Rank a list without a campaign

You’ve imported a few thousand accounts and want to know which ones to work first. You don’t need a campaign for that: rank the list against a product’s fit criteria, look at the best ones, then add them to a campaign or plan next steps.

Before you start

The product needs fit criteria for the purpose you’re ranking for (sales, press, partnership…). If it has none, generate them first: in Settings → Products, click Generate fit criteria in the product’s Fit criteria section, pick the purpose, add any hints and click Generate. Review, edit and approve what the AI wrote. (If you open Rank for a product and purpose without fit criteria, the dialog links you there.) See Fit criteria and fit score.

1. Pick the accounts

On Accounts, filter the list down to what you want ranked: a country, a business type, an import, a saved view, the accounts not in any campaign. Ranking uses exactly the list you see, with the same filters.

To rank only some of them, tick their rows instead: a bar appears above the table with the number selected and its own Rank button, which ranks just those accounts. When every row on the page is ticked and more accounts match, Select all N matching in that bar extends the selection to the whole filtered list.

Pick the product under Fit for first if you like: the Rank dialog starts from it, and the list gains a Fit column. Coming from an import, Rank these N accounts on its result step brings you here with the import’s accounts and the dialog open.

2. Check the estimate

Click Rank at the top of Accounts (on a narrow window it is in the More menu), or Rank in the selection bar when you ticked rows, then pick the Product and the Purpose. The dialog says what it will rank: the accounts the list shows, or the selected accounts. Before anything is spent you see:

• Accounts matching, Accounts to score and Already scored: accounts that already have an up-to-date fit score for that product and purpose are skipped (tick Rescore accounts already scored with the current criteria to score them again);
• the Batches of AI calls that takes, the Model and the Estimated cost, based on what scoring has cost in this project lately, or on the size of the prompts when there’s no history yet If the AI model has no price set, the dialog says so and the run can’t start: without a price the budget cap can’t be enforced. The estimate also shows what’s set aside for Website lookups (finding websites for accounts that have none), which counts toward the cap.

A run needs a price for its model: without one LeadHunter can’t count what the run spends, so it can’t hold it to a budget, and the run doesn’t start until a price is configured. The run keeps the model it started with to the end, even if the default model changes meanwhile.

A run takes at most 20,000 accounts. For a bigger list, narrow it or rank a sample.

3. Try a sample first

Not sure the fit criteria are right? Tick Try a sample first (it is on by default for lists of more than 200 accounts): the run scores the first 50 accounts (or any number you choose) that need a score. When it finishes, look at the scores and reasons. If they’re off, edit the fit criteria and try again; if they look right, rank the rest of the list.

4. Set a budget cap and start

Every run has a Budget cap (USD), prefilled at one and a half times the estimate plus the website-lookup reserve (at least $1). Click Start to begin; a banner links to the run (Follow it in Runs) and to View results.

The cap covers everything the run pays for: finding websites for accounts that have none, and the scoring calls. Before website lookups start, and before each batch of scoring calls, the run sets aside what the step may cost (one and a half times its estimate) and checks it, with what it has already spent, against the cap. When the step would go over, the run stops there and keeps what it already scored.

The cap is where the run stops, not an exact ceiling. The run checks its spend before every AI call (website lookups and scoring batches) and stops when the cap is reached. Calls already in flight finish, so the final spend can go a little over the cap, typically by less than one batch. Several chunks run in parallel, so the overshoot can be up to one call per running chunk. What the run already scored is kept. If the cost of a call can’t be recorded (for example, the model’s price was removed during the run), the run stops rather than spend money it can’t count.

If the list changed between the estimate and the start (an import landed, a colleague scored some accounts), the dialog shows the new estimate and nothing starts until you click Start again.

Only one ranking run per product and purpose runs at a time: if one is already running, the dialog says so and links to Runs.

5. Follow the run

The run appears under Runs → Ranking with its progress (accounts done of the total), what it has spent against the cap, and how it ended. Cancel run stops a running one.

Result	Meaning
Starting / Running	It is preparing its accounts, or scoring them.
Completed	Every account was scored, had too little data to score, or was deleted or merged into another account before the run reached it.
Partial	It stopped early (Stopped at the budget cap, or its spend could not be recorded) or some accounts could not be scored; what it scored is kept.
Cancelled	You stopped it. It stops within one AI call; what it scored is kept.
Failed	It could not run (for example, the fit criteria were deleted).

Runs survive deploys and restarts: work in progress picks up where it stopped, and accounts already scored are not scored (or billed) twice.

Merging or deleting accounts while a run is going is safe. An account merged into another before the run reached it counts as merged (the account it was merged into is scored only if it is in the run too); a deleted one counts as deleted; the totals still add up, and accounts already scored stay in the run’s results.

6. Act on the best accounts

When the run is done, View results opens Accounts filtered to the accounts the run took on (a From ranking run chip; its × shows every account again), sorted best fit first. From there:

• narrow it to the top band with Min fit (8+, for example);
• Add matching accounts to campaign (or tick the rows you want and click Add to campaign) to work them in a campaign, where you shortlist them (see Review thousands of accounts);
• open an account and add a next step for the ones you’ll handle one by one.

The scores also show in the Fit column of Accounts, on each account page, in every campaign of that product and purpose without its own scoring brief, and in Today’s Best accounts to work next.