Choosing an objective
The objective is the single question the daily split answers. Same budget, same campaigns, three different answers — because each objective measures "worth it" differently.
Pick one of the three cards in the Strategy box on Global settings. Profit is the default. It applies to every day the group runs unless a change point or a day's own settings say otherwise.
The three objectives
| Objective | Maximizes | The budget flows to |
|---|---|---|
| Profit | Contribution profit — revenue after fees, COGS, and shipping | Products that keep the most money per baht spent |
| GMV | Attributed revenue | High-ticket products, even at thin margins |
| Orders | Incremental conversions | Cheap, fast-converting products |
How DataGlass maximizes it
Every night, for each group:
- Forecast each campaign. The model predicts how many orders each budget level would bring. The curve rises, then flattens — each extra baht buys a little less than the one before.
- Score a split. It turns those orders into the objective's number:
Objective Score Profit Orders × contribution per order, minus the ad spend GMV Orders × price Orders Orders - Search for the best split. It tries different ways to divide the budget and keeps the one with the highest group score — inside your budget, mode and constraint. At that split, moving a baht from one campaign to another would not raise the score.
Because the curves flatten, the best split rarely puts everything on one campaign: once a campaign's next baht earns less than another's, the money moves.
What actually changes
Take two campaigns and ฿1,000 a day:
- Campaign A — ฿1,200 item, 8% margin.
- Campaign B — ฿250 item, 35% margin.
Profit pushes the budget to B: each sale keeps ฿87 against A's ฿96, but B converts far more often per baht. GMV pushes it to A: one sale is worth almost five of B's. Orders pushes it to B harder still, and stops caring what B is worth once it converts.
None is wrong. They optimize different businesses.
Picking one
- Profit — the default, and right for most shops. Use it whenever the group's products have COGS on file.
- GMV — use when you're chasing a revenue target, a platform tier, or share in a category, and you accept thinner margins to get there.
- Orders — use for launches, review velocity, or ranking momentum, where the count of sales matters more than what each one earns.
GMV and Orders need a bound
Profit is self-limiting: it counts what the ads cost, so once the next baht of ads earns back less than a baht, profit falls and the solver stops. GMV and Orders are not. They don't count the ad cost at all, so any baht expected to add a little more revenue or one more order scores higher — however much it costs. An unbounded group would keep spending until the extra bought nothing.
DataGlass refuses to save that. A GMV or Orders objective needs at least one of:
- a Group Monthly Budget or Group Daily Budget, or
- a constraint.
Without either you get "A GMV or Orders objective needs a bound" and the save is blocked — on the group, on a day, and on a change point alike.
Pair it with a constraint
An objective on its own has no floor — maximizing orders will accept unprofitable ones. Add a constraint to fence it: maximize orders, but never below 15% margin.