Here's how we automate project distribution for a 60-person writing team (something our client's team originally thought was impossible, now they love it)
(This is relevant for sales teams as well - to auto-distribute leads)
We've tested this over the course of months in our client's business, and it outperforms the humans in ACCURACY by 30% (and obviously is much faster).
They continue to love and use it years later
Here's how to do it
To rank/distribute projects accurately, we needed to take into consideration:
-basic airtable setup - projects, team assigned to each project, when the project was due, when it was submitted, size of project, quality mistakes found & their severity, and the log of the past attempts to assign a project to a given person (and whether or not they accepted and how long it took)
-topic/project type preferences for each writer (ie: some project types require a specialist, while some can be done by anyone)
-planned holiday/sick time for the writer
-planned holiday/sick time for the quality/proofreading team (when the QA team is lower in capacity, more projects are biased to go towards expert writers)
Once you have all that tracked, you can calculate:
-"timeliness" (based on a "lateness factor" - ie: a score based on how large the project was, how long the freelancer had to do it, and when they submitted it in relation to the deadline..late isnt always late)
-"receptivity" (ie: how quickly they say yes to projects and how often they say yes)
-"quality" - the average number of errors & severity of those errors found in past projects (link in the comments to a recent youtube video where I go over how to track this)
-estimated capacity (if on holiday capacity is zero, otherwise we estimate current capacity based on past output, whether projects were submitted with errors or late, and whether the writer accepted a lot of projects or denied some - all of which we use as signals to either estimate that their actual capacity is slightly greater or less than previous output - ie: if they did things on time and well, we think they can handle more work...if they deny lots of projects and submit late or with mistakes, we think they are overloaded, and this estimated capacity is therefore constantly adjusting)
-all of the above is tracked in Airtable on a rolling 14 day basis, so the algorithm is constantly adjusting to account for variances in individual schedule, motivation, etc
Then, we have 3 relevant Make scenarios to assign any given project
Part 1: Onboarding automation
-gets the project details
-pass information into the "set topic" to classify the project based on the text (so we know which subset of writers is relevant)
-when the topic is returned, pass to the "autoassign writer(s)" scenario
-once a writer is determined, create a draft email (so that the suggested writer can always be changed
Part 2: Determine topic
-this part was added later, as the team told us that some writers had far better quality on some topics (and were faster) than others
Part 3: Autosuggest a writer
-there are 3 different levels of writer - noobies, mid-level writers, and experts. First, we need to find which level is required (ie: if it's a large project or with a tight timeline, then it must go to an expert writer...if it's an easy project & long timeline, and there is at least one noobie available then it must go there to make sure the beginners are trained up)
-once we get the writer level, we determine whether the topic narrows the list at all
-find the highest ranked writer (ie: we just pick the first record, sorted by the auto-suggest score)
-...that we think has availability (ie: where the workload for the project does not make them exceed their estimated capacity)
-If we can't find anyone that we estimate has the availability to do the work, then return a list of the top 5 ranked people so the project can be sent to all of them at once
And that's how you automate project distribution (btw our client scaled to 100 projects per day with this system)
Happy automating