How to Prioritise Accounts: A Simple Scoring Model
A simple, explainable account scoring model for founders: five weighted factors, a 0–100 score and clear thresholds, and how to run it by hand in a spreadsheet.
Why score accounts at all?
When you are watching dozens of companies and several of them show signals in the same week, you need a consistent way to decide whom to contact first. Without one, founders tend to chase the most exciting news or the most famous name, which is not always the best opportunity.
An account score turns your judgment into a repeatable rule. It does not need to be sophisticated. It needs to be consistent, explainable, and updated when things change.
What makes a good account scoring model?
- Few factors. Five factors you understand beat twenty you cannot explain.
- Explainable. For any score you should be able to say exactly which factors produced it.
- Deterministic. The same inputs always give the same score, so you can trust changes in the number.
- Fit and timing together. ICP fit says whether a company is right for you; buying signals say whether now is the moment.
- Clear thresholds. A score is only useful if it tells you what to do.
What are the five factors?
| Factor | Weight | Question it answers |
|---|---|---|
| ICP fit | 30 | How closely does the company match your ideal customer profile (industry, size, region, model)? |
| Signal strength | 25 | How relevant and significant are its recent buying signals for what you sell? |
| Signal recency | 20 | How recent is the newest relevant signal? |
| Persona relevance | 15 | Are the people you need to reach (your target roles) present or involved? |
| Data confidence | 10 | How reliable, verifiable and complete is the evidence? |
Fit gets the largest weight because a perfectly timed signal at the wrong kind of company rarely becomes a good customer. Signal strength and recency together make up nearly half the score, because timing is what separates “someday” from “this week”. Persona relevance rewards accounts where you can reach the right buyer, and data confidence stops thin or unverified evidence from pushing an account to the top.
What do the numbers mean?
- Below 55: keep watching. The company may fit, but there is not enough recent, relevant evidence to act.
- 55 to 69: an opportunity. Worth research and outreach this week, with some uncertainty.
- 70 and above: high confidence. Good fit, strong recent signals and solid evidence. Contact these first.
These are the thresholds SLOE uses. If you build your own model, pick thresholds that match how many accounts you can actually work each week, and adjust them as you learn.
How do you run this model in a spreadsheet?
Rate each factor from 0 to 5, then convert the rating into points with rating ÷ 5 × weight. Because the weights are 30, 25, 20, 15 and 10, each rating step is worth 6, 5, 4, 3 and 2 points respectively, which keeps the arithmetic simple. The rubric below is one way to rate each factor by hand; adapt it to your market.
| Factor | Rated 5 when… | Rated 1–2 when… |
|---|---|---|
| ICP fit | Matches every key ICP criterion | Matches only one or two criteria |
| Signal strength | Signal is directly tied to your buyer or problem (e.g. hiring the role that uses your product) | Signal is generic or loosely related |
| Signal recency | Newest relevant signal is from the past week or two | Newest relevant signal is several months old |
| Persona relevance | You know the target role exists and who holds it | You are not sure the right role exists |
| Data confidence | Evidence is from official, dated sources | Evidence is second-hand or undated |
Set up one row per company and one column per factor, add a total column, and sort by the total each week.
Worked example: scoring three accounts
| Company | ICP fit (30) | Strength (25) | Recency (20) | Persona (15) | Confidence (10) | Score |
|---|---|---|---|---|---|---|
| Harbourline Logistics | 5 → 30 | 4 → 20 | 5 → 20 | 4 → 12 | 5 → 10 | 92 |
| Copperleaf Retail | 4 → 24 | 3 → 15 | 3 → 12 | 3 → 9 | 4 → 8 | 68 |
| Brightmoor Foods | 3 → 18 | 4 → 20 | 1 → 4 | 2 → 6 | 2 → 4 | 52 |
Harbourline (92) fits perfectly, posted the exact roles that use the product last week and has a clear buyer: high confidence, contact first. Copperleaf (68) is an opportunity: decent fit and a new leader a few weeks ago, worth outreach this week. Brightmoor (52) had a strong-looking signal, but it is months old and the evidence is thin, so it stays on the watch list.
Notice that Brightmoor’s signal strength is as high as Harbourline’s. The score keeps a single impressive signal from outranking a better-fitting, fresher account.
Where does AI fit into account scoring?
AI is good at reading job descriptions and news, summarising what changed and explaining it in plain language. It is less suited to producing the score itself, because a model’s output can vary between runs and is hard to audit.
A practical split is: a fixed formula produces the number; AI explains it. That is how SLOE works. Every watched company gets a deterministic 0–100 score from the five factors above. For each opportunity, AI writes a short “why now” explanation grounded in the evidence, with a recommended next action and who to contact, and it never changes the score.
What mistakes should you avoid?
- Too many factors. Each extra factor dilutes the ones that matter and makes the score harder to explain.
- No recency decay. Without a recency factor, old signals keep accounts at the top long after the moment passed.
- Scoring only fit. A fit-only score gives you the same ranking every week and says nothing about timing.
- Never checking results. Compare scores with what happened: did high-scoring accounts reply and convert? Adjust your rubric and weights if not. See also lead scoring.
- Letting the score replace judgment. A score ranks your list; it does not tell you what the prospect needs. That still comes from research and conversation.
Key takeaways
- Score companies on fit and timing together; either alone gives a misleading ranking.
- Five factors are enough: ICP fit 30, signal strength 25, recency 20, persona relevance 15, data confidence 10.
- Use clear thresholds, such as 55 for “act on this” and 70 for “high confidence”, so the score drives decisions.
- A 0–5 rating per factor makes the model easy to run in a spreadsheet.
- Keep the score deterministic and explainable; use AI to explain the evidence, not to move the number.
Frequently asked questions
What is account scoring?
Account scoring gives each target company a number that combines how well it fits your ideal customer profile with how likely it is to buy soon, so you know which companies to contact first. Unlike lead scoring, it scores the whole company rather than one person.
What factors should an account score include?
A simple, explainable model uses five: ICP fit, signal strength, signal recency, persona relevance (whether you can reach the right buyer) and data confidence (how reliable the evidence is). Fit and signals should carry most of the weight.
How does SLOE score accounts?
SLOE gives every watched company a deterministic 0–100 score from five weighted factors: ICP fit (30), signal strength (25), signal recency (20), persona relevance (15) and data confidence (10). A company becomes an opportunity at 55, and 70 or more is high confidence. AI writes the explanation but never changes the score.
Can I score accounts without software?
Yes. Rate each factor from 0 to 5 in a spreadsheet, convert each rating into points using the factor’s weight, and add them up. It takes a few minutes per company and works well for a list of a few dozen companies.
Should I use AI to score accounts?
AI is useful for reading evidence and explaining it in plain language. For the score itself, a deterministic formula is easier to trust: the same inputs always give the same number, and you can see exactly why an account ranks where it does.
How often should account scores be updated?
Whenever new signals appear, and at least weekly if you are actively selling, because recency is part of the score. A company that scored highly a month ago may have cooled off.