How to map your sources before connecting
The data inventory that avoids reconnecting everything later and makes you start with the right source.
- Make a simple inventory of your sources
- Prioritize what to connect first
- Arrive at the connection knowing what to expect
The temptation is to open Nekt and connect everything you have: HubSpot, Meta Ads, the order bank, the goals spreadsheet. It seems productive, but it usually turns into rework. You connect a source that no one uses it, leaves out the one that unlocked the first report, and In two weeks the setup will be redone. Mapping beforehand is a step of ten minutes that saves days. And the map doesn't start at the source: starts with the business question you want to answer, and works its way back backwards to the source that answers it.
Why map before connecting
Mapping solves three problems at once. First, avoid connect the wrong source: the one that seemed important but doesn't feed any real use case. Second, avoid rework: reconnect, rename and rearrange later because you started without a plan. Third, and most importantly, it helps to prioritize: almost always a handful of sources unlocks most of the value, and the map makes this obvious before you waste energy.
Start with the question, not the source
The most common mistake is to look at the connector list and ask "what can I call?" The way that works is the opposite: start from a business question concrete and do engineering reverse to the source. First the question, then the data answers this question, and only then which source delivers this data. So you never connect something without knowing what it's for.
An example. The question is "which channel brings in the most revenue?". To respond, you you need two things: the sales value and the origin of each lead. Sales live in e-commerce or in the order bank. Origin of the late lead in CRM. That's it: the question took you, alone, to the two sources that matter.
The inventory questions
After the business question pointed out the candidate sources, a good inventory answers five questions per source. What data does it bring, where does this data live, how recent does the data need to be to be useful, who is it? the owner (who do you look for when something breaks or lacks access) and what use case it unlocks. No tools needed: a table in one sheet already solves it. See a completed example.
| Source | What data does it bring? | Necessary current affairs | Owner | Use case |
|---|---|---|---|---|
| HubSpot | Deals, contacts, funnel stages | Diary | Commercial team | Pipeline and conversion report |
| Meta Ads | Investment, clicks, leads per campaign | Diary | Marketing | Cost per lead by channel |
| Order bank | Orders, items, values, status | Schedule | Engineering | Revenue and average ticket |
| Goals worksheet | Monthly target per salesperson | Monthly | Financial | Accomplished versus goal |
Note that the current situation varies a lot: the order bank needs be almost in real time, while the goal sheet changes once per month. In other words, how often does each source need updating is very different. This column alone changes how you configures the synchronization of each source up front.
Prioritization: value versus effort
With inventory in hand, add two readings per source: how much value does it unlock and how much effort you can connect and treat. Play every source in a 2x2 matrix and what to do first appears alone.
Decide where to start
An e-commerce client arrived wanting to connect the eleven sources which I had in the first week. In the inventory, it was clear that the first objective was a revenue and acquisition cost dashboard. Only the order bank and Meta Ads unlocked this, around 80% of the requested value. They connected these two, delivered the panel in few days, and the other nine entered little by little, according to new use cases appeared. Without the map, they would have spent the entire week in sources that no one was going to look at in the beginning.
When in doubt between two sources, ask which one your first report needs to exist. What the use case does not dismissal comes first. The rest can wait at no cost.