A CRO I worked with last year opened our first call by sharing a spreadsheet. Six AEs, six tabs, one tab per rep. Column A was a list of US states. Column B was a list of countries in EMEA. That was the territory plan.
I asked one question: how many accounts that match your ICP sit inside each of those boxes?
Nobody knew. So we counted. Two reps had 41 and 47 qualified accounts. One rep had 610. Another had 22. The rep with 22 accounts had missed quota four quarters in a row and was on a performance plan. He was not a bad seller. He had been handed a territory that could not physically contain his number.
That is what most B2B territory planning actually is. Somebody drew lines on a map in year one, the company grew, nobody redrew them, and now the plan is an artifact rather than a decision. The Sales Management Association found in 2024 that 58% of B2B companies rate their own territory design as ineffective. I believe that number is generous, because it relies on companies noticing.
What a territory actually is now
Stop thinking about geography first. For most B2B companies under 200 people selling software or services remotely, geography is a timezone constraint and a language constraint. Nothing more. Nobody is driving anywhere. The old logic of territory design was built to minimise windshield time for field reps, and that logic quietly stopped applying somewhere around 2015.
A territory is a defined set of accounts one seller is accountable for, sized so that working all of them properly is possible in the hours that seller actually has.
Read that again, because the second half is where every plan I have audited falls apart. Companies get very rigorous about defining the set and then completely hand-wave the capacity math. They assign 600 accounts to a rep who can run maybe 8 real discovery conversations a week and then act surprised when coverage is shallow.
A territory nobody can physically work is not a territory. It is a wish list with a name on it.
The test is simple. Take the account count, divide by the number of touches per account per quarter your motion requires, and see whether the answer fits inside a working week. If it does not, you have not designed a territory. You have hidden a capacity problem inside a spreadsheet.
The four inputs that decide a territory
The RevOps Co-op published a piece arguing that SaaS territory design has gone backwards, and I agree with the core of it. Modern teams obsess over dividing total opportunity into equal-looking slices and forget to ask whether each slice can be worked efficiently. Equal is not the same as workable.
Four inputs decide the shape of a territory. Get these and the map falls out of the math.
Most plans use input one and skip the other three. That is the whole failure mode in a sentence.
Getting account count right
This is where your ICP definition earns its keep. If your ICP is a paragraph of adjectives, you cannot count anything. If it is a set of filters (employee count band, industry codes, tech installed, funding stage, region) you can build a list and get a real number.
I run this in Clay for most clients. Pull the universe from a provider, apply the ICP filters, dedupe against the CRM, and you get a count. It takes a day. In one case the client believed their addressable market was around 18,000 accounts. After filtering to accounts that had actually ever converted from a similar profile, the number was 2,900. Their territory plan had been built on the 18,000 figure, which meant every rep's list contained mostly accounts that would never buy. See TAM SAM SOM market sizing for how to do this counting properly.
Getting capacity right
Ask a rep how many hours a week they sell. They will say 30. Then track it. It is closer to 16 to 20 once you subtract pipeline reviews, forecast calls, CRM admin, internal Slack, proposal building and the two hours a day nobody accounts for.
Then divide. If your motion requires 6 meaningful touches to get an account to a first meeting, and a rep has 18 selling hours, and each touch cycle costs 20 minutes of real work including research, you get roughly 54 touches a week. That is 9 accounts moving per week at full effort. Over a quarter, with everything else they are also doing, a rep can genuinely work somewhere between 80 and 150 named accounts. Not 600.
That Harvard Business Review figure gets quoted in every vendor blog on this topic, usually without the important caveat: 2% to 7% is the gain from fixing the design, not from buying software. It is free revenue sitting inside your existing market. It is also not huge. Anyone promising you 30% from a territory redesign is selling something.
How to actually build the plan
Here is the sequence I use. It takes about three weeks for a team of 5 to 15 reps.
Score on two axes, not one
Almost everyone scores accounts on potential value and stops. That is how you end up with a territory full of huge logos that will each take 14 months and never close.
Score every account twice. Potential value is your estimate of ACV if they buy, which for most companies is a simple function of employee count or seat count. Effort cost is how expensive it is to work them, driven by sales cycle length, buying committee size, whether an incumbent is in place, and whether you have any relationship or case study in that vertical.
Then balance on effort cost first. A patch with 90 high-effort enterprise accounts and a patch with 90 low-effort mid-market accounts are not equal, even if the potential value totals match. The mid-market rep will hit number and the enterprise rep will not, and you will conclude the enterprise rep is weak.
Then decide your split dimension
Only after the counting do you choose how to cut. The options, roughly in order of how often they are the right answer for a company under 200 people:
By segment or size. Cleanest for most SaaS companies. SMB, mid-market, enterprise, split on employee count or revenue band. Effort profiles differ sharply between bands, which makes balancing easier, and it lets you run different motions per band.
By vertical. Strong when your product genuinely lands differently per industry and reps build real domain knowledge. Weak when you have six verticals and four reps, because someone ends up owning three unrelated industries and knows none of them.
By named accounts. Best for teams doing serious account-based work. Each rep gets a fixed list of 50 to 150 accounts and owns them regardless of where inbound comes from. Highest quality coverage, most work to maintain.
By geography. Reasonable when timezone or language is a real constraint (a DACH rep who speaks German, an APAC rep awake at the right hours). Unreasonable when it means "you get the states west of the Mississippi" for a company selling on Zoom.
Round robin. Not a territory. It is the absence of one. Fine for pure inbound SMB volume, actively harmful anywhere a relationship matters.
Where territory plans die: the CRM
This is the part vendors skip and it is the part that decides whether any of the work above survives contact with reality.
A territory plan that exists only as a slide is fiction. It has to become data. In HubSpot that means a Territory property on the company object, populated by workflow rather than by hand, with contact and deal ownership inheriting from the parent company. In Salesforce it means either Enterprise Territory Management (heavy, but real) or a custom Territory field with assignment rules on top.
Three things break most often.
Ownership does not cascade. A rep owns the company record but a new contact from that company comes in through a demo form and gets assigned to someone else by round robin. Now two reps are emailing the same buying committee. Fix this with lead-to-account matching that fires before routing, so any new contact checks for a parent company first and inherits that owner.
The territory field is not the source of truth. People update the Owner field directly during a rebalance and leave Territory stale. Six months later nobody can tell you which accounts belong to which patch, because the two fields disagree on 30% of records. Make Territory the field humans set and Owner the field automation derives. Never the other way round.
New records arrive unclassified. An account gets created with a blank industry or employee count, so no territory rule matches, so it sits unowned. Build a catch-all rule that routes unmatched records to a queue a human checks weekly, and fix the underlying data decay problem that caused the blank field.
We build this layer as part of most CRM and RevOps engagements, usually with HubSpot workflows for the simple rules and n8n for the enrichment and matching logic that HubSpot cannot do natively. The routing mechanics themselves are covered in more depth in our guide to lead routing rules.
Ratio between the largest and smallest ICP-matched account count across six supposedly equal territories at one Series B client. Nobody had counted before we did.
The rebalancing question
Every founder asks the same thing: how often do we redraw?
Once a year, aligned with your fiscal planning, at the same time you set quota and comp. Territory, quota and compensation are one decision made three times, and splitting them across three months guarantees they contradict each other. Do the territory count first, then set quota from that capacity, then build comp on top.
Mid-year changes should be exceptions with a written reason: a rep leaves, you open a new segment, a merger dumps 400 accounts in your lap. Every unplanned reshuffle costs you weeks of momentum, because reps stop prospecting while they wait to find out what they own.
The thing to monitor between redraws is not fairness complaints. It is coverage. Build one report: accounts in territory, accounts touched in the last 90 days, accounts with open pipeline. If a rep is touching 20% of their patch, the territory is too big regardless of what the plan says. If they are touching 95% and pipeline is thin, the territory is too small or the ICP is wrong. Pair that with your pipeline coverage ratio by patch and you will see problems a quarter before they show up in the forecast.
One more thing that gets ignored: grandfathering. When you redraw, existing open deals stay with the original rep through close. Ripping an in-flight deal away from the person who sourced it to satisfy a new map is the fastest way to lose both the deal and the rep. Write the grandfathering rule into the plan before you announce it, not after the first rep escalates.
What I would tell a 50-person company
Do not buy territory management software. Not yet. At your size the entire plan fits in a spreadsheet and a set of CRM rules, and the tools cost more than the problem.
Do the counting exercise. Genuinely count the ICP-matched accounts per patch. In every audit I have run, this single step surfaced an imbalance nobody knew about, and it takes a day.
Balance on effort, not just on logos or revenue potential. Load the result into the CRM as a real field with real rules. Report coverage monthly. Redraw annually alongside quota and comp.
That is the whole thing. It is unglamorous and it works, and it will do more for your number next year than another sales methodology rollout.
Not sure your territories are balanced?
We will count your ICP-matched accounts per rep, show you the imbalance, and map the CRM rules that would fix it. Free 30-minute session.
Book an audit →FAQ
How many accounts should one AE own?
For most B2B companies with a mid-market motion, 80 to 150 named accounts per AE is workable. Enterprise reps running long cycles with large buying committees should sit at 25 to 60. SMB reps working a high-velocity inbound motion can handle 300 or more, because the effort per account is a fraction of what an enterprise deal costs. Do the capacity math for your own motion rather than copying a benchmark, because the number depends entirely on how many touches your sales cycle requires.
Should territories be based on geography or industry?
For remote B2B sales, geography only matters where timezone or language is a genuine constraint. Segment by company size is the cleanest default for most SaaS companies, because effort profiles differ sharply between SMB and enterprise, which makes patches easier to balance. Vertical splits work well when your product lands differently per industry and you have enough reps to give each one a real specialism. If you have four reps and six verticals, do not split by vertical.
How do I handle territory conflicts between reps?
Prevent them in the CRM rather than resolving them in Slack. Every account record gets one Territory value and one Owner derived from it. Any new contact runs through lead-to-account matching before routing, so it inherits the parent company's owner instead of getting round-robined to whoever is next in the queue. For the genuine edge cases, publish a written rule ahead of time covering multi-entity companies, subsidiaries and accounts that change size band mid-year.
What is the right time to redraw sales territories?
Once a year, at fiscal planning, in the same cycle as quota and compensation. Redrawing territories separately from quota setting almost guarantees the two contradict each other. Mid-year changes should be reserved for real events: a departure, a new segment opening, an acquisition. Every unplanned reshuffle stalls prospecting for weeks while reps wait to learn what they own.
Do we need territory management software?
Below roughly 25 reps, no. A spreadsheet for the design plus CRM fields and assignment rules for the enforcement will do the job, and the enforcement layer is the part that matters. Dedicated territory tools become worth the cost when you have many reps, overlays, multiple hierarchies, or complex crediting rules that your CRM cannot express. Buying the tool before you have done the counting exercise just automates a bad plan faster.