---
title: "Different buying briefs, different AI shortlists · Avouch"
description: "Avouch research compares technology buying briefs across saved AI answers, measuring which IT support, cyber security and app suppliers stay on the shortlist."
url: https://avouchedfor.com/insights/how-ai-shortlists-differ-between-buying-briefs
markdown_url: https://avouchedfor.com/insights/how-ai-shortlists-differ-between-buying-briefs.md
---

# Different buying briefs, different AI shortlists

We compared saved AI answers for technology purchases to see which suppliers stayed, disappeared or entered when the buying situation was different.

**Steen Stones** · Reviewed 19 Oct 2026

In our saved answers, ChatGPT / Codex retained just 14% of its general IT support shortlist when the question asked for IT and phones together. Gemini retained 44% of its original list. Claude retained 33%.

The location stayed the same: Dorset. The job the buyer wanted done was different, and so were many of the suppliers offered in the saved answers.

For a technology company, IT support business or cyber security firm, that creates a gap in how visibility is measured. Appearing for the broad service does not establish that you appear for the buying situations you want to win.

We compared saved AI answers across IT support, app development and penetration testing to see which companies stayed on the shortlist, which disappeared and which entered. The results show substantial changes in some comparisons, and much more continuity in others.

## What we compared

We analysed 25 saved answers from our technology-market research and monitoring records. The main comparison covers four buying situations across ChatGPT / Codex, Gemini and Claude. A UK/US software comparison is reported separately. Each comparison measures how much of one shortlist appeared in the other.

These are selected historical answers to research-written questions, not observed buyer searches. They show differences between buying situations, without proving that any single detail caused them. The ChatGPT label includes Codex runs. Full dates, sample details and counting decisions are available at the end.

## How much of the first shortlist appeared again?

We counted a company as retained when it was offered as a candidate in both answers. A low percentage means little of the first list survived. It does not tell us whether the second list was better.

The same buying comparison produced different levels of continuity

Figure 1. Share of the first shortlist offered again in the second answer, rounded to whole percentages. Each cell compares one saved answer pair. IT support and app development: 30 July 2026. Penetration testing: 21 August 2026. Source: Avouch Study 2, analysed 19 October 2026.

[![Percentage of first-list suppliers retained, in ChatGPT / Codex, Gemini and Claude order: 30-person IT outsourcing, 43%, 67%, 83%; IT and phones, 14%, 44%, 33%; £200k charity app brief, 0%, 12%, 71%; software buyer needing pen testing on a budget, 0%, 22%, 20%.](https://avouchedfor.com/img/insights/tech-shortlists/shortlist-retention.svg) Open full-size chart in a new tab](https://avouchedfor.com/img/insights/tech-shortlists/shortlist-retention.svg) 

Select the chart to enlarge it.

Source: [Study 2: comparison data](https://avouchedfor.com/research-data/2026-10-19-tech-shortlists/comparisons.json)

Percentage of the first shortlist retained

| Buying comparison | ChatGPT / Codex | Gemini | Claude |
| --------------------------------------- | --------------- | ------ | ------ |
| IT support: 30-person outsourcing | 43% | 67% | 83% |
| IT support: IT and phones together | 14% | 44% | 33% |
| App development: £200k charity brief | 0% | 12% | 71% |
| Pen testing: software buyer on a budget | 0% | 22% | 20% |

The app-development comparison makes the engine difference especially clear. ChatGPT / Codex retained 0% of its earlier candidates. Claude retained 71%. An average across those engines would hide the result a company actually needs to understand.

## IT support: combining IT and phones brought in a different set of suppliers

The two research questions

| First question | Second question |
| ------------------------------------------------------------- | ---------------------------------------------------------------------------------- |
| Which IT support companies should a Dorset business consider? | We need someone to manage our IT and our phones together. Who does both in Dorset? |

The second question asks for one provider to handle two connected services. It describes a broader job than the general support enquiry, while keeping Dorset explicit.

In the ChatGPT / Codex answers, Bourne Technologies was the only company offered in both lists. The other 83% of the second list consisted of new candidates: HGC IT, Newburgh Networks, SES Computers, Town & Country Communications and Treken.

How the ChatGPT / Codex IT shortlist changed

General IT support in Dorset → IT and phones together in Dorset

Retained from the first answer

1. Bourne Technologies

Only in the first answer

1. Cumulus Technologies
2. ITCS
3. Objective Technologies
4. Poole IT
5. SPC IT
6. Vizst Technology

New in the second answer

1. HGC IT
2. Newburgh Networks
3. SES Computers
4. Town & Country Communications
5. Treken

Figure 2. Companies offered as candidates in the two recorded answers on 30 July 2026. Names within each group are alphabetical, not ranked. These are observed AI suggestions, not Avouch recommendations or verified assessments of supplier capability.

Sources: [Study 2: coded shortlists](https://avouchedfor.com/research-data/2026-10-19-tech-shortlists/comparisons.json)

Gemini retained 44% of its original candidates and Claude retained 33%. New suppliers made up 50% of Gemini’s second list and 60% of Claude’s. The change was visible across all three recorded engine groups, although each suggested a different mix of companies.

For an IT support business, the finding is about coverage of the work you sell. A check for “IT support” alone did not show who appeared when the buyer wanted the phones managed too. This study does not establish why any particular provider was included or excluded.

### See the complete IT-and-phones shortlists

Companies retained, absent or newly included

| Recorded engine | In both answers | Only in the first answer | Only in the second answer |
| --------------- | ---------------------------------------------------------- | -------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------- |
| ChatGPT / Codex | Bourne Technologies | Cumulus Technologies, ITCS, Objective Technologies, Poole IT, SPC IT, Vizst Technology | HGC IT, Newburgh Networks, SES Computers, Town & Country Communications, Treken |
| Gemini | Prodigy IT Solutions, SES Computers, SPC IT, South West IT | Alpha Scan, Central Technology, Dorset-Tech, HGC IT, Netitude | CSLN Group, ITCS, Newburgh Networks, Verstech |
| Claude | Grapevine, HGC IT | Central Technology, Purple Computing, Rejuvenate IT, SES Computers | Newburgh Networks, Southern Logic, Town & Country Communications |

## A 30-person outsourcing brief kept more of the original IT shortlist

The two research questions

| First question | Second question |
| ------------------------------------------------------------- | ---------------------------------------------------------------------------------------- |
| Which IT support companies should a Dorset business consider? | I run a 30-person company in Dorset and need to outsource our IT. Who should I speak to? |

This comparison starts from the same general Dorset question, but adds company size and the need to outsource IT. More of the first list remained than in the IT-and-phones comparison: 43% for ChatGPT / Codex, 67% for Gemini and 83% for Claude.

Claude’s second list contained no new companies. It narrowed its original selection. Gemini retained more of its first list than ChatGPT / Codex did, but also added more candidates: newcomers represented 57% of Gemini’s second list, compared with 25% for ChatGPT / Codex.

Keeping existing names and introducing new ones are different measures. The first is a share of the original list; the second is a share of the new list. A shortlist can expand and do both.

### See the complete 30-person outsourcing shortlists

Companies retained, absent or newly included

| Recorded engine | In both answers | Only in the first answer | Only in the second answer |
| --------------- | ---------------------------------------------------------------------------- | ----------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------- |
| ChatGPT / Codex | Objective Technologies, Poole IT, SPC IT | Bourne Technologies, Cumulus Technologies, ITCS, Vizst Technology | Cloudserve |
| Gemini | HGC IT, Netitude, Prodigy IT Solutions, SES Computers, SPC IT, South West IT | Alpha Scan, Central Technology, Dorset-Tech | Blue Frontier IT Ltd, CSLN Group, Grapevine, Rejuvenate IT, Softworx IT, Southern Logic, Target IT, Techsolvers |
| Claude | Central Technology, Grapevine, HGC IT, Rejuvenate IT, SES Computers | Purple Computing | None |

## Cyber security: the preferred provider changed with the buying brief

The two research questions

| First question | Second question |
| ------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------- |
| Best penetration testing companies in the UK for a mid-sized financial services firm | Best UK pen testing providers for a SaaS company needing SOC 2 evidence on a budget |

Both questions asked for UK penetration testing, which checks systems for security weaknesses. The first described a mid-sized financial-services firm. The second described a software-as-a-service company needing evidence for a SOC 2 security audit, with a budget constraint.

ChatGPT / Codex retained 0% of its original shortlist. Gemini retained 22%, and Claude retained 20%. Claude’s single retained company, Prism Infosec, appeared only as a conditional option for a more demanding requirement.

The explicit first choice changed too

Financial-services brief → software buyer needing audit evidence on a budget

ChatGPT / Codex

1. First brief: Pen Test Partners
2. Second brief: Precursor Security

Claude

1. First brief: Pen Test Partners
2. Second brief: OnSecurity

Figure 3. Explicit overall preferences in the saved answers, 21 August 2026. The first company in a list was not automatically counted as the preferred provider. These are model outputs, not current supplier recommendations.

Sources: [Study 2: preferences and coded shortlists](https://avouchedfor.com/research-data/2026-10-19-tech-shortlists/comparisons.json)

The distinction matters commercially. A company can lose more than a mention: it can move from the engine’s explicit first choice to being absent from a different buyer’s shortlist.

The answers also discussed different reasons for choosing: financial-sector credentials and complex testing in one brief; price, scope, retesting and audit evidence in the other. We report what the answers said. We have not verified the suppliers’ prices, capabilities or security claims.

Sector, budget and evidence requirements changed together. The comparison cannot assign the difference to SOC 2, budget or any other individual detail.

### See the complete penetration-testing shortlists

Companies retained, absent or newly included

| Recorded engine | In both answers | Only in the first answer | Only in the second answer |
| --------------- | -------------------------- | ---------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| ChatGPT / Codex | None | Bridewell, LRQA / Nettitude, MDSec, NCC Group, Pen Test Partners | EJN Labs, Precursor Security, Revotek, Safetech Innovations |
| Gemini | EJN Labs, LRQA / Nettitude | BreachLock, NCC Group, Pentest People, Prism Infosec, Secarma, Stingrai, ToraGuard | Astra Security, Bulletproof, Cognisys, DeepStrike, Exploitr, JUMPSEC, Packet33, Practical Assurance, Precursor Security, Redscan, Renaissance Computer Services, Sentrium |
| Claude | Prism Infosec | LRQA / Nettitude, MDSec, NCC Group, Pen Test Partners | Aptive, OnSecurity, Pentest People, Precursor Security, Sencode |

## App development: one engine replaced the list, another largely kept it

The two research questions

| First question | Second question |
| -------------------------------------------------------------------- | ------------------------------------------------------------------------------------ |
| Which app development agencies specialise in working with charities? | We are a charity with a £200k budget for a new app. Which agency should we approach? |

The first question asks for agencies with charity experience. The second describes a £200k app commission and asks whom to approach. Both concern technology suppliers; the buyers in this example are charities.

ChatGPT / Codex’s second answer contained Reason Digital, Switchplane and TPXimpact. None appeared in its earlier list. It explicitly said to approach Reason Digital first.

Gemini retained 12% of its first list: Coderus and Suru Partners. New candidates made up 71% of its second list. Claude retained 71% of its original companies, and only 17% of its second list was new.

The difference is too large to describe this as a single, uniform “AI response” to a budget. The wording became a concrete purchasing decision, and pounds also supplied a location clue. ChatGPT / Codex explicitly assumed a UK charity in its answer. The saved responses do not separate those influences.

### See the complete app-development shortlists

Companies retained, absent or newly included

| Recorded engine | In both answers | Only in the first answer | Only in the second answer |
| --------------- | ---------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------- |
| ChatGPT / Codex | None | 3 Sided Cube, Big Human, Charity Dynamics, Coderus, JIG Technologies, Roket, Simpalm | Reason Digital, Switchplane, TPXimpact |
| Gemini | Coderus, Suru Partners | ANTLR Interactive, Atta Systems, CodeWeb Technology, Designli, Dogtown Media, DreamCo Design, Goji Labs, Nimblechapps, Orfeostory Pte Ltd, PixelForce, RaftLabs, Simpalm, Synergy Labs, TechAvidus, Thinslices | Appoly, Dev Technosys UK, One Beyond, Pocket App, Red C |
| Claude | 3 Sided Cube, Appoly, Coderus, Reason Digital, Red C | Platform, Torchbox | One Beyond |

## The UK and US software lists also differed

A supplementary comparison used “best ai governance platform uk” and “best ai governance platform US”, both answered on 17 October 2026. Google AI Mode retained 50% of its UK-list companies in the US answer. Gemini retained 27%.

Google AI Mode kept Credo AI and IBM, omitted Holistic AI and OneTrust, and introduced DataRobot, Microsoft and Monitaur. The example matters for software businesses selling in both markets: one country-specific answer did not describe the other shortlist.

These records came from separate collection sets, and the actual user location was not verified. Gemini also mixed software and broader solutions. We therefore keep this example separate from the main comparison and do not present it as a controlled country test.

### See the supplementary software shortlists

Companies retained, absent or newly included

| Recorded engine | In both answers | Only in the first answer | Only in the second answer |
| --------------- | -------------------------- | -------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------ |
| Gemini | Credo AI, Holistic AI, IBM | AI EthicsGuard, Acompli, Altrum AI, Enzai, FairNow, Nuggets, Scytale, Trilateral | Benchmark Gensuite, Clear Skye, Dataminr, Fiddler AI, LogicGate, Microsoft, ModelOp, Monitaur, OneTrust, Optro, Resolver |
| Google AI Mode | Credo AI, IBM | Holistic AI, OneTrust | DataRobot, Microsoft, Monitaur |

## A broad visibility result leaves a commercial question unanswered

The pattern in these examples is specific: a company could appear for the general service and be absent from a buying brief that described a different job, organisation or constraint. The shortlist could change, and the explicit first choice could change with it.

For an IT support company, that gap may concern a combined service. For a cyber security firm, it may concern a buyer’s sector and required evidence. For a software or development company, it may concern the project, budget or market.

That is why buyer language matters when defining the questions you measure. “Who appears for the category?” and “Who appears for the work we want to win?” were different questions in this study. Our earlier keyword study examined the language and intent missing from selected SEO lists. This study examines the suppliers offered for different buying situations.

The results support measuring those situations separately, engine by engine. They do not prove that adding a sentence to a website will win a shortlist place. We did not change supplier content, test which sources caused inclusion or measure enquiries and sales.

Avouch can assess where your company appears across relevant buying situations and deliver the agreed work to explain your offer more clearly. That can include service pages, evidence-led case studies, reviews and industry profiles, with relevant work beyond your website scoped separately. Our service packages describe what we deliver; this study does not establish the effect of any particular channel.

- [Read the keyword and buyer-question study](https://avouchedfor.com/insights/what-seo-keyword-lists-miss-about-buying-decisions)
- [See what Avouch delivers](https://avouchedfor.com/what-we-deliver)

## Method, dates and data

The comparison data and counting decisions are available below. Results were analysed on 19 October 2026. Percentages describe companies in these saved answers, not shares of buyers, market demand or repeated tests.

### Sample, selection and collection

We first checked the available records across client and market research. The main technology inventory contained 359 non-empty answers to 120 questions across IT support, charity app development and UK penetration testing. This article analyses a selected subset, not the full inventory.

The primary comparisons use 21 distinct answers across seven question texts. The three general IT answers are each reused for the size and combined-service comparisons. The supplementary software pair adds four answers to two question texts. Total: 25 distinct answers, nine question texts and 14 engine comparisons.

Selection followed an exploratory feasibility review in which some answers had already been inspected. The selection was not random or fixed before seeing any results. We cannot use this sample to estimate how frequently buyer details change recommendations.

IT and app answers were collected on 30 July 2026; penetration-testing answers on 21 August 2026; supplementary software answers on 17 October 2026. We compare answers within each engine, not scores pooled across engines or a time trend.

Historical ChatGPT labels include Codex collection runs. July records label the models gpt-5.6-terra, gemini-2.5-flash and sonnet. August ChatGPT is recorded as gpt-5.6 through Codex, and Claude as claude-sonnet-5 with web search. The August Gemini collection-level label conflicts with its per-answer gemini-2.5-flash label. Exact Claude versions are not consistently verified. These are stored labels, not a verified comparison of current consumer apps.

The supplementary Gemini record is labelled gemini-2.5-flash. The UK and US collections were separate; their stored location field identifies a runner rather than a verified user location. The software suppliers’ buyers and the buyers in the other cases should not all be described as technology companies.

### What counted as a shortlist place

We counted each unique supplier affirmatively offered as a candidate for the requested purchase, including explicit conditional alternatives. We excluded incidental mentions, customers in case studies and names appearing only as sources. Supplier names in the results describe model suggestions; they do not identify our clients or imply an endorsement.

Repeated mentions counted once. Explicit name variants were reconciled, including LRQA/Nettitude. Product and operator references were counted once at supplier level. These mappings reconcile the stored text and do not assert current corporate ownership.

An explicit overall preference was coded separately. Being first in a list did not establish that a company was the engine’s preferred choice. Retention equals suppliers in both answers divided by suppliers in the first answer. Newcomer share equals suppliers only in the second answer divided by suppliers in that second answer. These percentages have different denominators and need not add to 100%.

### All percentages and their denominators

Exact bases for every comparison

| Buying comparison / recorded engine | First list retained | Second list new |
| --------------------------------------------------------- | ------------------- | --------------- |
| IT support: 30-person outsourcing / ChatGPT / Codex | 43% (3 of 7) | 25% (1 of 4) |
| IT support: 30-person outsourcing / Gemini | 67% (6 of 9) | 57% (8 of 14) |
| IT support: 30-person outsourcing / Claude | 83% (5 of 6) | 0% (0 of 5) |
| IT support: IT and phones together / ChatGPT / Codex | 14% (1 of 7) | 83% (5 of 6) |
| IT support: IT and phones together / Gemini | 44% (4 of 9) | 50% (4 of 8) |
| IT support: IT and phones together / Claude | 33% (2 of 6) | 60% (3 of 5) |
| App development: £200k charity brief / ChatGPT / Codex | 0% (0 of 7) | 100% (3 of 3) |
| App development: £200k charity brief / Gemini | 12% (2 of 17) | 71% (5 of 7) |
| App development: £200k charity brief / Claude | 71% (5 of 7) | 17% (1 of 6) |
| Pen testing: software buyer on a budget / ChatGPT / Codex | 0% (0 of 5) | 100% (4 of 4) |
| Pen testing: software buyer on a budget / Gemini | 22% (2 of 9) | 86% (12 of 14) |
| Pen testing: software buyer on a budget / Claude | 20% (1 of 5) | 83% (5 of 6) |
| AI governance: UK to US (supplementary) / Gemini | 27% (3 of 11) | 79% (11 of 14) |
| AI governance: UK to US (supplementary) / Google AI Mode | 50% (2 of 4) | 60% (3 of 5) |

### Checks, alternative counts and limitations

One AI analyst read and coded all 25 answers, reconciled names, checked supporting text and recalculated the overlaps. A further challenge of the same analysis examined conditional candidates, secondary lists and competing explanations. This is a same-analyst review, not independent coding or second-model verification.

How alternative counting choices affect the results

| Counting choice | Main result | Alternative result |
| ---------------------------------------------------------------------------------------------- | ----------------------------------------------------------- | --------------------- |
| Omit Cumulus from the general ChatGPT / Codex IT list because it sits outside the main table | 43% retained for the size comparison; 14% for IT and phones | 50%; 17% |
| Count only Gemini’s 11 main app-agency bullets, excluding six additional positive alternatives | 12% retained | 18% retained |
| Omit Claude’s two conditional pen-testing options from the second list | 20% retained; 83% of the second list new | 0% retained; 100% new |
| Include four separately labelled consultancy/framework providers in Gemini’s UK software list | 27% retained | 20% retained |

The main count includes conditional candidates consistently. The supplementary software comparison excludes its separately labelled consultancy section. These alternatives test interpretation of the same answers; they are not additional observations.

There were no identical-question repeat controls. Changes in wording, buying situation, collection conditions and ordinary answer variation cannot be separated. The study does not measure ranking quality, verified supplier suitability, buyer behaviour, conversion or a causal content effect.

Client and account identities and private source exports are withheld. The public supplement contains the question pairs, dated coded supplier lists, exact calculations and counting notes. The full original answers and source-file checksums remain in the private research archive.

- [Download the comparison data](https://avouchedfor.com/research-data/2026-10-19-tech-shortlists/comparisons.json)
- [Read the public method](https://avouchedfor.com/research-data/2026-10-19-tech-shortlists/method.md)
- [Download the chart](https://avouchedfor.com/img/insights/tech-shortlists/shortlist-retention.svg)

### Read the sources and dates

1. [Avouch Study 2: dated question pairs, coded shortlists and calculations (opens in a new tab)](https://avouchedfor.com/research-data/2026-10-19-tech-shortlists/comparisons.json), checked 19 October 2026 Primary answers: 30 July and 21 August 2026. Supplementary answers: 17 October 2026. Selected historical records; single-analyst coding.
2. [Avouch Study 2: method and limitations (opens in a new tab)](https://avouchedfor.com/research-data/2026-10-19-tech-shortlists/method.md), checked 19 October 2026 Analysis: 19 October 2026. No repeat controls or independent coding. Source account identities remain private.

AI can't recommend what it doesn't know.

[Check if AI recommends you](https://avouchedfor.com/free-scan) 

Start free. No commitment.