Short answer: Treat "lost to a competitor" as a claim that needs evidence on the account, such as a competitor signal, and not only a rep's closed-lost field. Label a loss "no decision" only after you have checked fit and found no sign of a competitor. Leave unclear losses as unknown.
What changed
Outlevo's October 19, 2026 release added two things that bear on this problem [[1]](#sources) [[2]](#sources). You can import past deals from CSV or your CRM and get a win/loss analysis that proposes ICP hypotheses [[1]](#sources). And competitor tracking arrived, with quoted facts from competitors' public pages, changes, win/loss by competitor and competitor signals on accounts [[2]](#sources).
The import covers the common data sources:
- Outlevo reads HubSpot deals with their pipeline, stage, amount and closed-lost reason [[3]](#sources).
- It reads Salesforce accounts and opportunities with their stages for ICP and win/loss analysis [[4]](#sources).
- Without a connected CRM, you can import won and lost deals from a CSV of up to 200,000 rows [[5]](#sources).
- A CSV import maps deal fields such as stage, amount, outcome and loss reason [[5]](#sources).
Start with the closed-lost reason field, because it is where your current labels live.
Why one label hides different problems
In HubSpot, a deal carries a closed-lost reason [[3]](#sources). Check whether your reps fill that field in quickly, with the answer that is easiest to defend: "price" or "went with a competitor." Those two answers may cover deals that have little in common.
A deal that went to a competitor says something about your positioning against that vendor. A deal that ended in no decision says something about urgency, the champion, or whether the buyer had a problem worth solving. A bad-fit deal says something about who you chased in the first place. These point to different fixes. If they share one label, you test the wrong thing.
The three labels and what supports each
Lost to a competitor. The claim is that the buyer chose another vendor. The strongest support is evidence on the account that a named competitor was involved. Outlevo's competitor tracking includes competitor signals on accounts and win/loss by competitor [[2]](#sources). Treat a competitor signal on the account as stronger support than a rep's closed-lost field alone. If you have no such evidence, treat the label as a guess.
Lost to no decision. The claim is that the buyer chose nobody. This label is the easiest to misuse, because it is where deals go when no one can explain what happened. Support for it is mostly the absence of evidence for the others. There is no competitor signal on the account, and the account fits your ICP. The deal simply stopped moving. Be honest that this is an inference, and write it down as one.
Bad fit. The claim is that the account should not have been in the pipeline. Support comes from the account itself: its size, industry, and the traits your won customers share. Outlevo writes fit and intent scores to HubSpot company fields [[3]](#sources). A low fit score on a lost account points to bad fit before it points to a competitor.
A workable evidence bar
Pick a rule and apply it to every loss. A Readiness Card in Outlevo needs at least three observations, each with a source and a date [[6]](#sources). That is a workable bar for any loss label. Before a deal gets "lost to competitor," you should be able to point to a few dated, sourced observations that a competitor was involved.
Two habits help:
- Keep facts and guesses apart. Outlevo keeps facts read from a website only when their quote is found on the page, and labels guesses as hypotheses, never findings [[6]](#sources). Do the same in your loss notes: a quote from the buyer is a fact, and "probably price" is a hypothesis.
- Allow an "unknown" label. A loss you cannot classify is better left unclassified than forced into "price."
What to do
1. Export the loss reasons as they are. Pull closed-lost deals with their reason field before you clean anything. Count how many sit under each label. The shape of that count tells you how much of your picture rests on one overused answer. For file preparation, see how to prepare past deal data for a win/loss import.
2. Add a second column for evidence. Next to each reason, record what supports it: a competitor signal, a buyer quote, a fit score, or nothing. Deals with nothing stay unclassified for now.
3. Re-label with a rule. Competitor losses need competitor evidence on the account. Bad-fit losses need a mismatch with your ICP. No-decision losses are what remains once the other two are ruled out and the account fits. Write the rule down so the next quarter's labels match this one's.
4. Count companies, not deals. Outlevo's win/loss analysis counts companies, not deals, so repeated deals at one account do not inflate a pattern [[7]](#sources). If one account reopened three times, it is still one account. Apply the same discipline in a spreadsheet.
5. Check the sample size. Outlevo labels fewer than 10 won companies as very low confidence [[7]](#sources). If your data is thin, read how to run a win/loss analysis on few deals before drawing a conclusion from any one label.
6. Look at competitor win/loss only where the evidence exists. Use win/loss by competitor on accounts that carry a competitor signal, and treat the rest as open [[2]](#sources). For a wider view of the category, see competitor tracking software for B2B.
Why the split changes what you test
Each label points to a different kind of change.
- Competitor losses suggest messaging and positioning tests against a specific vendor. If one competitor shows up repeatedly, that is where comparison content and objection handling belong.
- No-decision losses suggest tests on qualification and urgency. Messaging aimed at a competitor will not help a buyer who never felt a reason to act. Look at who the champion was and what trigger, if any, started the deal.
- Bad-fit losses suggest ICP changes: who you target, not what you say. If a segment loses often and its fit scores were low, the fix is to stop pursuing it or to test it deliberately.
Mix the three and every result looks like a messaging problem.
Treat any ICP change that comes out of this as a hypothesis. Outlevo labels AI-proposed ICP segments experimental until a person approves them [[7]](#sources). A segment you propose from loss data deserves the same label until a test confirms it.
A note on what the data cannot tell you
Deal data records what happened in the CRM, not why the buyer decided. A competitor signal shows a competitor was present, not that it won on any particular point. A missing signal does not prove no competitor was involved. Read the split as a way to find where to look next, then confirm with buyer conversations where you can.
If you want a starting read on your own ICP before you dig into past deals, the free ICP assessment reads up to 5 public pages of your website [[8]](#sources).