Frequently asked questions
AutoModel: data, privacy, and what the list contains
Short answers to the questions finance, ops, and sales leads ask before a first run.
What file do I upload to AutoModel?
Upload the dealer order file your team already uses — CSV, Excel, a billing export, or an ERP extract. No special template and no implementation project are required. If your export carries the order history and your account names, it is enough to produce a first ranked list.
Which fields does AutoModel check before analysis?
AutoModel confirms dealer code and name, order date, count, quantity or value, product SKU or category, and state, region, and account owner before analysis. Partial data is workable — order date, count, quantity, or value are each acceptable. If a column is missing or inconsistent, it is surfaced before the run rather than silently guessed.
What signals does AutoModel read?
AutoModel checks more than 150 signals across the dealer relationship, covering buying rhythm, revenue movement, product behaviour, seasonality, field engagement, service friction, payment stress, returns, and delivery. Weak movements in several of these are read together, so an account that looks flat on any single metric can still surface when the overall pattern has changed.
Can AutoModel read data beyond my order file?
Yes. The order file is enough for a first run, covering buying rhythm, revenue movement, product behaviour, and seasonality. When you also upload or connect field visits, loyalty, complaints, payment, returns, or delivery data, AutoModel reads those signals too, so the ranking reflects the whole dealer relationship rather than orders alone.
What if I only have distributor billing data?
AutoModel will not pretend to see dealer-level movement it cannot see. If your data is distributor-only, it flags distributor territories and product lines that are losing movement, so the field can audit them directly rather than acting on an invented dealer ranking.
How long does the first run take?
A first run takes about five minutes from upload to a ranked list. The work is reading an existing export, not building pipelines, so there is nothing to install and no data engineering to schedule before you can judge the output.
What exactly do I get back?
You get an explained priority list, not a raw score. Each dealer row shows who to contact, why the account was flagged, and the next field check for the account manager. Ordering is by possible loss, so the list reads as a set of calls rather than a table to interpret.
Can I download and edit the list?
Yes. The list downloads as an editable CSV, and it can also be sent to the field through WhatsApp or SMS. It can be assigned to owners, annotated in the field, and combined with your own notes, so the output travels with your existing account-management process rather than replacing it.
Is my order data kept after the run?
Uploads, runs, and reports auto-delete after the session. DotAI is built for a first run on a normal export rather than for storing your commercial data, so there is no long-lived copy of your order history sitting behind the list.
Who should upload the file?
Use your approved company email address to upload the order data your team already has. In practice this is the ERP or finance owner who already produces the export, which is why the first run usually needs no new access or procurement step.
Is AutoModel another dashboard?
No. AutoModel produces a ranked, explained action list for account managers, not a dashboard of charts to interpret. The output answers who to contact first and why, which is the question a monthly review usually leaves open until the numbers are already old.
Do I need an IT project or ERP integration to start?
No. The first run starts from an export you already produce, so there is no ERP replacement, no connector build, and no months-long implementation. Integration can remove the upload step later, but it is not required to see whether the list is useful.
Does AutoModel prove a competitor took my business?
No. AutoModel identifies a loyalty-risk pattern from order behaviour; your rep confirms the cause on the call. It does not prove a switch to a named competitor. That distinction matters, because the reason on each row is a prompt for a conversation rather than a conclusion.
How is this different from a single Excel rule?
Excel catches rules such as no order in 90 days. AutoModel ranks combined behaviour changes across more than 150 signals covering timing, quantity, SKU mix, local peers, field engagement, payment, and delivery, so accounts that look flat on any single metric can still surface when the pattern as a whole has changed.
Has AutoModel actually saved an account?
Yes. In one case AutoModel flagged a distributor who had not ordered in 20 days against a usual rhythm of every 10 days; the account team found delivery problems in the last three orders, fixed them, and retained the customer. In another, a distributor whose volume had fallen over three cycles was found to be considering a competitor, and a better offer saved the account.
Still deciding?
If your question is about whether the signal is useful on your data, one free run answers it faster than another page.
Try AutoModel free