1 in 3
job items placed on the map far from their own job.
The addresses said Birmingham. The map said Birmingham, Alabama.
Check your data before you bet on it.
Data problems are usually found by the worst person: a client, an auditor, or the AI you’ve just switched on. And once they’re found, nobody owns the fix. We check your data before a decision rests on it, put a cost on what’s wrong, and make sure someone owns fixing it.
For companies that rely on their data but don’t have a data team to check it. What it reads: Salesforce in depth (see the Salesforce check), and other CRMs, warehouses and spreadsheets on request.
8 yearstesting data for a living
3 industriesclinical trials, fintech, sport
Read-onlynothing in your system is changed
1 weekfrom login to findings
When it matters
The moments a business is about to trust its numbers more than usual.
Moving systems
Before the data is moved, and after, so nothing is lost or changed on the way.
Buying or selling
Data due diligence: what the records actually show, before a price rests on them.
Rolling out AI
An AI tool is only as good as the data it reads. Find out what it would be reading.
Scaling fast
Small errors in a process get copied with every new client. Find them while they are cheap.
What a check finds
Four findings from one field service company
From Ashgrove Facilities, our sample org: invented, but every finding is a pattern we’ve met in real ones. Every record was complete, so nothing in the system flagged any of them. Each one only showed up when the records were tested against the system’s own rules.
1,962
of the last 2,000 visits due exactly 14 days after they could start.
A default, read everywhere as a deadline.
116%
of the data storage the company pays for, in use.
More than it pays for. The rest has to be bought or cleared.
18,204
cancelled jobs whose reason is no longer on the list.
Loaded years ago. Tidying them up is, presumably, on a list somewhere.
See the whole thing
A full sample census, for an invented org.
See one first
What you get, for an invented company
Ashgrove Facilities doesn’t exist. Its census is drawn by the same tool and layout a client receives, from figures modelled on real orgs.
The censusEight pages of measured fact: what the org costs, what stands in the way of change, and where its rules and records disagree.Open the PDF →
The one-page summaryWritten by a person, for the decision in front of you: what changes the size of the job, what’s fine, and three questions.Open the PDF →
How it works
Five steps, built on established practice (Ten Steps to Quality Data, and DAMA), not invented here.
1. Decide
A 30-minute call: what you’re about to decide, what data it rests on, and what “good enough” means. Sometimes the honest answer is no.
2. Check
Your administrator creates one read-only user, about ten minutes. Within a week: the census and a one-page summary. Then the login is switched off.
3. Cost
A walkthrough, together. The census only states facts; what they cost your business comes from the conversation, because only you know that part.
4. Fix
A short plan in order of importance, each item with an owner. If you want help seeing it through, we can stay on.
5. Watch
The same check after go-live, or each quarter, so you can see what changed, and nothing quietly slides back.
Case studies
What went wrong elsewhere, and what a check would have shown
Public failures, read the way a data check would read them.
Your AI will believe whatever your data says
An AI that learned from one customer’s bad data, and a $110 million hole in the forecast.
Nobody told them it had to be right
Product data that was right about 30% of the time, and shelves that sat empty.
The rows that were never counted
16,000 positive tests missing from the daily figures for eight days.
Have a decision coming up?
Book a short call. Tell us what you’re about to decide and what data it rests on, and we’ll tell you honestly whether it’s worth checking.
Book a callOr email nicholas@dataikenga.com