Your scoring model defines how we score and enrich the accounts in your active account list and the net new accounts in your discover tab. This model populates the tier, overall fit, individual score, and signal fields in Keyplay.
The scoring model applies to accounts both within and outside your SAM. You can build and update your scoring model by clicking your account name in the top right corner, then Account Scoring, in your Keyplay account as shown here:
Edit your scoring model:
From there, you’ll be able to add/remove signals, customize signal weights, and see the math on a real account as you tweak and tune.
Overview
When you navigate to account scoring, you'll be taken to the scoring modal where you can set up each piece of your custom account scoring within Keyplay. The overview tab gives you a quick glance at your scoring set up, as well as allows you to choose which score mode you want to use.
There are 5 score modes available in Keyplay:
For most customers, the "Signal score + bonus points for similarity gives the best possible results as it accounts for both the number of signals matched as well as your lookalikes, giving the best holistic overall score.
Lookalikes
On the Lookalikes tab, you can upload or manually add accounts that you want Keyplay to use in order to surface similar accounts. Most customers use their customer list for their lookalikes. The most important criteria is that it's a known good fit account for your ICP.
If uploading a CSV, make sure you include a "domain" column with the domains of the accounts you want to upload.
Signals
You can specify if a signal is positive or negative. Positive signals add points to a specific account when found, and negative signals lose points on an account when found. Drag the relevant signals into each bucket in order to add them to your model.
Signal Weights
You can also change how signals are weighted in your model. Signal weighting gives you the ability to change how much a specific signal counts towards your overall scoring.
You can click directly on a signal pill to update the weightings, or you can click "Customize all weightings" to edit weightings on your signals in bulk.
Signals can have a weighting of anywhere from 1x to 20x. We suggest that signals which are most significant to your ICP have a higher weighting, while signals that are less significant to your weighting have a lower weighting.
Signal Weighting Tips
Tip 1: Try not to overweight any one signal, as this will give that signal more emphasis in the model and make scoring less holistic.
Tip 2: Try to use 30 signals or less in your model in order to laser in on the best accounts and not muddy the waters.
Tip 3: Overall, expect to see best accounts scoring as As and Bs. Use the test tab and the example account feature to test that your best accounts are scoring as you expect.
Test
You can also test your model in real time by uploading lists to the test tab. We suggest uploading a customer list, a high ACV customer list, and a list of disqualified accounts that you know are bad fits. You can also generate a control list that is a sample based on your SAM definition.
We recommend using this as a barometer for how well the scoring model is matching your ICP. Your customer list and high ACV account lists should score mostly As and Bs, while your control and bad fit account lists should score lower.
Account Preview
You can also click the eye in the top right corner next to "Exit and Score" to preview the score for any account and get a breakdown of why an account scored a certain way. This can be helpful in understanding why accounts are receiving scores and how the different factors are contributing to the scoring model.
Exit and Score
When you are ready to score your accounts, you can hit the exit and score button. This will trigger a refresh of all your account in order to update them with the latest scoring model.
