How much of my CRM can I trust?
Eight quick questions about what is in your CRM. You get back how many contacts you could act on today, what the rest is costing you, and what to fix first.
- How many of your contacts you could act on today.
- What the duplicates and the dead records are costing you a month.
- The one thing to fix first, and why that one.
- Whether a clean up is worth it, or whether the way records get in is the problem.
The problem this settles
Somebody says the CRM has twelve thousand contacts, and everybody in the room treats that as an asset. Then a list gets pulled for a campaign and it turns out a few thousand are the same people twice, a few thousand have not been touched since the year before last, and nobody can say which of them ever asked to hear from us. So the campaign goes out to everybody, or it does not go out at all.
The instinct is to clean it, and a clean up is the one thing that reliably does not work on its own. Records arrive by four routes with nothing checking them against what is already there, so a database tidied in January is untidy again by spring, and the second clean up costs exactly what the first one did. The fault is at the door, not in the data.
This check counts how many contacts you could act on today rather than how many you have, separates the faults a script fixes in an afternoon from the ones needing a decision per record, and names the way in that is causing the most of it. It will tell you not to bother cleaning when the honest answer is to fix the door first, and it will not pretend a deduplication project is a strategy.
- Your CRM
- Your answer
- Your clean up
Question 1 of 8
How to use it
- 1Answer eight questionsQuick picks about what you use, roughly how many contacts, what is wrong with them, how records get in and who owns it being right. One short line about the business. Two minutes.
- 2Run the prompt in your own ChatGPTThe page writes the prompt and opens ChatGPT with it already sent. A free account is fine, and Claude or Gemini work too. You never export a record, a name or an address anywhere, and nothing comes to us.
- 3Bring the answer backCopy the code block at the end of its reply and paste it back on the page. Ctrl+V anywhere works. The page reads it and scores it against the rubric published here.
- 4Read the verdict and take the reportYou get the share you could act on, the faults ranked by what they cost, which are a script and which are a decision, and the way in to fix first. The report is on the page in full. If you want the image file to keep or forward, that is where we ask for an email.
How the score works
This score is ours, not an industry standard. It is the same seven things we check before quoting any CRM work: does anything stop the same person going in twice, how many unchecked ways in are there, does one person own it being right, what share could you act on today, can the faults be fixed in bulk, are the people who opted out marked rather than deleted, and can you tell where a record came from. Four of the seven are about the door rather than the data, because that is what decides whether a clean up lasts. It is published here so you can disagree with it.
- 20Something stops the same person going in twice
- 15Records do not arrive by four different unchecked routes
- 15One person is responsible for it being right
- 15Most of the contacts could be acted on today
- 10What is wrong can mostly be fixed in bulk
- 15People who asked not to be contacted are marked, not deleted
- 10You can tell where a record came from
Questions
How do I know if my CRM data is any good?
Count what you could act on rather than what you have. Pick a hundred contacts at random and ask three things of each: is this the only record for this person, is there an email or number that still works, and could somebody tell what to do next from what is on it. The share that passes all three is the real size of your database, and it is usually a good deal smaller than the number on the dashboard.
Should I clean the data or fix how it gets in?
The door, first and almost always. Records arriving by several routes with nothing matching them against what is already there means a clean up is undone at the rate new records arrive, so you pay twice for the same tidy database. Fix the matching, then clean, and the clean one stays clean.
What is the fastest way to deal with duplicate contacts?
Stop making new ones before merging the old ones, otherwise you are bailing out a boat with the tap running. Then merge on email first, because it is the one field that is genuinely unique, and on company domain for the records with no email. Leave the pairs the tool is not confident about for a person to look at, because a wrong merge loses history and is much harder to undo than a duplicate is to live with.
Do you see my answers or any of my records?
No. The questions, the prompt and the scoring all run in your browser. You never export or upload a record, a name or an address, and you are not asked for access to anything. Nothing is sent anywhere unless you fill in the form at the end, and that form shows you the exact note before it goes.
Do I need a paid ChatGPT account?
No. A free account works, and so does Claude or Gemini. The prompt asks for a code block at the end of the reply, which all three can do.
Should I delete the contacts who asked not to be contacted?
No. Mark them and keep them, because a deleted record is one the next import puts straight back, and then you contact somebody who told you not to. Keeping a suppression list is the only version of this that works, and it is the one mistake in here with a legal edge on it rather than just an embarrassing one.
Who is this not for?
If you have under about two thousand contacts, most of this is an afternoon with a spreadsheet and a filter, and you do not need a check to tell you what is wrong. It is for the database that has become something nobody fully believes.
How long does it take, and what do I get?
Eight questions, one prompt and one paste. You get the share you could act on today, the faults ranked by what they affect, which ones a script can fix, and a drawing of how records should get in so the mess stops coming back.
What if the answer is that it is not worth cleaning?
That is a real outcome and it happens when the way in is the problem. Cleaning then buys a tidy database for a quarter at the same cost as doing it properly, so the page tells you to start at the door instead. None of that first step needs us.
Tell us what you are building.
Thirty minutes, free, with the people who would do the work.
Book a call