diff --git a/.evergreen/config.yml b/.evergreen/config.yml
index 6f83976f..a043c638 100644
--- a/.evergreen/config.yml
+++ b/.evergreen/config.yml
@@ -113,6 +113,15 @@ functions:
         add_expansions_to_env: true
         args:
           - .evergreen/run-granular-benchmarks.sh
+  run custom benchmarks:
+    - command: subprocess.exec
+      type: test
+      params:
+        working_dir: src
+        binary: bash
+        add_expansions_to_env: true
+        args:
+          - .evergreen/run-custom-benchmarks.sh
   run spec benchmarks:
     - command: subprocess.exec
       type: test
@@ -246,6 +255,19 @@ tasks:
       - command: perf.send
         params:
           file: src/test/bench/etc/resultsCollectedMeans.json
+  - name: run-custom-benchmarks
+    commands:
+      - func: fetch source
+        vars:
+          # This needs to stay pinned at Node v18.16.0 for consistency across perf runs.
+          NODE_LTS_VERSION: v18.16.0
+      - func: install dependencies
+        vars:
+          NPM_VERSION: 9
+      - func: run custom benchmarks
+      - command: perf.send
+        params:
+          file: src/customBenchmarkResults.json
   - name: run-spec-benchmarks
     commands:
       - func: fetch source
@@ -300,3 +322,4 @@ buildvariants:
     tasks:
       - run-granular-benchmarks
       - run-spec-benchmarks
+      - run-custom-benchmarks
diff --git a/.evergreen/run-custom-benchmarks.sh b/.evergreen/run-custom-benchmarks.sh
new file mode 100644
index 00000000..2984cb35
--- /dev/null
+++ b/.evergreen/run-custom-benchmarks.sh
@@ -0,0 +1,6 @@
+#!/usr/bin/env bash
+
+source "${PROJECT_DIRECTORY}/.evergreen/init-node-and-npm-env.sh"
+set -o xtrace
+
+npm run check:custom-bench
diff --git a/.gitignore b/.gitignore
index 39b98b7b..b2ee050b 100644
--- a/.gitignore
+++ b/.gitignore
@@ -33,3 +33,4 @@ docs/public
 .nvmrc
 
 benchmarks.json
+customBenchmarkResults.json
diff --git a/package.json b/package.json
index fd1fa890..6fed86fa 100644
--- a/package.json
+++ b/package.json
@@ -105,6 +105,7 @@
     "check:web-no-bigint": "WEB=true NO_BIGINT=true mocha test/node",
     "check:granular-bench": "npm run build:bench && node ./test/bench/etc/run_granular_benchmarks.js",
     "check:spec-bench": "npm run build:bench && node ./test/bench/lib/spec/bsonBench.js",
+    "check:custom-bench": "npm run build && node ./test/bench/custom/main.mjs",
     "build:bench": "cd test/bench && npx tsc",
     "build:ts": "node ./node_modules/typescript/bin/tsc",
     "build:dts": "npm run build:ts && api-extractor run --typescript-compiler-folder node_modules/typescript --local && node etc/clean_definition_files.cjs",
diff --git a/test/bench/custom/benchmarks.mjs b/test/bench/custom/benchmarks.mjs
new file mode 100644
index 00000000..74170fb9
--- /dev/null
+++ b/test/bench/custom/benchmarks.mjs
@@ -0,0 +1,22 @@
+/* eslint-disable strict */
+import { BSON } from '../../../lib/bson.mjs';
+
+const ObjectId_isValid = [
+  function objectid_isvalid_wrong_string_length() {
+    BSON.ObjectId.isValid('a');
+  },
+  /** wrong character at the start, could be the most short circuited code path */
+  function objectid_isvalid_invalid_hex_at_start() {
+    BSON.ObjectId.isValid('g6e84ebdc96f4c0772f0cbbf');
+  },
+  /** wrong character at the end, could be the least short circuited code path */
+  function objectid_isvalid_invalid_hex_at_end() {
+    BSON.ObjectId.isValid('66e84ebdc96f4c0772f0cbbg');
+  },
+  function objectid_isvalid_valid_hex_string() {
+    BSON.ObjectId.isValid('66e84ebdc96f4c0772f0cbbf');
+  }
+];
+
+// Add benchmarks here:
+export const benchmarks = [...ObjectId_isValid];
diff --git a/test/bench/custom/main.mjs b/test/bench/custom/main.mjs
new file mode 100644
index 00000000..b375418f
--- /dev/null
+++ b/test/bench/custom/main.mjs
@@ -0,0 +1,39 @@
+/* eslint-disable strict */
+
+import util from 'node:util';
+import fs from 'node:fs';
+import os from 'node:os';
+import benchmark from 'benchmark';
+import { benchmarks } from './benchmarks.mjs';
+
+const hw = os.cpus();
+const ram = os.totalmem() / 1024 ** 3;
+const platform = { name: hw[0].model, cores: hw.length, ram: `${ram}GB` };
+
+const systemInfo = () =>
+  [
+    `\n- cpu: ${platform.name}`,
+    `- cores: ${platform.cores}`,
+    `- arch: ${os.arch()}`,
+    `- os: ${process.platform} (${os.release()})`,
+    `- ram: ${platform.ram}\n`
+  ].join('\n');
+console.log(systemInfo());
+
+const suite = new benchmark.Suite();
+
+for (const bench of benchmarks) suite.add(bench.name, bench);
+
+suite
+  .on('cycle', function logBenchmark(event) {
+    console.log(String(event.target));
+  })
+  .on('complete', function outputPerfSend() {
+    const data = Array.from(this).map(bench => ({
+      info: { test_name: bench.name },
+      metrics: [{ name: 'ops_per_sec', value: bench.hz }]
+    }));
+    console.log(util.inspect(data, { depth: Infinity, colors: true }));
+    fs.writeFileSync('customBenchmarkResults.json', JSON.stringify(data), 'utf8');
+  })
+  .run();
diff --git a/test/bench/custom/readme.md b/test/bench/custom/readme.md
new file mode 100644
index 00000000..ed69ae21
--- /dev/null
+++ b/test/bench/custom/readme.md
@@ -0,0 +1,38 @@
+# Custom Benchmark Tests
+
+In this directory are tests for code paths not covered by our spec or granular (de)serialization benchmarks.
+
+## How to write your own
+
+In `benchmarks.mjs` add a new test to an existing array or make a new array for a new subject area.
+Try to fit the name of the function into the format of: "subject area", "method or function" "test case that is being covered" (Ex. `objectid_isvalid_bestcase_false`).
+Make sure your test is added to the `benchmarks` export.
+
+### Example
+
+```js
+const ObjectId_isValid = [
+  function objectid_isvalid_strlen() {
+    BSON.ObjectId.isValid('a');
+  },
+  // ...
+];
+
+export const benchmarks = [...ObjectId_isValid];
+```
+
+## Output
+
+The JSON emitted at the end of the benchmarks must follow our performance tracking format.
+
+The JSON must be an array of "`Test`"s:
+
+```ts
+type Metric = { name: string, value: number }
+type Test = {
+    info: { test_name: string },
+    metrics: Metric[]
+}
+```
+
+The metric collected is always "ops_per_sec" so higher is better.