When tasks and docs share one record, work stops scattering

A typical software or agency team works like this: the card lives in Trello, the decision in Slack, the spec in Notion, the due date in an Excel row, the client brief in an email attachment. Nobody is malicious. Each tool does its job. The problem is that the record is split five ways. “Which one is current?” is asked again at every stand-up, and the answer drifts a little each time.

The cost is invisible at first. For two weeks everyone says “we’re fast” because copy-paste does not hurt yet. In week three a date slips. In week four the client thinks they approved an old PDF. In week six a newcomer walks three tools and still opens the wrong file. The villain is not the process — it is the split record. When the date slips the team argues about speed; the real loss is that the decision has no home.

If you want to measure it, a simple count is enough: how many decisions stayed in Slack last sprint, how many specs were never bound to a task, how many “which doc is current” arguments happened. Most teams do not keep this number because it is embarrassing. Ontra turns that embarrassment into a report. The decision sits on the task comment, the spec on the same work’s page, the date on the card’s single due field. Three sources collapse into one record.

In Ontra a task opens in the list, changes column on the board, sits under a feature on the roadmap, becomes an effort row on a report, may be a pin on the map, and carries a bound page in the docs tree. None of these is a copy. A view is a lens. Assignee, status, due date, and permission are read from the same record on every surface. The board is not “a second Trello”; it is another look at the list card.

In practice this split changes one thing: you do not update one copy “for the board” and another “for the report.” You write the card once, correctly. A wrong status looks wrong on every surface — that is a feature. There is no secret Excel truth. A manager does not refresh a Friday-night slide while the board moves on Monday. The milestone report opens five minutes before the meeting and computes percent from card count.

The document is part of this tree, not an island. Spec, meeting notes, and acceptance criteria bind to the same work. Page access flows from the project; you do not mint a separate share link “so the intern can see the PDF.” When tag or assignee changes, document access follows. The archive is not a graveyard the task is moved into when it finishes; the record stays put and the status closes.

A Slack line that says “let’s do it this way” vanishes in two days. Search works for a while, then the channel grows and the sentence is buried. In Ontra the same sentence sits under the task. The next sprint reads it. A new member three months later does not hunt a PR for “why is this endpoint like this?” — they read the card comment. The decision also binds to the doc; you do not open a second wiki page.

This is not a ban on chat. Fast debate can stay in Slack. The acceptance sentence lands on the card. If the team is not taught the “debate / decision” split, everything still dies in chat. Ontra does not force a required field; it forces a ritual: when the sprint closes, no card is left without a decision. The activity report shows who wrote what and when. “Nobody looked” cannot stand against the log.

Do not inflate setup. On day one pick a real project — not a showcase “sample workspace,” the work that goes to a client this week. Open one list. Write three tasks: in flight, waiting, due this week. Put an assignee and a due date on each. If a spec exists, bind the page to that task. Paste the latest Slack decision as a comment. That is ten minutes. Do not hunt a tutorial; use the panel.

On day two leave the old tool read-only. New cards open in Ontra. Going “just once” back to Trello brings the habit back in two weeks. On day three look at the report. If it is empty, the tool is not missing — the record is not bound yet. On day four sit the docs tree in the project folder. On day five lock the sprint. On the weekend read effort and delay, not empty menus.

The active sprint, live client work, and this month’s docs move. A five-year archive, “maybe later” cards, and dead channels do not. The archive can wait a month. Do not spend the demo or the first month touring history. Every card you move should have an owner and a date; an ownerless card is ownerless in Ontra too — the tool does not magically assign it.

Importing a Notion page as-is can also be a trap. If the page is not bound to a task it is an island again. Move the work first, then the spec. Empty features bloat the roadmap; bind at least one task to a feature or do not open the feature yet. “We’ll tidy later” still produces a messy workspace six months on.

The success metric for month one is not “everyone liked the board.” It is this: stand-up finishes in fifteen minutes, the delay list talks only about red dates, the date sent to the client matches the card, a new member opens the right spec on day three. If those are missing, the tool is not missing — the record is still split. A team that never moves Slack decisions onto the card will use Ontra as a second chat.

The report should be full at the end of this month. Effort, repeats, close speed, and delay read from the live list. An Excel copy leaves only if a client demands their format. The exec slide is generated from the plan, not the reverse. This discipline looks boring; what is actually boring is the Friday tour of “which file is current.”

Say you are shipping a payment screen for client A. In the old setup the card is “payment UI” in Trello, the spec is “checkout v3” in Notion, the decision is “show VAT on the last step” in Slack, the date is Friday in Excel. Three days later the designer opens Notion v2, the developer misses the Slack decision, the manager tells the client the Excel date. Delivery slips. Nobody is at fault; the record is five.

In the Ontra setup there is one task: “Client A — payment screen, VAT on last step.” Due Friday, assignee set, the spec is that card’s page, the Slack sentence is pasted as a comment. The board shows “doing,” the roadmap sits under the feature, the report reads effort from this card. The designer cannot open the wrong version because v2 is not a separate island. A new member on day three opens the same card and sees spec, decision, and date together.

Apply this example to your own work. Pick a delivery that currently lives in three places. Write the card in Ontra, bind the spec, paste the decision, put the date in the single field. Leave the old tools read-only. Two days later you will see the “which is current?” question disappear. If it does not, the record is still split; check the binds, not the tool.

A task carries three texts and scatters because the three homes differ. Decision: “why this way.” Spec: “how it will look / behave.” Acceptance: “what must be true to count as done.” Decision goes on the comment, spec on the page, acceptance at the end of the same page or as bullets. An issue title is not acceptance. A Slack emoji is not a decision. A Figma comment is not a spec — unbound, it is an island.

Without an acceptance bullet, “done” is a lie and the repeat report bloats. The same card closes three times because everyone carries a different “done.” Ontra does not fix this with magic. Ritual does: when the sprint closes, no card is left without a decision and acceptance. Activity shows who changed the spec. Stand-up does not reread the spec; it talks the red date and the missing acceptance.

Docs AI can draft; the home is still the task. If a draft sits in the tree unbound, the wiki island returns. Put the draft on the card’s page, mark acceptance bullets, keep the date. The tool produces text; record discipline is yours. A generated paragraph is not a spec; a spec is the sentence the team accepted.

Trap 1: “Let’s keep Trello open a while.” Two weeks later everyone is back there. Leave it read-only; require new cards in Ontra. Trap 2: importing Notion as-is. A hundred unbound pages are still an island. Bind this month’s specs. Trap 3: pasting every comment onto the card. Debate can stay in Slack; the acceptance sentence lands on the card. Trap 4: leaving dates “flexible.” The report goes blind and the sprint lies.

Trap 5: a separate slide for management. It updates Friday; the board moves Monday. Open the milestone report five minutes before the meeting. Trap 6: “this work is small, no doc needed.” Small work still carries a decision. Two sentences are enough; the home is still the card. Trap 7: setting access later. An intern downloads the spec. Pick the role at invite, grant the folder on day one.

Week 1: one project, one list, the live sprint, three real tasks, spec binds, old tool read-only. Week 2: roles, invites, docs tree, first delay report. Week 3: feature binds, a goal sentence, three saved views. Week 4: move the next sprint, not the archive. At month-end stand-up is fifteen minutes, the client date is on the card, a new member finds the spec. If those are missing, do not extend the timeline — finish the binds.

The success sentence is not “we liked Ontra.” It is “decisions are not homeless.” To measure it, count decisions left in Slack last sprint, unbound specs, and “which is current” arguments. All three should drop. If they do not, it is not missing training — it is missing ritual. Sprint close requires a decision and acceptance. The tool helps; the enforcement is yours.

On sprint day one the card opens. The title is in the customer’s language, not internal jargon. Assignee and Friday’s date are not left empty. The spec page is bound at the same time; “we’ll add the doc later” is forgotten in two days. The latest Slack decision is pasted as a comment. These five moves take ten minutes and carry the rest of the sprint. Skip them and Wednesday’s “which is current” returns.

The developer drags the card to “doing” on the board. List, roadmap, and report change at once. There is no separate Trello update. The designer reads the spec on the same card’s page; they do not fall into Notion v2. The manager does not produce a Friday slide; they open the milestone report. The client lead is a viewer: they see the draft, they do not download the file. The date that goes outside is the date on the card.

Wednesday the spec changes. Page history writes into task activity. “Who changed it?” is the record, not chat. If acceptance bullets are still missing the card cannot be dragged to done; even with `tasks.status` the ritual stops it. Without acceptance the repeat report will bloat next sprint. Thursday the client sees the draft on a passworded link. There is no workspace invite. Friday lock: unfinished work is not deleted, it moves. The decision is written on the card.

Slack debate is not banned during this sprint. What is banned is leaving acceptance in chat. The PR number is written on the card; review may stay on GitHub. A photo or screenshot lands on the card, not a separate gallery. A new member opening this card on Monday sees spec, decision, date, acceptance, and who wrote what together. They do not tour three tools. That is the concrete form of one record. The metric is this: on day three the spec is found, stand-up finishes in fifteen minutes, the client date is on the card.

A Slack line that says “let’s do it this way” vanishes in two days. In Ontra the same sentence sits under the task; the next sprint reads it. The decision also binds to the doc — you do not open a second wiki page.

When tag and assignee change, document access follows. You do not mint a separate share link “so the intern can see the PDF”; the record already belongs to the project.

Move the live Trello card into Ontra, bind the Notion page to the task, paste the Slack decision as a comment. If the report is empty on day three, the record is not bound yet — the tool is not missing.