Which Industries Search for Direct Mail? 2026 Keyword Demand Data for Agencies

Three findings from our analysis

Demand and competitor coverage point to different niches

1. Insurance has less volume, but a larger low-KD share

Real estate contributes 24,740 estimated monthly searches, of which 9,030 (36.5%) sit at KD ≤15. Insurance contributes 3,020, with 2,440 (80.8%) at that threshold. Median known KD is 15 for real estate versus 3 for insurance.

That makes insurance worth reviewing before simply choosing the largest niche. This is a lower bound on low-KD volume: missing KD is excluded, not treated as zero. KD is available for 174 of 328 real estate terms and 20 of 54 insurance terms.

2. Positive CPC and low KD overlap with a tracked coverage gap

Our long-tail filter returns 88 keywords totaling 9,230 estimated monthly searches. Of those, 74 keywords (7,920 searches) have no top-10 organic result observed for any of the four tracked domains. Their median CPC is $12.23 and median KD is 11.

The overlap is a useful page-research queue: advertisers have a positive estimated CPC, difficulty is low by our chosen cutoff, and these four providers are not observed on page one. It does not mean the search results are empty or that a page will rank quickly. The table below includes both gaps and already-covered terms so the difference stays visible.

3. Automotive and mortgage have different visible leaders

Lob has top-10 rows for 38 of 81 automotive keywords; Postalytics has 2. In mortgage, Postalytics has 5 of 52, while Lob has 0. Across all four providers, observed top-10 coverage reaches 49.4% of auto keywords versus 9.6% of mortgage keywords.

An agency investigating automotive pages should inspect Lob's ranking pages; for mortgage, Postalytics is the more visible reference in this sample. This is coverage within supplied exports, not market share or a judgment of either service.

Industry demand, KD and CPC

A zero CPC median is not evidence of free advertising or weak demand. We retain the export's zeroes and show how many terms have positive CPC. Dental is separated from healthcare; mortgage and insurance are separated from other financial services. Small categories, especially legal, are too thin for broad market conclusions.

The first long-tail terms to review

Our proxy for the fastest-to-win long-tail: at least four whitespace-separated words, volume ≥20, known KD ≤15, CPC >$0 and no navigational intent label. Sort by volume descending, KD ascending, CPC descending, then keyword. Take one term per industry first, then fill to 12. This prioritizes investigation; we did not measure ranking speed.

Use related phrases to improve one useful industry page rather than building a page for every word-order variation. CPC helps flag advertiser interest, but it is not an expected acquisition cost. The dental and mortgage phrases, for example, have positive CPC and observed competitor positions outside the top 10; current search results still need a fresh review.

Which competitors rank in each industry?

Each cell counts unique retained keywords with a top-10 / top-100 organic position for that domain. Domains can overlap on the same keyword. A zero means no qualifying row in these files, not proof of absence from Google.

The sample is partly built from competitor exports, so it is biased toward terms these providers already cover. Thanks.io has very little observed coverage in the retained industry sample. This does not measure its full keyword footprint, service range or customer base.

Intent is not the same as buying readiness

We normalize Semrush's intent labels and preserve mixed labels as one exclusive bucket. Missing labels stay Unknown. We do not infer buying intent from a high CPC or from an industry name.

Informational labels can still include service-shaped phrases. Read the actual query and search results before deciding whether it belongs on a service page, an example gallery or an explanatory guide. Keyword difficulty describes organic competition; CPC describes a separate paid-search estimate.

Methods, source dates and reproducible data

We analyzed our saved Semrush exports on October 19, 2026. The upstream pipeline deduplicated by database and normalized keyword, taking the maximum volume, KD and CPC across seed exports, then removed nonpositive volume and noise. Its cleaned file contains 54,462 US keywords; stats.json also records a separate Spanish database that is excluded here. keyword-opportunities.csv is a cluster-level planning file, so we inspected it for context without adding its aggregated volume to keyword totals.

1. Normalize lowercase and whitespace, then require a channel term (postcard, post card, direct mail, mailer or EDDM) and a recognized industry expression. The public script contains every regular expression.
2. Assign exactly one industry by first match: mortgage, insurance, real estate, dental, nonprofit, political, automotive, home services, retail/hospitality, healthcare, other finance, legal. Specific categories precede broad ones. This puts mortgage protection insurance in mortgage and Medicare in insurance.
3. Apply the upstream noise rules for brands, jobs, free downloads/templates, non-US locations and unrelated queries. Also exclude older years, tax e-postcards, identified names and malformed terms, postal-product uses of EDDM Retail, plus question and statistics queries. This page studies industry demand, not response-rate or question demand.
4. Deduplicate exact normalized keywords. Prefer metrics from keywords-clean.parquet, then competitor-gap-keywords.csv, then the most recently timestamped organic competitor row (domain and position break ties). Do not average conflicting sources. Preserve missing KD as null and reported zero CPC as zero. Keep positive-volume records only.
5. The script found 1,461 matching source observations covering 873 unique candidate queries. It excluded 112 queries, leaving 761. Metrics come from the cleaned keyword file for 605, the gap file for 94, and position exports for the remaining 62.
6. Join organic positions 1 to 100 by exact normalized keyword. Keep the best exported position per domain. These are historical observations within the exports, not a fresh, simultaneous SERP audit. Medians are unweighted; sums count each retained keyword once, including word-order variants.

What the dates actually establish

Exact export/download dates are not embedded in the supplied files and could not be verified. The files' UTC modification dates are October 19, 2026; that establishes file metadata only. We record those timestamps and SHA-256 hashes in the raw snapshot. Access date means when we read the local files, not when Semrush measured search volume.

Sources and downloads

Every source below was accessed on October 19, 2026. Semrush links identify report locations for the supplied exports; they are not public, immutable copies of the reports. Each keyword observation in our raw JSON has a source URL, access date and source row. The snapshot also preserves upstream cleaning rules and source hashes.

• analysis/keywords-clean.parquet
• analysis/competitor-gap-keywords.csv
• semrush/competitors/lob.com/positions.csv
• semrush/competitors/postalytics.com/positions.csv
• semrush/competitors/thanks.io/positions.csv
• semrush/competitors/handwrytten.com/positions.csv
• analysis/keyword-opportunities.csv
• analysis/stats.json
• analysis/scripts/build.py
• Fer Marketing published postcard offer (fetched public page; $10 per postcard, minimum 50)

Raw observations and provenance (JSON) · Clean classified keywords (JSON) · Published aggregates (JSON) · Analysis script (Python)

In the repository, rebuild from the committed snapshot with Python's standard library. To recollect from the supplied source files, use --collect /workspace/seo; collection requires pandas, pyarrow and curl and verifies the published postcard offer.

Limits

This is a seed- and competitor-derived convenience sample, not all US direct mail demand. 365 retained keywords have no KD, so difficulty comparisons have uneven coverage. Semrush's volume and CPC are estimates, not first-party Google counts. Keyword variants may overlap; volume sums are not unique searchers. A low KD, positive CPC or missing tracked competitor cannot establish lead quality, resale margin or campaign performance. Rule-based classification can miss synonyms or misassign ambiguous queries. We did not verify current rankings, seasonality or changes over time.

From a niche shortlist to a postcard offer

Start with a sector your agency already understands. Use the keyword table to define the audience and the page it needs, then choose an offer the recipient can act on. Our industry guides cover real estate postcards, insurance direct mail, postcard postage and direct mail postcards.

Agencies can resell Fer Marketing's personalized postcard campaigns under their own brand. The published offer is $10 per postcard, with a minimum of 50 postcards ($500). It includes prospect research, website review, design, printing and USPS First-Class postage. This is a campaign for agency client acquisition; the industry search estimates do not promise results for a mailing.