Territory planning season is here, and the teams we work with are approaching 2027 differently than they did a year ago. Fewer are redrawing the whole map from scratch. More are starting with how their reps actually spend time, testing a few designs side by side and fixing their data before the first scenario runs.
We pulled together what we've learned from recent territory planning sessions with RevOps teams at B2B software companies. Here are 12 practices that keep coming up.
(Editor's note: In most of these cases, we're using our Bookbuilder reporting for most of these metrics, and Carve to do the actually territory design.)
Start with the model, not the map
1. Pick a planning model per region or business unit. There's no rule that says one model has to fit the whole company. One security software company plans the Americas segment-first (strategic, enterprise, commercial) and plans international geo-first, then by segment. Their old software couldn't handle both at once, which is why they turned to Gradient Works. Your tools should fit your go-to-market model, not the other way around.
2. Start from rep capacity, not the account list. Before deciding how many accounts each rep gets, find out what share of their current book they actually work and where their time goes. One team with 1,500+ accounts per outbound rep is now piloting focused books of about 200 against a control group. Capacity first, carve second.
3. Separate hunters from farmers. Net-new prospecting and account growth are different jobs and need different designs. One software company is moving hunters to dynamic books pulled from a large shared pool, with no geographic territories at all. Account growth reps get named books instead, and accounts move from hunter to farmer on a quarterly review once they cross a spend threshold.
4. But build a buffer into named books. Even named books for account growth reps need slack. As one ops leader put it: if a rep is expected to deliver $1M from 10 accounts and one doesn't come through, that rep is stranded. A few extra accounts per book protects both the rep and the number.
Define "balanced" before you balance
5. Write balancing targets with a tolerance and a priority order. "Balanced territories" isn't a spec. "$2M of potential per rep, within plus or minus 5%" is. When you have more than one target (revenue potential, account count, customer count), decide which one wins when they conflict. Vague targets produce vague scenarios. Carve can easily handle all these requirements and priorities.
6. Balance at more than one level if your org needs it. Larger teams often balance across managers first, then across reps within each team. One data company we work with plans a team of roughly 60 reps under 9 managers this way: managers’ books are balanced against each other, then rebalanced rep by rep.
7. Plan around real constraints instead of ignoring them. The cleanest model on paper breaks on contact with reality. Constraints teams are explicitly planning around:
- Territories that follow regulatory groupings rather than contiguous geography
- Weighting capacity by rep level (for example, a senior rep counts as 110% of a baseline rep)
- Never moving accounts with an open opportunity
- Not reassigning an account whose rep changed recently, to avoid customer whiplash
Get your data and scoring right first
8. Fix data before the planning cycle starts. Teams that struggled with planning consistently said the bottleneck was data, not tooling. The usual suspects: duplicate accounts, broken parent-child hierarchies, industry codes that don't match how they segment, and missing opportunity or activity history. At one company, managers were spending about a full day per book on spreadsheets and outside research to fill those gaps.
9. Gate on a hard ICP rule before you rank by score. Similarity and lookalike scores are powerful, but they rank on resemblance, not fit. One team found IT outsourcing firms scoring highly even though they only sell to software companies. The fix was a simple yes/no "sells software" check applied before any scoring. Validate rules like this on a labeled sample of 20–30 accounts before running them across the full universe.
10. Keep scores fresh in two different ways. Score net-new accounts continuously as they enter your CRM. Do a full re-score only when your reference set changes, such as new wins, churn or a shift in ICP. And make that reference set big enough: one team grew its reference customers from about 26 to about 108 by loosening tenure filters and combining adjacent segments, and its Market Map account clusters got much cleaner.
Test, collaborate and publish
11. Run scenarios side by side, and have managers react to their own books. Compare a vertical-based design against a geographic one before you commit. Duplicate a scenario before changing it, and keep a recorded reason for every assignment so the plan holds up when someone asks "why did this account move?" Then give each manager their own book to review, comment on and override. Feedback on real accounts is faster and more concrete than feedback on a slide. One team is using this approach to replace a two-week back-and-forth on new-hire books. (You can collect all this feedback directly in Carve now, by the way.)
12. Pilot outside live books, then publish to the system you already run. Don't disrupt reps' books in the closing quarter. Pilot new structures on a separate pool of accounts first. When the plan is final, publish it into the system your team already uses.
A quick checklist for your 2027 plan
- What share of their current book do reps actually work?
- Are hunters and farmers on different models, with a clear handoff rule?
- Have you written down your balancing targets, tolerances and which one wins?
- Is your account data deduplicated, with clean hierarchies and the right industry codes?
- Does your scoring screen out non-ICP accounts before ranking?
- Have you compared at least two territory scenarios before picking one?
- Do managers get to react to their own books before rollout?
- Is there a pilot plan that leaves Q4 books alone?
If you're working through 2027 planning and want a second set of eyes, we're happy to help. This is exactly what we built Carve for: modeling scenarios, balancing books and getting manager feedback in one place before you publish to your CRM. We'd love to talk to you! You can try Carve for free, no strings attached.
%20(3).png)
