If you’re running account based marketing (ABM), you already know bad data is the enemy. Messy CRM exports, duplicate company records, and mismatched firmographics will kill your campaign targeting. This guide is for marketers, data ops folks, and sales ops teams who are tired of fighting spreadsheets and want to actually use their data. We’ll walk through how to use Tamr, a data mastering tool, to prep your ABM data so it’s actually useful—warts, workarounds, real-world caveats and all.
Why bother with data prep for ABM?
ABM only works if you’re targeting the right accounts with the right info. But most company and contact data is a mess:
- Duplicates everywhere (“Acme Inc.” vs. “Acme Incorporated”)
- Outdated firmographics (company size, industry, revenue)
- Fragmented account hierarchies (parent/child relationships are fuzzy)
- Multiple sources (CRM, LinkedIn, purchased lists) that don’t sync
If you don’t fix this, your campaigns will miss the mark, your reps will waste time, and you’ll never trust your metrics. Tamr helps by cleaning, unifying, and enriching your data—if you set it up right.
Step 1: Gather Your Data Sources (and Know Their Flaws)
Before you even log into Tamr, get your arms around your data. Here’s what matters:
- CRMs: Salesforce, HubSpot, or whatever you use. Export company, contact, and opportunity data.
- Purchased lists: ZoomInfo, Dun & Bradstreet, others. Realize these are full of duplicates and stale records.
- Marketing automation: Marketo, Pardot, etc. Contact engagement data is often siloed.
- Other sources: Website form fills, LinkedIn, event lists.
Pro tip: Make a quick spreadsheet listing each data source, what fields they have, and any quirks (“List A has no domains”; “CRM is full of old junk”).
What to ignore: Don’t bother with tiny, outdated lists or sources you can’t match to accounts. Focus on the 80% that matters.
Step 2: Clean Up Your Data Before Tamr (Seriously)
Tamr is powerful, but it’s not magic. Garbage in, garbage out. Pre-clean your data to save headaches:
- Remove exact duplicates. You don’t need Tamr to spot “Acme Inc.” listed twice with the same domain.
- Standardize columns. Make sure “company_name” is “company_name” everywhere, not “Account Name” or “BusinessName.”
- Fix obvious errors. Nulls where you need domains, broken email addresses, etc.
- Sample your data. Open a few records from each source. You’ll spot weird formatting or issues fast.
Pro tip: Do NOT try to “over-clean” at this stage. Just get rid of the worst offenders so Tamr can do its job.
Step 3: Load Data into Tamr and Set Up Projects
Now, bring your data into Tamr.
- Upload each source as a separate dataset. Tamr works best when it can see the differences.
- Set up a “Golden Records” project. This is Tamr’s way of building a clean, unified view.
- Map fields. For each dataset, tell Tamr which columns are which (“company_name,” “website,” “industry,” etc.).
This step is pretty straightforward, but don’t rush field mapping. If you mismatch fields, your results will be junk. Tamr’s UI helps, but double-check.
What to ignore: Don’t upload every single field. Focus on what matters for ABM: company name, domain, industry, employee count, revenue, and whatever segments you actually use.
Step 4: Configure Matching Rules (The Real Work)
This is where Tamr earns its keep. The platform uses machine learning and rules to find duplicates, variations, and related entities. But you have to set the logic.
- Set up blocking rules. These are basic filters—e.g., only compare records with the same domain extension. This keeps things fast.
- Define match rules. Decide what counts as a “match.” Examples:
- Same website? Strong match.
- Same company name and city? Possible match.
- Everything matches except industry? Flag for review.
- Train the model. Tamr will show you pairs of records and ask, “Are these the same company?” You approve or reject, and the system learns.
Pro tip: Don’t trust the default settings. Every company’s data is different. Spend time reviewing matches—especially the first few batches.
What works: Tamr is strong at catching the “Acme Inc.” vs. “Acme Incorporated” vs. “ACME, LTD” mess. It’s also good at finding parent/child relationships if you have enough data.
What doesn’t: If your data is missing key fields (like domains), Tamr can only guess so much. And if your sources are inconsistent (one uses legal names, one uses trade names), you’ll get false positives.
Step 5: Review and Validate Matches (“Human in the Loop”)
No matter how fancy the AI, you need to check its work.
- Review edge cases. Tamr will flag uncertain matches (“Acme Corp” in New York vs. “Acme Corp” in Texas). Decide if they’re the same.
- Spot check big accounts. Make sure your top targets (the ones sales cares about) are matched correctly.
- Iterate. Update rules, retrain the model, and rerun. This isn’t a “set it and forget it” process.
Pro tip: Get a marketing or sales ops person involved—they know which accounts matter, and which ones are just noise.
Step 6: Enrich and Fill Gaps
Once you’ve got a unified list of accounts, you’ll probably notice gaps—missing industry codes, no employee count, etc.
- Feed in enrichment data. Tamr can merge in fields from trusted sources (like Dun & Bradstreet, LinkedIn, etc.).
- Set up priority rules. If two sources disagree on revenue, pick the one you trust more.
- Create standard segments. Tag accounts by industry, size, or whatever matters for your ABM plays.
What to ignore: Don’t try to fill every single blank. Focus on the fields you’ll actually use for targeting or personalization.
Step 7: Export “Golden Records” for ABM
Now you’ve got a clean, deduped, and enriched set of accounts—the “golden records.” Time to put them to work.
- Export to your CRM and marketing tools. Use Tamr’s export options or APIs. Make sure you map fields correctly.
- Set up regular refreshes. Data decays fast. Schedule Tamr runs monthly or quarterly, not just once a year.
- Document your process. Write down what rules you used, which sources you trust, and what “good” looks like. You’ll thank yourself later.
Pro tip: Don’t do a big bang update. Start with a test batch—say, your top 1,000 accounts—before pushing to your entire CRM.
What to Watch Out For
- Don’t believe the “fully automated” hype. Tamr is powerful, but you’ll need to roll up your sleeves.
- Garbage in, garbage out. Bad sources or missing data will still trip you up.
- Siloed teams kill this process. Marketing, sales, and data ops need to actually talk to each other.
- Cost and complexity. Tamr isn’t cheap or trivial to implement. Make sure your team has the appetite (and budget) for it.
Keep It Simple, Iterate, and Don’t Chase Perfection
Perfect data doesn’t exist. The real win is getting your ABM team a clean enough list to target the right accounts, then improving it over time. Start with your top segments, fix the worst issues, and add bells and whistles later. Don’t let the quest for “single source of truth” stall your campaigns. Clean, actionable, and “good enough” beats pretty dashboards every time.