Back to blog
7 min read

CRM Hygiene Runbook: Re-Enrich on a Schedule

By Kooperativa Engineering

CRM Hygiene Runbook: Re-Enrich on a Schedule

A CRM hygiene routine that lasts is a schedule, not a clean-up project. Give every record a stable key, store the date it was last verified, re-check employer and title on a fixed rhythm with a script, and decide in advance which fields the script may overwrite.

This runbook covers the part of hygiene that goes stale because people change jobs. It does not cover bad email addresses or phone numbers: Kooperativa holds professional profile data and no contact details, so email validity needs a separate verifier. Everything below is about the employer, the title and the seniority sitting in your CRM.

The rhythm matters because people change employers on a timescale you can put a number on. The US Bureau of Labor Statistics reported in September 2026 that wage and salary workers had a median of 4.1 years with their current employer in January 2026. For workers aged 25 to 34 the median was 3.0 years, so half of them had started their current job within the last three years. Management occupations had the longest tenure of the occupation groups it compares, at 6.1 years. These figures describe all US workers, not buyers in particular, but they show the pattern: younger contacts move often and senior ones less so, which is one reason to sweep contacts on open deals weekly and everyone else less often.

Start with a key that does not change

Names collide and companies get renamed, so a hygiene routine needs one identifier per person that survives both. Store the profile username and the internal id the first time a contact is matched, and use them for every later check. Two CRM contacts that resolve to the same id are duplicates, whatever their names and email addresses say.

Then add a few fields of your own, so each record can tell you how old your belief about it is.

Custom fields on a CRM contactjson
{
  "profile_username": "EXAMPLE-USERNAME",
  "profile_id": "PERSON_ID",
  "company_id_checked": "COMPANY_ID",
  "title_checked": "Director of Operations",
  "last_checked_at": "2026-11-21",
  "record_fetched_at": "2026-10-28",
  "check_result": "verified"
}

The two dates are the point. last_checked_at is when your script looked. record_fetched_at is when the data behind the answer was collected, and it can be much older.

A weekly sweep in batches of 100

Bulk enrich takes up to 100 profiles per call and returns each person's current company, current title and a fetched_at timestamp. Compare those with what the CRM holds and sort every record into one of five outcomes: left the account, changed title, verified, verified but old, or unmatched.

Sweep CRM contacts and classify each onejs
const BASE = 'https://kooperativa.io/api/v1';
const headers = {
  Authorization: 'Bearer ' + process.env.KOOPERATIVA_API_KEY,
  'Content-Type': 'application/json',
};
const MAX_AGE_DAYS = 90;

const norm = (s) => (s || '').toLowerCase().replace(/\s+/g, ' ').trim();

function chunk(list, size) {
  const out = [];
  for (let i = 0; i < list.length; i += size) out.push(list.slice(i, i + size));
  return out;
}

// crmRows: [{ crm_id, username, company_id, title }]
export async function sweep(crmRows) {
  const actions = [];
  for (const batch of chunk(crmRows, 100)) {
    const res = await fetch(BASE + '/people/bulk-enrich', {
      method: 'POST',
      headers,
      body: JSON.stringify({ profiles: batch.map((r) => ({ username: r.username })) }),
    });
    if (!res.ok) continue; // leave the batch unchecked and try again next run

    const { profiles } = await res.json();
    const byUsername = new Map(profiles.map((p) => [p.username.toLowerCase(), p]));

    for (const row of batch) {
      const p = byUsername.get(row.username.toLowerCase());
      if (!p) {
        actions.push({ crm_id: row.crm_id, action: 'unmatched' });
      } else if (p.current_company_id && p.current_company_id !== row.company_id) {
        actions.push({ crm_id: row.crm_id, action: 'left_account', now_at: p.current_company });
      } else if (norm(p.current_title) !== norm(row.title)) {
        actions.push({ crm_id: row.crm_id, action: 'title_changed', title: p.current_title });
      } else {
        // fetched_at is an ISO string on full records
        const ageDays = (Date.now() - new Date(p.fetched_at)) / 86400000;
        actions.push({
          crm_id: row.crm_id,
          action: ageDays > MAX_AGE_DAYS ? 'verified_but_old' : 'verified',
        });
      }
    }
  }
  return actions;
}

Read the outcomes with care. left_account is the valuable one: the contact has moved, the account owner needs a task, and the account may need a new champion. Unmatched means the call did not return the person, either because the profile is not indexed yet or because the lookup failed, so it is never evidence that someone left. A bulk call does not say why a person is missing, so check each unmatched record with a single lookup before you decide anything. And verified_but_old is the honest answer when the profile still shows the same employer but the record was collected months ago. It is the last known state, not today's.

One parsing detail: fetched_at is an ISO timestamp on full records and a number of seconds in search results, so parse it according to the endpoint that returned it.

Spend live lookups where they pay

A record older than you can tolerate is a candidate for a live lookup. The realtime person endpoint fetches the current profile from the source and costs one credit, $0.001, per call, and it needs an account with billing set up. Keep it for the contacts whose accuracy changes a decision: people on open opportunities, champions, and anyone about to be contacted. Everyone else can wait for the next sweep.

Let monitors catch the moves that cannot wait

A weekly sweep sees a job change up to a week late. For the few people where that matters, a monitor sends a signed webhook when a tracked field changes, so there is nothing to poll. A workspace can hold 500 monitors, so reserve them for champions and economic buyers on open deals and let the sweep cover the rest.

Rules for what the script may overwrite

Write these rules down before the first run, because the first time a script overwrites a rep's careful note is the last time the team trusts it. The article on CRM data enrichment write strategy covers the four write policies in depth. What follows is the short version for a script that only touches job fields.

  • Never overwrite a field a person edited by hand. Compare the CRM value with the value your script wrote last time, and propose a change only when they match.
  • Keep the previous value in a history field, so a bad update can be undone.
  • A title change updates the title. A company change creates a task and leaves the account alone, because moving a contact between accounts is a decision.
  • If a single person lookup answers 451, the person has asked for removal. Suppress the record from future checks and outreach and do not retry. The article on 404 versus 451 explains why the two codes need different handling.

Measure the routine, not the data

Three numbers tell you whether the routine works: the share of records checked in the last 90 days, the share that changed on each sweep, and the share that stayed unmatched. A rising unmatched share means your identifiers are drifting. A change rate near zero for months can mean the sweep is reading old records and not that nobody moved, so look at record_fetched_at before you celebrate.

Frequently asked questions

What is CRM hygiene?

The routine of keeping CRM records accurate, deduplicated and current: a stable key per record, a verified date, a fixed re-check rhythm and written rules for what automation may overwrite.

How often should CRM data be cleaned?

Re-check employer and title weekly for active contacts and monthly or quarterly for the rest. People change roles continuously, so the rhythm matters more than any single clean-up.

Does this fix email addresses and phone numbers?

No. It checks employer, title and seniority. Kooperativa does not provide email addresses or phone numbers, so use an email verifier for those.

How do I find duplicate contacts?

Match records on a stable profile identifier instead of name or email. Two contacts that resolve to the same id are one person.

Get started

Try Kooperativa

One API key. Person and company enrichment, structured search, and monitors under one flat license.

Keep reading

We use only cookies that are strictly necessary to run this site (staying signed in, and protecting forms from abuse). Our analytics tool doesn't use cookies or collect personal data. See our Privacy Policy for details.