Login
Revenue & Growth

Customer Expansion Strategy: Test AI Before Hiring

Adi Gorelik
-
Base AI
6
min read
-
Published
November 20, 2026
Blog cover titled "Test AI before hiring" on customer expansion strategy, with a Read more button

The short answer

A customer expansion strategy for today starts by testing what AI can take on, before adding CSMs to the old playbook. Give AI the watching and drafting, then run four moves between renewals: find the signal, build the proof, give the champion a program, and hand off to sales. Each move has one owner, one AI job and one metric.

Look at the pattern. Median NRR fell even as companies hired more customer success roles. The two moved together, which doesn't prove a cause. Base AI's read: more people running the same motion bought more of the same result.

What is a customer expansion strategy?

A customer expansion strategy is the plan for growing revenue inside existing accounts through more usage, a higher tier (upsell) or an added product (cross-sell). In Base AI's read, the 2021 version leaned on seat growth. When customers stopped hiring, that growth disappeared, and the playbook had no step for building a business case.

Accounts grow in three ways, as Base AI's glossary entry on account expansion explains:

  • Depth: more users or more usage on the product they own
  • Height: a higher tier of the same product
  • Breadth: a second product that complements the first

In 2021, depth often arrived on its own. Height and breadth never did. They need a reason, a champion and a business case, and someone has to build all three.

More CS hires, lower NRR

Median net revenue retention in B2B SaaS fell from 105% in 2021 to 101% in 2024, across 563 companies in Benchmarkit's 2025 SaaS Performance Metrics report (via Ray Rike's summary).

Median net revenue retention fell from 105% in 2021 to 101% in 2024
Median NRR in Benchmarkit's 2025 report, 563 companies.

Hiring went the other way, by Bain's account. Its December 2025 report says NRR "has declined even as companies have invested in hiring more customer success roles," though it gives no hiring figure.

2025 brought no recovery. The 2026 benchmarks from Aleph and Benchmarkit, on full-year 2025 data from 342 companies, put median NRR at 102%. That edition charts 102%, 101%, 101% and 102% for 2022 to 2025, on a different sample from the 2021 figure.

So most of the drop came early and the line has been flat since. The 25th percentile fell to 92%.

Most expansion only refills churn

The same 2026 benchmarks put median gross revenue retention (GRR) at 84%. GRR counts what you keep before any expansion. NRR adds expansion back in.

So a company at 84% GRR needs about 18 points of expansion to finish at 102%. About 16 of them refill what churn and downgrades took. About 2 are growth.

Of about 18 points of expansion at the median, about 16 refill churn and downgrades and about 2 are growth
Our arithmetic on the 2026 medians. Illustrative.

Medians do not add exactly, and the NRR median comes from 230 of the 342 companies, so treat this as an illustration. Still, close to nine in ten expansion points look defensive. Adding CSMs to that motion may only buy more backfill.

Three assumptions to retest with AI

Most expansion plans rest on three beliefs. Each was reasonable in 2021. AI now makes them cheap to test.

  1. A green account is ready to expand. A health score is built to predict retention. Forrester's Laura Ramos, writing on AI in post-sale work, lists two uses for signals: mitigating churn or spotting growth. Most scores were built for the first.
  2. Renewal is the moment to ask. In most enterprise renewals, next year's budget is set by renewal week. The behaviour that predicts expansion shows up mid-contract: a new team logs in, usage hits the plan ceiling, a champion gets active. AI can watch for that every week, on every account.
  3. CSM time is the bottleneck. Bain's August 2025 survey found CSMs spend about two-thirds of their time on lower-value work AI could automate. A new CSM inherits the same calendar.

AI alone doesn't fix this. In competitor ChurnZero's 2026 study of 580 CS and revenue leaders, all self-reported, 81% said AI cut time on manual tasks, and half had no consistent way to measure its impact. The release adds that AI's impact has not yet shown up in NRR. Saved hours aren't expansion until someone points them at accounts.

Buyers moved too. Benchmarkit's 2026 report, with NRR data from 230 companies, names budget scrutiny, slower upsell cycles and commitments deferred while buyers evaluate AI-native options as headwinds on expansion.

Four moves, one AI job each

Run these instead, usually in this order. If a move has no owner, it won't happen. If it has no metric, you won't know it stopped.

Four moves and owners: signal (CS operations), proof (CSM), program (customer marketing), handoff (CSM to AE)
The owner of each move. AI supports all four.
  • Signal (CS operations). AI watches product analytics, CRM, tickets and program activity. People decide which signals count. Metric: accounts with a live signal in the last 30 days. The Base AI platform connects those sources.
  • Proof (CSM). AI drafts the value summary, and the CSM confirms every number with the customer. Metric: signaled accounts with a confirmed summary in the last 90 days. Base's retention solution generates success plans and QBRs from customer data.
  • Program (customer marketing). AI spots who is ready and drafts the invitation. A person makes the ask. Metric: signaled accounts with a champion active in a program this quarter. See Base's advocacy and community solution.
  • Handoff (CSM to AE). AI assembles the packet in the CRM, the CSM makes the introduction, and the AE owns the deal. Metric: handoffs the AE accepts. Base's expansion solution delivers expansion-ready accounts to AEs.

AI reads, drafts and routes. People decide, ask and commit. On the Base AI customer growth platform, you choose what each AI teammate does on its own, what it asks first and what it only suggests.

Three things stay with people, however good the model gets: outcome numbers, the timing of the executive ask, and price.

What AI-powered QBRs did at Vidyard

Vidyard's customer story is the closest live example of the Proof move. Vidyard is a Base AI customer. It tells this as a QBR story, and the mapping is ours.

  • Before: reps pulled data and compiled decks by hand, so quality varied by rep and high-quality QBRs were limited to select accounts.
  • Build: the team built data-driven QBR hubs in AI Studio through prompting, with no developers.
  • Coverage: Vidyard reports 100% QBR coverage, with personalized QBRs for every customer, SMB included.
  • Action: the hubs help customers see usage, identify gaps and take action, like assigning more licenses.

The story reports no expansion number. Vidyard's renewal hubs are still in progress, and license expansion is named as what comes next.

The lesson for your strategy is coverage. When hubs built with AI take the prep out of a review, long-tail accounts get proof they never had, and a license gap can reach the customer before the renewal does.

How do you test it in 90 days?

Test a customer expansion strategy in 90 days with leading measures: more accounts with a live expansion signal, more signaled accounts with customer-confirmed proof, and AEs accepting most CS handoffs. Expansion ARR and NRR lag by a renewal cycle or more, so they confirm the strategy later rather than test it early.

  • Days 1-30: wire three signals you already have data for. Record baseline coverage.
  • Days 31-60: for the top 20 signaled accounts, have AI draft a value summary and a one-page plan. The CSM confirms the numbers. Invite each champion into one program.
  • Days 61-90: make the first handoffs and track acceptance. Review every rejected one with the AE.

Set a kill criterion now. If AEs reject most handoffs at day 90, your signals measure activity, not intent. Go back to Signal before you hire anyone.

Start with your last ten deals

Pull your last ten closed expansion deals. For each one, find the date someone first noticed the opportunity and what they noticed.

If most were first raised in the 60 days before renewal, you are running the 2021 playbook. Start with Signal. If most started mid-contract, the gap is usually proof or program.

To see the four moves on your own customer data, book a Base AI demo and bring the list.

Key Takeaways

  • NRR fell despite hiring. Median NRR slid from 105% in 2021 to 101% in 2024.
  • Most expansion is defensive. Illustrative, on 2025 medians: about 16 of 18 points refill churn.
  • AI gets one job per move. Signal, proof, program, handoff, each with an owner and metric.
  • Test it in 90 days. If AEs reject most handoffs, fix the signals before hiring.

Frequently asked questions

What is a customer expansion strategy?

A customer expansion strategy is the plan for growing revenue inside existing accounts through three motions: depth (more seats or usage on the same product), height (an upsell to a higher tier) and breadth (a cross-sell of a second product). Depth once arrived with customer hiring. Height and breadth need a reason, a champion and a business case, and someone has to build all three.

How do you build a customer expansion strategy?

Build it in four moves. Signal: find growth behaviour mid-contract, defined by CS operations. Proof: build a customer-confirmed business case, owned by the CSM. Program: give the champion visibility through a speaker slot, mentorship, beta or advisory seat, owned by customer marketing. Handoff: pass expansion-ready accounts to the AE with context. Each move gets one owner, one job for AI and one metric.

Why is net revenue retention declining in B2B SaaS?

Median NRR in B2B SaaS slid from 105% in 2021 to 101% in 2024, according to Benchmarkit's 563-company report, and sat at 102% on 2025 data from a different sample. Bain reports the decline came even as companies hired more customer success roles. The benchmarks show the two moving together, not a proven cause. Base AI's read: seat growth stopped delivering expansion once customers stopped hiring.

Do AI-mature customer success teams have higher NRR?

AI-mature customer success teams may have higher NRR, but only by their own account. In ChurnZero's 2026 study of 580 self-reported CS and revenue leaders, 65% of the most AI-mature teams reported NRR of 100% or higher. With median NRR at 101-102%, at least half of all companies clear that bar anyway, and the study gives no figure for less mature teams. It shows correlation, not cause.

Can a customer community drive expansion revenue?

A customer community can feed expansion revenue, mostly as a signal source rather than a closer. Members active in a program show who is engaged and ready to be asked. At Phenom, a Base AI customer, program dashboards show how engagement maps to account health and which customers influence revenue. Phenom's example is retention risk, not a reported expansion figure, so track your own handoffs from community to AE.

How long does a customer expansion strategy take to show results?

Leading indicators should move within 90 days: more accounts with a live expansion signal, more signaled accounts with customer-confirmed proof, and AEs accepting most CS handoffs. Expansion ARR and NRR lag by a renewal cycle or more, so they confirm the strategy later rather than test it early. If AEs reject most handoffs at day 90, the signals are measuring activity, not intent.

◷
Contoso QBR written and shared in their hub
Renewal teammate
Ready
◎
Northwind went live 9 days early
Onboarding teammate
Done
↗
Fabrikam usage dipped, nudge sent to 3 teams
Adoption teammate
Sent
★
Priya Shah agreed to a reference call
Advocacy teammate
New
+
Acme asked about 20 more seats
Expansion teammate
Warm

Every customer gets a next step.

See how AI teammates run your post-sale programs, from onboarding to renewal, on your own customer data.

Related Articles