30 packages a week) online for the past 3 years or so. Recently I moved to registered post only due to postage being so bad. Last week my wife spent an hour and half in the post office. We then spent another 30 mins transcribing the Tracking number from the receipt onto our spreadsheet. All stuff I didn't want to be doing every week.

Most postal services offer a bulk upload option where you give it a specifically formatted file and generates labels for them, An Post don't have this.

What this does is automatically interacts with the An Post website to fill in the appropriate details, video of an early test I was doing here:
https://twitter.com/witnessmenow/status/1365948297026879489
(An Post Vox is a parody account, but I wasn't bothered deleting the tweet )

It just adds orders to your basket so they can still be reviewed and verified before committing to pay for them.

It still needs a good bit of work, but I'm interested to hear if there would be any demand for this? Not looking to charge for it or anything, it's open source using the MIT license (so you can use it commercially).

A small subsection of the future work needed:

• Currently only supports 'Large Letter' that are Registered. It shouldn't be a problem supporting more, but additional work would be needed.
• The an post website changes under the hood based on the country selected, and I'm not exactly sure what the criteria is ahead of time for this. I've come across 3 variants, EU seems to be one, US, Oz and Canada have another and then Switzerland and Russia have another. These 3 variants work, but each additional non EU country will need to get added to the appropriate list (these are just the countries contained in my 20 orders)
• It is not battle hardened at all! It has worked for my 20 orders but who know what future orders will throw up
• It currently only supports two different types of items, this could be extended relatively easily.

beauf

Its a good idea especially if you streamline the interface that people use to create orders.
The hard part is reaching your target audience, then convincing them if its value. People are slow to adapt to change.

witnessmenow

beauf wrote: »

Its a good idea especially if you streamline the interface that people use to create orders.
The hard part is reaching your target audience, then convincing them if its value. People are slow to adapt to change.

I guess my target audience is people running a part time e-commerce business similar to myself, if you were full time you'd surely want to be at the level where an post start engaging with you! For this reason I'm not interested in selling it, hopefully it might help a couple of little guys out and let them focus on more pressing things than data entry!

I'm not here to convince anyone to change, if people are experiencing the problem this solves they will know all too well about it! I'm not sure exactly how long it would take me to add 20 orders manually, but this does it in about 2 minutes with 0 interaction from me once started

Streamlining the generating of the csv is the key alright, but I won't be doing this for anything I don't use myself.

I could increase adoption by doing the most popular ones, but I'm not overly concerned with that. The time consuming part of this whole operation is the automation side.

But hopefully if other people see the value in it and they are using something like ebay or whatever they might push up the code they write for the csv.

The platform I use is super niche (tindie, think etsy but for hobby electronics like raspberry pis). I will push the code I used for that soon. I basically run a script and it automatically generates the csv in the correct format for me with all my unshipped orders. The only part of the process that is not great is I don't know for sure the weight of the package ahead of time, so I probably need to pack them first to see if it's over 100g or not. (Some I could say for sure by what is ordered, others would be close)

beauf

I'm just saying persuading people of doing things a faster and easier way is often surprisingly difficult.

witnessmenow

As someone who is moving from just printing labels with the addresses on them and letting the post office staff enter the details, unfortunately I've hit a new problem.

As mentioned I successfully entered my 20 orders and I purchased them last night. That was all fine, but that's when the fun begins.

(Will provide more info below)

• The labels they provide are not in any way compatible with address label printers
• I now have 20 orders with tracking numbers, and there is no good way of matching customers orders to tracking numbers.

Labels first, they are not designed to be printed with any type of label printer, they are intended to be printed on A4 paper (I guess) and cut out and sellotaped to your package. They come in a single PDF and each page of the PDF contains different combinations of the different labels. This might be fine to do for one or two orders, but not realistic for any sort of quantity.

Some images of how they come back here:
https://twitter.com/witnessmenow/status/1367230836744089604

You do have the option of bringing the email in and getting the labels printed by An Post in the post office, but there is a problem with that too. Each package has a barcode and the only piece of identifying information for each barcode is the destination country and the value. The only other way to verify your matching the correct barcodes to the package is the order your entered them in, which is madness! Really a customer reference number is needed at order time. [Pre-submit edit, this wasnt as bad of a previously thought as they just scanned the barcode and could match the label that it prints to my placeholder label)

This same thing is the problem with matching my customer orders to tracking numbers, I either have to use the email and match tracking numbers to orders based on the order I entered them, or I can manually go through the PDF address labels and match them that way, either way is pretty error prone.

I'll have to see if I can automate the extraction of the tracking number, cause its super slow at the moment!